Want to move to all 208V for server racks
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
On Thu, Dec 2, 2010 at 10:58 AM, Jay Nakamura <zeusdadog@gmail.com> wrote:
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
Hi Jay, Pretty much any little oddball piece of equipment with an external power brick is at risk. The hundred buck ethernet-based USB extender was my particular lesson. If you're talking about paying customers who bring in their own equipment, you'll run in to this a *lot*. Regards, Bill Herrin -- William D. Herrin ................ herrin@dirtside.com bill@herrin.us 3005 Crane Dr. ...................... Web: <http://bill.herrin.us/> Falls Church, VA 22042-3004
Dear Jay,
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
you mean 240V AC 50HZ and move from 120V 60Hz? (or also 50Hz) you will need to check each device if it supports 240V, commonly the specified power ratings are printed at a stricker on the device itself. Kind regards, Ingo Flaschberger
you mean 240V AC 50HZ and move from 120V 60Hz? (or also 50Hz)
In US, I think everything is 60Hz. But I mean 208v single phase. (Which is what you get when you combine two 120v single phase legs out of three phase, I believe. I am not an expert on AC...)
you will need to check each device if it supports 240V, commonly the specified power ratings are printed at a stricker on the device itself.
I have even been looking at USB HD AC adapter and all other odd ball equipment and I always see the label say "100~240v AC". Dell's old rack mount monitor/KB from 5 years ago even supports 208v (Just wrong connector.)
On 12/2/10 8:30 AM, Jay Nakamura wrote:
you mean 240V AC 50HZ and move from 120V 60Hz? (or also 50Hz)
In US, I think everything is 60Hz. But I mean 208v single phase. (Which is what you get when you combine two 120v single phase legs out of three phase, I believe. I am not an expert on AC...)
Correct, a L-N connection will get you 120V, a L-L connection will get you 208V. Everything in the US is 60Hz.
you will need to check each device if it supports 240V, commonly the specified power ratings are printed at a stricker on the device itself.
I have even been looking at USB HD AC adapter and all other odd ball equipment and I always see the label say "100~240v AC". Dell's old rack mount monitor/KB from 5 years ago even supports 208v (Just wrong connector.)
The vast majority of power adapters are switching these days and will run up to 240, it's when they have built in NEMA 1-15 or 5-15 prongs that you have to overcome. ~Seth
Dear Jay,
you mean 240V AC 50HZ and move from 120V 60Hz? (or also 50Hz)
In US, I think everything is 60Hz. But I mean 208v single phase. (Which is what you get when you combine two 120v single phase legs out of three phase, I believe. I am not an expert on AC...)
I got the point. 120 * sqrt(3), phase to phase, three-phase current in european;
you will need to check each device if it supports 240V, commonly the specified power ratings are printed at a stricker on the device itself.
I have even been looking at USB HD AC adapter and all other odd ball equipment and I always see the label say "100~240v AC". Dell's old rack mount monitor/KB from 5 years ago even supports 208v (Just wrong connector.)
Whats the idea behind todo this? You will also need circuit breakers that both phases are switched of simultaneous? Kind regards, Ingo Flaschberger
you mean 240V AC 50HZ and move from 120V 60Hz? (or also 50Hz)
In US, I think everything is 60Hz. But I mean 208v single phase. (Which is what you get when you combine two 120v single phase legs out of three phase, I believe. I am not an expert on AC...)
That would be considered a 2 pole, 208v receptacle, most commonly a L6-20 or L6-30.
No, I'm pretty sure he means "across the 2 high legs of a 120/208 3ph Wye service", and I'd never heard that idea suggested before. I can see why it reduces the amount of copper you need to run, but it seems as if it would have compensating disadvantages, though I can't think precisely what they might be at the moment. -- jra ----- Original Message -----
From: "Ingo Flaschberger" <if@xip.at> To: "Jay Nakamura" <zeusdadog@gmail.com> Cc: "NANOG" <nanog@nanog.org> Sent: Thursday, December 2, 2010 11:22:32 AM Subject: Re: Want to move to all 208V for server racks Dear Jay,
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
you mean 240V AC 50HZ and move from 120V 60Hz? (or also 50Hz)
you will need to check each device if it supports 240V, commonly the specified power ratings are printed at a stricker on the device itself.
Kind regards, Ingo Flaschberger
In a message written on Thu, Dec 02, 2010 at 11:32:16AM -0500, Jay Ashworth wrote:
No, I'm pretty sure he means "across the 2 high legs of a 120/208 3ph Wye service", and I'd never heard that idea suggested before. I can see why it reduces the amount of copper you need to run, but it seems as if it would have compensating disadvantages, though I can't think precisely what they might be at the moment.
In most residential / small business construction in the US you will find "240V single phase with neutral". There are two hot wires and a neutral from the provider. Hot to hot is 240, hot to neutral is 120. Most colos run their back end plant (e.g. UPS's, Gensets, etc) on 480v 3-phase power. The typical way they get 120v power is to transform that to a 3-phase Y wired output, also known as 3-phase 4 wire. Each hot leg is 120v to the neutral (the fourth wire). You can run hot to hot here as well, where the voltage is 208v. The trick with 208v loads in this situation is you want to keep the load across each pair of phases roughly balanced. What can be particularly confusiong here is the panels look exactly the same. The same physical panel layout your house gets with 2 phases in plus a neutral is now two of the three phases from the three phase power go in, plus a neutral. Same breakers are used, with hot to hot being 208 volt. The difference is, in the colo there are three of them: A N B B N C C N A | | | | | | | | | Panel 1 Panel 2 Panel 3 With A, B, and C being the 3 phases, and N being the neutral. You may also find this arrangement in larger multi-tennent buildings where they are fed with 3-phase power. -- Leo Bicknell - bicknell@ufp.org - CCIE 3440 PGP keys at http://www.ufp.org/~bicknell/
On Dec 2, 2010, at 8:46 AM, Leo Bicknell wrote:
In a message written on Thu, Dec 02, 2010 at 11:32:16AM -0500, Jay Ashworth wrote:
No, I'm pretty sure he means "across the 2 high legs of a 120/208 3ph Wye service", and I'd never heard that idea suggested before. I can see why it reduces the amount of copper you need to run, but it seems as if it would have compensating disadvantages, though I can't think precisely what they might be at the moment.
In most residential / small business construction in the US you will find "240V single phase with neutral". There are two hot wires and a neutral from the provider. Hot to hot is 240, hot to neutral is 120.
Most colos run their back end plant (e.g. UPS's, Gensets, etc) on 480v 3-phase power. The typical way they get 120v power is to transform that to a 3-phase Y wired output, also known as 3-phase 4 wire. Each hot leg is 120v to the neutral (the fourth wire).
You can run hot to hot here as well, where the voltage is 208v. The trick with 208v loads in this situation is you want to keep the load across each pair of phases roughly balanced.
What can be particularly confusiong here is the panels look exactly the same. The same physical panel layout your house gets with 2 phases in plus a neutral is now two of the three phases from the three phase power go in, plus a neutral. Same breakers are used, with hot to hot being 208 volt. The difference is, in the colo there are three of them:
A N B B N C C N A | | | | | | | | | Panel 1 Panel 2 Panel 3
With A, B, and C being the 3 phases, and N being the neutral.
It is not uncommon for three-phase panels to be different and have all three phases in the panel each phase feeding every third breaker slot. Owen
On Dec 2, 2010, at 11:06 AM, Owen DeLong wrote:
It is not uncommon for three-phase panels to be different and have all three phases in the panel each phase feeding every third breaker slot.
I was just recently trying to explain this to a European friend who thought I was hallucinating this system, so I took a picture. http://dl.dropbox.com/u/230717/temp/208YPanel.jpg That's a picture of one of the breaker boxes in our office, showing what you described. There are 3 phases coming into the panel, each a different coil off a Y transformer, as well as a "neutral". Those are the 4 black wires you see at the bottom. You can see how the three hot phases are staggered as they go up the breaker rails. For standard 110V service, you use a single-wide breaker and send one hot phase + neutral and you get 110V. The difference between two phases is 208 volts though, so you use a double wide breaker and can send to device without using a neutral wire. Just 2 hots and a ground. If that's all you're doing (you don't need legacy 110V service anywhere) you skip the ground wire going into the panel entirely. -- Kevin
----- Original Message -----
From: "Kevin Day" <toasty@dragondata.com>
On Dec 2, 2010, at 11:06 AM, Owen DeLong wrote:
It is not uncommon for three-phase panels to be different and have all three phases in the panel each phase feeding every third breaker slot.
I was just recently trying to explain this to a European friend who thought I was hallucinating this system, so I took a picture.
Precisely the same panel layout I had in my last facility, though we didn't use any 208V branch circuits; thanks for the pic, Kevin. Cheers, -- jra
I was just recently trying to explain this to a European friend who thought I was hallucinating this system, so I took a picture.
http://dl.dropbox.com/u/230717/temp/208YPanel.jpg
That's a picture of one of the breaker boxes in our office, showing what you described. There are 3 phases coming into the panel, each a different coil off a Y transformer, as well as a "neutral". Those are the 4 black wires you see at the bottom. You can see how the three hot phases are staggered as they go up the breaker rails.
For standard 110V service, you use a single-wide breaker and send one hot phase + neutral and you get 110V. The difference between two phases is 208 volts though, so you use a double wide breaker and can send to device without using a neutral wire. Just 2 hots and a ground. If that's all you're doing (you don't need legacy 110V service anywhere) you skip the ground wire going into the panel entirely.
that one looks dangerous. In europe: http://img406.imageshack.us/i/verteilerkasten.jpg/ 64A 240V 3-Phase input. Out to Servers single phase, output to airconditioners with 3 phase (not at this picture). Kind regards, Ingo Flaschberger
On Dec 2, 2010, at 12:08 PM, Ingo Flaschberger wrote:
For standard 110V service, you use a single-wide breaker and send one hot phase + neutral and you get 110V. The difference between two phases is 208 volts though, so you use a double wide breaker and can send to device without using a neutral wire. Just 2 hots and a ground. If that's all you're doing (you don't need legacy 110V service anywhere) you skip the ground wire going into the panel entirely.
that one looks dangerous.
Err, I meant "skip the neutral wire". It's still grounded. And there are normally significantly more covers over the panel than this, there were a dozen screws I had to remove to expose all of this. :) This is a much smaller scale panel though, not far up from a typical home system. The more current you start talking about, the more isolated everything becomes until you wouldn't even be able to see the bus bars like in this one. -- Kevin
Err, I meant "skip the neutral wire". It's still grounded. And there are normally significantly more covers over the panel than this, there were a dozen screws I had to remove to expose all of this. :)
This is a much smaller scale panel though, not far up from a typical home system. The more current you start talking about, the more isolated everything becomes until you wouldn't even be able to see the bus bars like in this one.
are "Residual-current device" (Fi in German) are common in us? I use for servers "Residual-current device" and circuit breaker integrated in one device; but I try to use the more expensive pulse tolerant ones. Kind regards, Ingo Flaschberger
On Dec 2, 2010, at 12:20 PM, Ingo Flaschberger wrote:
Err, I meant "skip the neutral wire". It's still grounded. And there are normally significantly more covers over the panel than this, there were a dozen screws I had to remove to expose all of this. :)
This is a much smaller scale panel though, not far up from a typical home system. The more current you start talking about, the more isolated everything becomes until you wouldn't even be able to see the bus bars like in this one.
are "Residual-current device" (Fi in German) are common in us? I use for servers "Residual-current device" and circuit breaker integrated in one device; but I try to use the more expensive pulse tolerant ones.
They're called "Ground Fault Interruptors" here, or GFI/GFCI. They're extremely common built into wall power outlets, and GFI outlets are required in wet areas (kitchens, bathrooms, hot tubs, outdoors, etc). Most wall outlets with GFIs built into them have a "daisy chain" system where one outlet in the kitchen has the circuitry and the Test/Reset buttons, and it protects all non-GFI downstream outlets from it. Downstream outlets usually have a sticker on them saying "GFI Protected" which is a hint that if the outlet stops working, check other outlets in the room to see if one of them tripped. Newer versions have a light that comes on to indicate when they've been tripped, which is handy for non-technical people to figure out what happened more easily. You can get breakers with GFIs built into them(called GFCIs), but they're favored less than putting them at the outlet. I haven't seen any datacenters using them, but I haven't looked that closely. An electrician I talked to once about it felt that the panel mounted variety were designed to be less sensitive/slower reacting due to much longer wire lengths, but I'm not sure if that's just urban legend, experience with a single product or fact. -- Kevin
Err, I meant "skip the neutral wire". It's still grounded. And there are normally significantly more covers over the panel than this, there were a dozen screws I had to remove to expose all of this. :)
This is a much smaller scale panel though, not far up from a typical home system. The more current you start talking about, the more isolated everything becomes until you wouldn't even be able to see the bus bars like in this one.
are "Residual-current device" (Fi in German) are common in us? I use for servers "Residual-current device" and circuit breaker integrated in one device; but I try to use the more expensive pulse tolerant ones.
They're called "Ground Fault Interruptors" here, or GFI/GFCI.
They're extremely common built into wall power outlets, and GFI outlets are required in wet areas (kitchens, bathrooms, hot tubs, outdoors, etc). Most wall outlets with GFIs built into them have a "daisy chain" system where one outlet in the kitchen has the circuitry and the Test/Reset buttons, and it protects all non-GFI downstream outlets from it. Downstream outlets usually have a sticker on them saying "GFI Protected" which is a hint that if the outlet stops working, check other outlets in the room to see if one of them tripped. Newer versions have a light that comes on to indicate when they've been tripped, which is handy for non-technical people to figure out what happened more easily.
You can get breakers with GFIs built into them(called GFCIs), but they're favored less than putting them at the outlet. I haven't seen any datacenters using them, but I haven't looked that closely. An electrician I talked to once about it felt that the panel mounted variety were designed to be less sensitive/slower reacting due to much longer wire lengths, but I'm not sure if that's just urban legend, experience with a single product or fact.
in europe GFIs are always needed for prection and by law. to avoid the cascading effects the GFCIs are better. break current ranges from 10mA (bath) up to 300mA; for servers I use the 30mA with pulse protection (internal delay) to avoid the server powersupply capacitor loading GFCIs "flip". Kind regards, Ingo Flaschberger
----- Original Message -----
From: "Ingo Flaschberger" <if@xip.at>
in europe GFIs are always needed for prection and by law. to avoid the cascading effects the GFCIs are better. break current ranges from 10mA (bath) up to 300mA; for servers I use the 30mA with pulse protection (internal delay) to avoid the server powersupply capacitor loading GFCIs "flip".
And that, indeed, is one of the circumstances in which Chris Lewis and Steve Bellovin's Wiring FAQ suggests that you should *not* use a GFCI: in places where the inevitable "nuisance trip" is troublesome, like powering servers. That FAQ is a bit dated, of course. And indeed, I never liked GFCI breakers for the usages for which they're mandated in the US, cause the milliamp currents they're supposed to trip on are no match for all that copper resistance... Cheers, -- jra
On Dec 2, 2010, at 3:54 15PM, Jay Ashworth wrote:
----- Original Message -----
From: "Ingo Flaschberger" <if@xip.at>
in europe GFIs are always needed for prection and by law. to avoid the cascading effects the GFCIs are better. break current ranges from 10mA (bath) up to 300mA; for servers I use the 30mA with pulse protection (internal delay) to avoid the server powersupply capacitor loading GFCIs "flip".
And that, indeed, is one of the circumstances in which Chris Lewis and Steve Bellovin's Wiring FAQ suggests that you should *not* use a GFCI: in places where the inevitable "nuisance trip" is troublesome, like powering servers.
That FAQ is a bit dated, of course.
Indeed; it's been unmaintained for quite a number of years at this point. The major place I personally have trouble with GFCIs is on things with big motors, and in particular my basement dehumidifier -- a place I really want a GFCI because we've occasionally had water problems...
And indeed, I never liked GFCI breakers for the usages for which they're mandated in the US, cause the milliamp currents they're supposed to trip on are no match for all that copper resistance...
Wire resistance shouldn't matter. A GFCI is measuring the current in the hot wire compared to the current in the neutral wire; if they differ by more than about 5 milliamps, the device trips. That's why motors cause problems: the inductance of the windings can cause a brief current imbalance. Anyway -- in response to the original question: the US electrical code requires GFCI protection for outlets in kitchens, bathrooms, or unfinished basements, for outdoor outlets, and for any other outlet near water. Canada has slightly different rules, or at least it did when we last updated the FAQ (Chris Lewis is Canadian): their code requires that every duplex kitchen outlet be served by two separate circuits, which generally share a common neutral. A simple outlet GFCI can't handle that setup, since the actual current flowing through the neutral will vary depending on the loads on the two hot wires. You'd need a specialized outlet or breaker GFCI that summed the current across all three wires; such devices may exist but I've never seen them. (Btw -- the usual reason for using outlet GFCIs is that they're much cheaper than breaker versions.) --Steve Bellovin, http://www.cs.columbia.edu/~smb
One thing to be aware of- if you are going to be connecting gear with bigger current draws- Cisco 6509's, most blade enclosures etc. come to mind- then many of them effectively require 208V C19 connectors. There are not as many power strips out there that provide sufficient numbers of C19 connectors as would be desired, particularly if you want remote switched power. In that case 3 Phase power becomes more attractive. Since many datacenters are moving towards consolidation on Blades with SAN backend storage, it is worth keeping in mind. In the current DC's we support, we find the only need for 120V is for laptops and such, which is solved by "convenience outlets" that are not on the UPS plant. We always get at least two 120V circuits just in case they are needed, but haven't had any requirements for them recently. --D On Thu, Dec 2, 2010 at 1:22 PM, Steven Bellovin <smb@cs.columbia.edu> wrote:
On Dec 2, 2010, at 3:54 15PM, Jay Ashworth wrote:
----- Original Message -----
From: "Ingo Flaschberger" <if@xip.at>
in europe GFIs are always needed for prection and by law. to avoid the cascading effects the GFCIs are better. break current ranges from 10mA (bath) up to 300mA; for servers I use the 30mA with pulse protection (internal delay) to avoid the server powersupply capacitor loading GFCIs "flip".
And that, indeed, is one of the circumstances in which Chris Lewis and Steve Bellovin's Wiring FAQ suggests that you should *not* use a GFCI: in places where the inevitable "nuisance trip" is troublesome, like powering servers.
That FAQ is a bit dated, of course.
Indeed; it's been unmaintained for quite a number of years at this point.
The major place I personally have trouble with GFCIs is on things with big motors, and in particular my basement dehumidifier -- a place I really want a GFCI because we've occasionally had water problems...
And indeed, I never liked GFCI breakers for the usages for which they're mandated in the US, cause the milliamp currents they're supposed to trip on are no match for all that copper resistance...
Wire resistance shouldn't matter. A GFCI is measuring the current in the hot wire compared to the current in the neutral wire; if they differ by more than about 5 milliamps, the device trips. That's why motors cause problems: the inductance of the windings can cause a brief current imbalance.
Anyway -- in response to the original question: the US electrical code requires GFCI protection for outlets in kitchens, bathrooms, or unfinished basements, for outdoor outlets, and for any other outlet near water. Canada has slightly different rules, or at least it did when we last updated the FAQ (Chris Lewis is Canadian): their code requires that every duplex kitchen outlet be served by two separate circuits, which generally share a common neutral. A simple outlet GFCI can't handle that setup, since the actual current flowing through the neutral will vary depending on the loads on the two hot wires. You'd need a specialized outlet or breaker GFCI that summed the current across all three wires; such devices may exist but I've never seen them. (Btw -- the usual reason for using outlet GFCIs is that they're much cheaper than breaker versions.)
--Steve Bellovin, http://www.cs.columbia.edu/~smb
-- -- Darren Bolding -- -- darren@bolding.org --
On 12/2/2010 13:42, Darren Bolding wrote:
One thing to be aware of- if you are going to be connecting gear with bigger current draws- Cisco 6509's, most blade enclosures etc. come to mind- then many of them effectively require 208V C19 connectors.
Even smaller stuff like a 2U server will have multiple ratings on the PSU these days: you will only get full capacity out of it at high voltage. Plus, almost any modern PSU will run at higher efficiency compared to 120V.
There are not as many power strips out there that provide sufficient numbers of C19 connectors as would be desired, particularly if you want remote switched power.
In that case 3 Phase power becomes more attractive. Since many datacenters are moving towards consolidation on Blades with SAN backend storage, it is worth keeping in mind.
Most blade enclosures can be found with three-phase power supply options as well, making it even more convenient. When they take three-phase directly it's usually a pair of 20A circuits and you're good for full capacity. ~Seth
On Thu, 02 Dec 2010 13:39:16 -0500, Kevin Day <toasty@dragondata.com> wrote:
You can get breakers with GFIs built into them(called GFCIs), but they're favored less than putting them at the outlet. ...
I think they are now a violation of the NEC. And they were delisted by UL years ago. They pose a hazard as they will not react fast enough to prevent a fatal shock. (and the only one's I've ever seen were outlawed as the breaker itself was a fire hazard.) --Ricky
On Thu, 2 Dec 2010, Ricky Beam wrote:
I think they are now a violation of the NEC. And they were delisted by UL years ago. They pose a hazard as they will not react fast enough to prevent a fatal shock. (and the only one's I've ever seen were outlawed as the breaker itself was a fire hazard.)
You sure about that? GFCI breakers as well as their close cousins AFCIs are still being sold and bought at hardware stores. Antonio Querubin 808-545-5282 x3003 e-mail/xmpp: tony@lava.net
On Thu, Dec 02, 2010 at 12:17:37PM -1000, Antonio Querubin wrote:
On Thu, 2 Dec 2010, Ricky Beam wrote:
I think they are now a violation of the NEC. And they were delisted by UL years ago. They pose a hazard as they will not react fast enough to prevent a fatal shock. (and the only one's I've ever seen were outlawed as the breaker itself was a fire hazard.)
You sure about that? GFCI breakers as well as their close cousins AFCIs are still being sold and bought at hardware stores.
A quick browse of www.homedepot.com and www.lowes.com shows that both of them sell GFCI breakers online and in the stores local to me. Moreover, the UL website (www.ul.com) doesn't say anything about GFCI breakers being delisted, and _does_ mention GFCI breakers as one of three types of GFCI devices. -- Mike Andrews, W5EGO mikea@mikea.ath.cx Tired old sysadmin
On Thu, 2 Dec 2010, mikea wrote:
A quick browse of www.homedepot.com and www.lowes.com shows that both of them sell GFCI breakers online and in the stores local to me. Moreover, the UL website (www.ul.com) doesn't say anything about GFCI breakers being delisted, and _does_ mention GFCI breakers as one of three types of GFCI devices.
Yep I just did the same check. I think the delisting may have applied to specific models from specific manufacturers. I just don't see UL delisting all GFCI breakers. Antonio Querubin 808-545-5282 x3003 e-mail/xmpp: tony@lava.net
----- Original Message -----
From: "Antonio Querubin" <tony@lava.net>
Yep I just did the same check. I think the delisting may have applied to specific models from specific manufacturers. I just don't see UL delisting all GFCI breakers.
Clearly, some intermediate gateway set the evil bit on Steven's message. Cheers, -- jra
On Thu, Dec 2, 2010 at 22:17, Antonio Querubin <tony@lava.net> wrote: ...
You sure about that? GFCI breakers as well as their close cousins AFCIs are still being sold and bought at hardware stores.
I am not sure I would call AFCIs a close cousin to the GFCI (except that they are both more expensive that a non-xFCI breaker). They serve different purposes. The (arc) faults that AFCIs are designed to interrupt would commonly be passed through the GFCI without notice. GFCIs are designed to protect people from shock, and AFCIs are designed to protect against fire from the arc (which also tends to protect people, but less directly).
On Thu, 2 Dec 2010, Ricky Beam wrote:
On Thu, 02 Dec 2010 13:39:16 -0500, Kevin Day <toasty@dragondata.com> wrote:
You can get breakers with GFIs built into them(called GFCIs), but they're favored less than putting them at the outlet. ...
I think they are now a violation of the NEC. And they were delisted by UL years ago. They pose a hazard as they will not react fast enough to prevent a fatal shock. (and the only one's I've ever seen were outlawed as the breaker itself was a fire hazard.)
They are???? Bought some at Grainger the other day.. http://www.grainger.com/Grainger/wwg/search.shtml?searchQuery=GFCI+breaker&op=search&Ntt=GFCI+breaker&N=0&sst=subset Home Depot also must have missed this: http://www.homedepot.com/webapp/wcs/stores/servlet/Search?keyword=gfci+breaker&langId=-1&storeId=10051&catalogId=10053
<> Nathan Stratton CTO, BlinkMind, Inc. nathan at robotics.net nathan at blinkmind.com http://www.robotics.net http://www.blinkmind.com
On Thu, 02 Dec 2010 17:26:51 -0500, Nathan Stratton <nathan@robotics.net> wrote:
They are????
Bought some at Grainger the other day.
Just because someone is selling them doesn't mean they meet building codes. (esp. for residential use.) None of the dozen or so licensed electricians I've ever talked to will use them. None of my local Lowes stock anything you'd use in a home. (60A breakers?) [of course, their website does lie.] And some of those available online are not UL listed. I know the one's I've seen installed (circa 1980) were delisted -- GE sent notice to the electricians that installed them. --Ricky
----- Original Message -----
From: "Ricky Beam" <jfbeam@gmail.com>
Just because someone is selling them doesn't mean they meet building codes. (esp. for residential use.) None of the dozen or so licensed electricians I've ever talked to will use them.
The breakers, I assume you mean.
None of my local Lowes stock anything you'd use in a home. (60A breakers?) [of course, their website does lie.] And some of those available online are not UL listed.
I know the one's I've seen installed (circa 1980) were delisted -- GE sent notice to the electricians that installed them.
This page: http://www.hilo-electric.com/blank?pageid=63 suggests that 2008 code still *permits* them, but neither it nor the concurring Wikipedia article mentions then category having been delisted or manufacture-decontinued. And indeed, I had little luck with Google trying to find evidence of mass delistings of GFCI breakers. Cheers, -- jr 'are we off-topic enough, now? :-)' a
Once upon a time, Ricky Beam <jfbeam@gmail.com> said:
Just because someone is selling them doesn't mean they meet building codes. (esp. for residential use.) None of the dozen or so licensed electricians I've ever talked to will use them.
I saw GFCI breakers installed in a new house this year, and it passed inspection. I think you experienced a recall of a specific device and are confusing that with a general removal. When Toyota recalled a model of car, that didn't mean all cars were banned. -- Chris Adams <cmadams@hiwaay.net> Systems and Network Administrator - HiWAAY Internet Services I don't speak for anybody but myself - that's enough trouble.
GFCI breakers are very common, the slightly less common version are arc fault breakers which are starting to show up more as well. GFCI breakers are often required on large services, most large (new) 480v services I have seen (1000A and larger) a have Ground fault breakers, in fact I have seen some bad outages on entire datacenters where the main breakers had a lower ground-fault current setting (for tripping) than a branch circuit that had a phase-to-ground fault resulting in the main breakers tripping instead of the branch circuit. I don't know if the ground-fault breakers are required just in Washington (I am in seattle) or if it is a NEC requirement. John -----Original Message----- From: Chris Adams [mailto:cmadams@hiwaay.net] Sent: Thursday, December 02, 2010 7:38 PM To: NANOG list Subject: Re: Want to move to all 208V for server racks Once upon a time, Ricky Beam <jfbeam@gmail.com> said:
Just because someone is selling them doesn't mean they meet building codes. (esp. for residential use.) None of the dozen or so licensed electricians I've ever talked to will use them.
I saw GFCI breakers installed in a new house this year, and it passed inspection. I think you experienced a recall of a specific device and are confusing that with a general removal. When Toyota recalled a model of car, that didn't mean all cars were banned. -- Chris Adams <cmadams@hiwaay.net> Systems and Network Administrator - HiWAAY Internet Services I don't speak for anybody but myself - that's enough trouble.
GFCI breakers are very common, the slightly less common version are arc fault breakers which are starting to show up more as well.
Partly because of a code requirement. Houses burning down, etc. Somehow, we all survived for a long time without them, but now there is a huge requirement. Perhaps Sq-D or Eaton paid the NFPA/NEC to put this in the code to sell pricier breakers. Yes, I believe in conspiracies.
GFCI breakers are often required on large services, most large (new) 480v services I have seen (1000A and larger) a have Ground fault breakers, in fact I have seen some bad outages on entire datacenters where the main breakers had a lower ground-fault current setting (for tripping) than a branch circuit that had a phase-to-ground fault resulting in the main breakers tripping instead of the branch circuit. I don't know if the ground-fault breakers are required just in Washington (I am in seattle) or if it is a NEC requirement.
I believe it to be any service 1200 amps or larger. And, you don't have to have GFI trip, you can have a GFI alarm, especially if you are under "engineering supervision." In fact, it is quite normal to have GFI Alarm on the generator mains, so as to prevent you from having a nuisance trip when you transfer to emergency power. As to the second part of your paragraph, that would be discovered (hopefully) in the commissioning process, where you have your coordination studies done. Anyway, back to topic: Vendors, please a) get all your gear to cool front-to-back, and b) let it take 480 polyphase and not require a neutral. I, for one, will be happier. The datacenter of tomorrow (hell, today) require this.
On Thu, Dec 2, 2010 at 8:58 PM, Alex Rubenstein <alex@corp.nac.net> wrote: ...
Anyway, back to topic: Vendors, please a) get all your gear to cool front-to-back, and b) let it take 480 polyphase and not require a neutral. I, for one, will be happier. The datacenter of tomorrow (hell, today) require this.
People are still feeding their gear with AC? Save on PS inefficiency, and feed direct 12/5vDC to the servers. Save space, save power, save cooling. Matt *quietly exits, singing "every watt is sacred, every watt is great..." under his breath*
On 12/03/2010 03:21 PM, Matthew Petach wrote:
On Thu, Dec 2, 2010 at 8:58 PM, Alex Rubenstein <alex@corp.nac.net> wrote: ...
Anyway, back to topic: Vendors, please a) get all your gear to cool front-to-back, and b) let it take 480 polyphase and not require a neutral. I, for one, will be happier. The datacenter of tomorrow (hell, today) require this.
People are still feeding their gear with AC? Save on PS inefficiency, and feed direct 12/5vDC to the servers. Save space, save power, save cooling.
If you're already in a datacenter, getting 208V AC from an existing AC infrastructure is a lot easier, cheaper, and sometimes more plausible than building a DC plant. If you have your own facility, it's a different story, but if you do colo, you probably have more customers expecting AC than DC, so you'll at least need to maintain both infrastructure. -- Kevin Stange Chief Technology Officer Steadfast Networks http://steadfast.net Phone: 312-602-2689 ext. 203 | Fax: 312-602-2688 | Cell: 312-320-5867
----- Original Message -----
From: "Kevin Stange" <kevin@steadfast.net>
People are still feeding their gear with AC? Save on PS inefficiency, and feed direct 12/5vDC to the servers. Save space, save power, save cooling.
If you're already in a datacenter, getting 208V AC from an existing AC infrastructure is a lot easier, cheaper, and sometimes more plausible than building a DC plant. If you have your own facility, it's a different story, but if you do colo, you probably have more customers expecting AC than DC, so you'll at least need to maintain both infrastructure.
It *is* Friday night, Kevin. :-) He said 12/5VDC, not -48. Cheers, -- jra
On Fri, 03 Dec 2010 13:21:07 PST, Matthew Petach said:
People are still feeding their gear with AC? Save on PS inefficiency, and feed direct 12/5vDC to the servers. Save space, save power, save cooling.
What does that do to customer equipment choices? I've got a quarter acre of boxes that I know want 12/5vDC inside the case, but that's not an easily available option from the vendor - most of the time the only option is autoswitching 120-240DC with your choice of power cables. (If anybody has a good TCO analysis for doing this with Dell/Sun/IBM/Apple servers, I'm willing to listen...)
On Dec 3, 2010, at 16:58, Valdis.Kletnieks@vt.edu wrote:
On Fri, 03 Dec 2010 13:21:07 PST, Matthew Petach said:
People are still feeding their gear with AC? Save on PS inefficiency, and feed direct 12/5vDC to the servers. Save space, save power, save cooling.
What does that do to customer equipment choices? I've got a quarter acre of boxes that I know want 12/5vDC inside the case, but that's not an easily available option from the vendor - most of the time the only option is autoswitching 120-240DC with your choice of power cables.
The 10,000amp bus for the 12v feed for a row of server racks would be a thing to behold. I don't think anyone but Paul Wall has seriously considered this.
(If anybody has a good TCO analysis for doing this with Dell/Sun/IBM/Apple servers, I'm willing to listen...)
On Fri, Dec 3, 2010 at 7:18 PM, Joel Jaeggli <joelja@bogus.com> wrote:
On Dec 3, 2010, at 16:58, Valdis.Kletnieks@vt.edu wrote:
On Fri, 03 Dec 2010 13:21:07 PST, Matthew Petach said:
People are still feeding their gear with AC? Save on PS inefficiency,
and feed direct 12/5vDC to the servers. Save space, save power,
save cooling.
What does that do to customer equipment choices? I've got a quarter acre of boxes that I know want 12/5vDC inside the case, but that's not an easily available option from the vendor - most of the time the only option is autoswitching 120-240DC with your choice of power cables.
The 10,000amp bus for the 12v feed for a row of server racks would be a thing to behold. I don't think anyone but Paul Wall has seriously considered this.
Some day I'd love to meet that guy--he sure has come up with some revolutionary ideas here! (OK, so it's not as practical when you have other customers to worry about... but it might not be so crazy when you're looking at the efficiency numbers for 100,000 small 1u power supplies vs a set of much larger ones.) Matt
On Dec 3, 2010, at 19:25, Matthew Petach <mpetach@netflight.com> wrote:
On Fri, Dec 3, 2010 at 7:18 PM, Joel Jaeggli <joelja@bogus.com> wrote:
On Dec 3, 2010, at 16:58, Valdis.Kletnieks@vt.edu wrote:
On Fri, 03 Dec 2010 13:21:07 PST, Matthew Petach said:
People are still feeding their gear with AC? Save on PS inefficiency,
and feed direct 12/5vDC to the servers. Save space, save power,
save cooling.
What does that do to customer equipment choices? I've got a quarter acre of boxes that I know want 12/5vDC inside the case, but that's not an easily available option from the vendor - most of the time the only option is autoswitching 120-240DC with your choice of power cables.
The 10,000amp bus for the 12v feed for a row of server racks would be a thing to behold. I don't think anyone but Paul Wall has seriously considered this.
Some day I'd love to meet that guy--he sure has come up with some revolutionary ideas here!
(OK, so it's not as practical when you have other customers to worry about... but it might not be so crazy when you're looking at the efficiency numbers for 100,000 small 1u power supplies vs a set of much larger ones.)
Ohm's law is a bitch. 10kamp -48v DC plants are bad enough as far as the amount of copper required, running 12v for significant distance is comical, this is the reason small boats airplanes and diesel trucks adopt 24v systems. There's probably some model where top of rack rectifiers makes sense but that's really pretty much what a blade server is. When you look at a motherboard in a server a big chunk of of real-estate is devoted to taking 12v and switching it down to 1.2-1.8 for distribution to the CPU/memory, a 4 socket server might have to carry 400amp around in a space of around 300cm^2 on a layer of the pcb. The justification for running 208 or 480 all the way to a cabinet is all about smaller conductors. Joel
Matt
On Fri, Dec 03, 2010, Joel Jaeggli wrote:
(OK, so it's not as practical when you have other customers to worry about... but it might not be so crazy when you're looking at the efficiency numbers for 100,000 small 1u power supplies vs a set of much larger ones.)
Ohm's law is a bitch. 10kamp -48v DC plants are bad enough as far as the amount of copper required, running 12v for significant distance is comical, this is the reason small boats airplanes and diesel trucks adopt 24v systems. There's probably some model where top of rack rectifiers makes sense but that's really pretty much what a blade server is. When you look at a motherboard in a server a big chunk of of real-estate is devoted to taking 12v and switching it down to 1.2-1.8 for distribution to the CPU/memory, a 4 socket server might have to carry 400amp around in a space of around 300cm^2 on a layer of the pcb.
The justification for running 208 or 480 all the way to a cabinet is all about smaller conductors.
Isn't this one area where Google have already (re-)pioneered recently? Besides, there's a reason why AC won over DC for carrying 0 < x < few hundred (or thousand? Amps) over a reasonable distance. IANA-PowerEngineer, but ISTR the behaviour/efficiency of voltage/current over distance for both AC and DC is well understood. (And no, ISTR it isn't "AC wins." :-) If you're at all serious about discussing this, I bet spending 15 minutes doing some research and then an hour or so crafting some simultaneous equations to solve/graph would be very very eye-opening. Come on guys/girls, you're a bright bunch, post some models and discuss those rather than un-substantiated datapoints! :-) 2c, Adrian
On Dec 3, 2010, at 9:17 PM, Adrian Chadd wrote:
On Fri, Dec 03, 2010, Joel Jaeggli wrote:
(OK, so it's not as practical when you have other customers to worry about... but it might not be so crazy when you're looking at the efficiency numbers for 100,000 small 1u power supplies vs a set of much larger ones.)
Ohm's law is a bitch. 10kamp -48v DC plants are bad enough as far as the amount of copper required, running 12v for significant distance is comical, this is the reason small boats airplanes and diesel trucks adopt 24v systems. There's probably some model where top of rack rectifiers makes sense but that's really pretty much what a blade server is. When you look at a motherboard in a server a big chunk of of real-estate is devoted to taking 12v and switching it down to 1.2-1.8 for distribution to the CPU/memory, a 4 socket server might have to carry 400amp around in a space of around 300cm^2 on a layer of the pcb.
The justification for running 208 or 480 all the way to a cabinet is all about smaller conductors.
Isn't this one area where Google have already (re-)pioneered recently?
Besides, there's a reason why AC won over DC for carrying 0 < x < few hundred (or thousand? Amps) over a reasonable distance. IANA-PowerEngineer, but ISTR the behaviour/efficiency of voltage/current over distance for both AC and DC is well understood. (And no, ISTR it isn't "AC wins." :-)
If you're at all serious about discussing this, I bet spending 15 minutes doing some research and then an hour or so crafting some simultaneous equations to solve/graph would be very very eye-opening.
Come on guys/girls, you're a bright bunch, post some models and discuss those rather than un-substantiated datapoints! :-)
2c,
Adrian
This isn't rocket science and doesn't require much math... 1. For long distances, you need higher voltages to overcome line loss. 2. For larger loads, you want to use higher voltages to have lower amperages so that you can use reasonable wire sizes. 3. It's a whole lot easier to change AC voltages than DC. The system that won is a system of very very very high voltages for the core distribution with transformers converting that to intermediate distribution voltages which are then further transformed down to even lower voltages for service delivery. This is easily done with AC and would be quite complex and inefficient (especially with the technology available at the time this decision was made) with DC. It would probably be more efficient to run the entire country on 200,000 VDC, but, the dangers of exposing the general public to that kind of voltage are, well, probably just one of the reasons we use 110 VAC instead. Owen
On Fri, Dec 3, 2010 at 22:28, Owen DeLong <owen@delong.com> wrote:
... This is easily done with AC and would be quite complex and inefficient (especially with the technology available at the time this decision was made) with DC.
Correct. Now, of course, with switched mode conversion and power FET technology DC-to-DC converter efficiency can be greater than 95% in optimized designs, but back when Edison and Tesla were arguing the merits, DC conversion was very inefficient compared to AC.
On 12/3/2010 9:25 PM, Matthew Petach wrote:
(OK, so it's not as practical when you have other customers to worry about... but it might not be so crazy when you're looking at the efficiency numbers for 100,000 small 1u power supplies vs a set of much larger ones.)
Particularly if you're running your AC power through UPSes -- especially online ones (where there's a constant AC-DC-AC conversion happening). Go to DC for the batteries, never come back. It's a tempting notion. Jima
Your battery stack isn't like 12v either, unless it's one battery. Joel's widget number 2 On Dec 3, 2010, at 20:02, Jima <nanog@jima.tk> wrote:
On 12/3/2010 9:25 PM, Matthew Petach wrote:
(OK, so it's not as practical when you have other customers to worry about... but it might not be so crazy when you're looking at the efficiency numbers for 100,000 small 1u power supplies vs a set of much larger ones.)
Particularly if you're running your AC power through UPSes -- especially online ones (where there's a constant AC-DC-AC conversion happening). Go to DC for the batteries, never come back. It's a tempting notion.
Jima
On 12/3/2010 10:10 PM, Joel Jaeggli wrote:
Your battery stack isn't like 12v either, unless it's one battery.
Try connecting the batteries in parallel rather than in series, then. ;-) Regarding your other message:
The justification for running 208 or 480 all the way to a cabinet is all about smaller conductors.
So 208/480 to each cabinet's UPS, and DC from there? I'm not sure how feasible that is. Jima
----- Original Message -----
From: "Jima" <nanog@jima.tk>
On 12/3/2010 9:25 PM, Matthew Petach wrote:
(OK, so it's not as practical when you have other customers to worry about... but it might not be so crazy when you're looking at the efficiency numbers for 100,000 small 1u power supplies vs a set of much larger ones.)
Particularly if you're running your AC power through UPSes -- especially online ones (where there's a constant AC-DC-AC conversion happening). Go to DC for the batteries, never come back. It's a tempting notion.
And in fact, much carrier class equipment can be had with -48V power, there are ATX and similar power supplies for PCs that are -48, and I *think* I've commercial small UPSs (<3kVa) that give with -48 as well... using 48V battery strings, obviously. Cheers, -- jra
On Fri, Dec 3, 2010 at 10:33 PM, Jay Ashworth <jra@baylink.com> wrote:
And in fact, much carrier class equipment can be had with -48V power, there are ATX and similar power supplies for PCs that are -48, and I *think* I've commercial small UPSs (<3kVa) that give with -48 as well... using 48V battery strings, obviously.
Take a look at the Solar/Renewable energy systems, Xantrex (Schneider actually) makes the XW series inverter/chargers which use 48V battery strings and can be paralleled up to a rated total of about 18kW@120/240. This is done by paralleling 3x 6kW inverter/chargers. They've an integrated transfer switch, load shaving/sharing (IE if you've got say 6kW of generator, but 12kW of Inverter, the system capacity is up to 12kW, with battery assist). And that's just one option, Magnasine makes parallel inverter/charger and inverter systems up to around 12kW, also using 48VDC (or 24VDC) strings. Both of these are sinewave inverters. There's also a telco oriented 48V inverter rack system thats escaping my mind at the moment. It can be setup with A/B 48V strings, and you plug in inverter modules up to IIRC around 8kW. Not parallel capable between racks AFAIK.
----- Original Message -----
From: "Michael Loftis" <mloftis@wgops.com>
On Fri, Dec 3, 2010 at 10:33 PM, Jay Ashworth <jra@baylink.com> wrote:
And in fact, much carrier class equipment can be had with -48V power, there are ATX and similar power supplies for PCs that are -48, and I *think* I've seen commercial small UPSs (<3kVa) that give with -48 as well... using 48V battery strings, obviously.
Take a look at the Solar/Renewable energy systems, Xantrex (Schneider actually) makes the XW series inverter/chargers which use 48V battery strings and can be paralleled up to a rated total of about 18kW@120/240. This is done by paralleling 3x 6kW inverter/chargers. They've an integrated transfer switch, load shaving/sharing (IE if you've got say 6kW of generator, but 12kW of Inverter, the system capacity is up to 12kW, with battery assist).
And that's just one option, Magnasine makes parallel inverter/charger and inverter systems up to around 12kW, also using 48VDC (or 24VDC) strings.
Both of these are sinewave inverters.
There's also a telco oriented 48V inverter rack system thats escaping my mind at the moment. It can be setup with A/B 48V strings, and you plug in inverter modules up to IIRC around 8kW. Not parallel capable between racks AFAIK.
I phrased my comment poorly, which mislead you. I was suggesting a UPS which took 208VAC on on the charge side, and charged 48VDC batteries with it, providing -48 to a rack full of equipment which took that. People actually call those "48VDC UPSs", though in fact they're just Little Teeny Battery Plants. :-) Cheers, -- jra
On Sat, Dec 4, 2010 at 12:45 PM, Jay Ashworth <jra@baylink.com> wrote:
I phrased my comment poorly, which mislead you. I was suggesting a UPS which took 208VAC on on the charge side, and charged 48VDC batteries with it, providing -48 to a rack full of equipment which took that.
People actually call those "48VDC UPSs", though in fact they're just Little Teeny Battery Plants. :-)
Ah, well, the XW (6048's) *do* have a 100A charger each (so up to 300A @ ~48VDC) so they could be used for that too :D-- but that same industry segment, solar/renewable, makes 48VDC charger only/rectifier only systems as well. So my answer still sort of stands :D
Cheers, -- jra
There's also a telco oriented 48V inverter rack system thats escaping my mind at the moment. It can be setup with A/B 48V strings, and you plug in inverter modules up to IIRC around 8kW. Not parallel capable between racks AFAIK.
48V (and some more when batteries are full) are slightly below the limit of non harmfull voltage. Thus you have a voltage with less power loss at short transports and a secure voltage. (creating a short is still not a great idea). Kind regards, Ingo Flaschberger
Date: Sat, 4 Dec 2010 23:26:46 +0100 (CET) From: Ingo Flaschberger <if@xip.at>
There's also a telco oriented 48V inverter rack system thats escaping my mind at the moment. It can be setup with A/B 48V strings, and you plug in inverter modules up to IIRC around 8kW. Not parallel capable between racks AFAIK.
48V (and some more when batteries are full) are slightly below the limit of non harmfull voltage.
Thus you have a voltage with less power loss at short transports and a secure voltage. (creating a short is still not a great idea).
Saying that 48V is not a harmful voltage is a very dangerous statement. It is unlikely to be a threat of electrocution (though even that has exceptions), but people have lost fingers to 12V systems. Lead-acid batteries can deliver way over 100 amps of current and a conductor across "safe" voltage will get hot and, if not heavy enough, will vaporize. The temperatures attained can cause major burns and, should the metal vaporize, can damage tissue so severely that fingers have been lost when the blood vessels were cauterized. While safety rules often list voltages under 50 as being safe, it is still important to exercise caution like removing rings, bracelets and the like. -- R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: oberman@es.net Phone: +1 510 486-8634 Key fingerprint:059B 2DDF 031C 9BA3 14A4 EADA 927D EBB3 987B 3751
Kevin Oberman wrote:
Lead-acid batteries can deliver way over 100 amps of current and a conductor across "safe" voltage will get hot and, if not heavy enough, will vaporize. The temperatures attained can cause major burns and, should the metal vaporize, can damage tissue so severely that fingers have been lost when the blood vessels were cauterized.
While safety rules often list voltages under 50 as being safe, it is still important to exercise caution like removing rings, bracelets and the like.
I can't remember what I was trying to accomplish, but when we were building a telco office, and after making sure I was completely "demetalicized", I had to climb up the ladder and sit on one of the 48V 1/4"x4" (2-sandwiched) copper buss-bars and lay out accross the others, everything being already 'hot'. Unnerving to be sure. I can also recall one morning at the S.P. Railroad when they called all us 'Diesel Electricians' together and showed us a wrench from graveyard shift. Most of one end was burned off, and the other end was welded to the thick, gold, wedding-band which had been cut off the guy's finger on the way to the hospital. They reiterated the mantra, 'when working with batteries, always disconnect the grounded/carbody side first'. At IBM, we had a ritual before working on -anything-. Take off rings, watches/bracelet, tie-clasp and put into pocket. Tuck tie into top opening of shirt (white) so your neck doesn't get broken when tie catches on all the spinning crap. Even after the 360/370 came along you could always tell the old hands...the guys with their tie tucked in.
On Saturday, December 04, 2010 05:52:09 pm Kevin Oberman wrote:
Lead-acid batteries can deliver way over 100 amps of current and a conductor across "safe" voltage will get hot and, if not heavy enough, will vaporize.
Our smallish 540Ah -48VDC plant has a 35,000A short circuit rating; important to know when sizing the disconnect breaker, as 50,000AIC breakers are required for that. The A and B side rectifiers are Lorrain 200A three phase units, built like tanks. We have a secondary 12V plant at one solar location that is using six 2,320Ah cells which required two disconnects in series to meet AIC ratings, since the nearly 100,000A short circuit current makes it difficult to get small (<100A) breakers with 100,000+AIC ratings. We're doing the solar thing for our optical telescopes, using Xantrex inverter/chargers and Outback solar charge controllers, 24VDC nominal strings. Works great; DC input switches make it even nicer, although you then need low voltage cutoffs to prevent battery damage when there have been several days in a row of dark skies. At the 5ESS in Buckhead/Brookhaven I recall seeing an operating A buss current of >20KA years ago; the AIC on that plant has to be huge (of course, that's been 25+ years, and that's my memory, which could be mistaken as to the exact current value). A technician there told me he had seen an 18 inch adjustable wrench totally vaporized when it bridged from B- to ground. Yeah, not something to play with.
48V (and some more when batteries are full) are slightly below the limit of non harmfull voltage.
I suspect you have never seen the pictures of a wrench that "exploded"/"splattered" all over someones body. 50V may not (usually, but your mileage will vary) be able to produce enough current in a body to kill via fibrillation, but as usually deployed it has enough joules to kill in other ways. 50V is the number in the regs below which certain controls are not required. In some jurisdictions, it also allows those that are not "electricians" to perform work. Anyone regularly working around that many joules, no matter the voltage, has either been properly trained in a safety regimen, or is extremely lucky. It is no different than people who work around high pressure compressed air/steam. There is a lot of stored energy there, and you need to treat it with respect (same with heavy weights suspended above your head, or lots of other examples). Gary
On 12/2/10 8:02 PM, John van Oppen wrote:
GFCI breakers are very common, the slightly less common version are arc fault breakers which are starting to show up more as well.
Arc fault breakers are a very new code requirement which I believe is primarily targeted at sleeping areas. My place has them (built about 4 years ago) on the bedroom outlet circuits. If I spin the socket switch on one of the table lamps too fast it'll trip. ~Seth
On Thu, Dec 2, 2010 at 22:39, Seth Mattinen <sethm@rollernet.us> wrote: ...
Arc fault breakers are a very new code requirement which I believe is primarily targeted at sleeping areas. My place has them (built about 4 years ago) on the bedroom outlet circuits. If I spin the socket switch on one of the table lamps too fast it'll trip.
The NFPA priority is to protect life (property/equipment are there too, but lower in priority). (Note that while NFPA 70 is not required, most jurisdictions eventually turn it into their law/codes. But exceptions exist, and your specific requirements may vary, and not all jurisdictions adopt the new rules immediately. Some still (only) require NFPA 70-2005, and not NFPA 70-2008. There is no known case where applying more recent practices has resulted in liability, so some contractors may build to 2008 when only 2005 is being enforced by the inspector). Now that most outlets are grounded, and GFCIs are in locations where people are likely to be the source to ground ("wet" areas), one of the bigger remaining issues for loss of life in the home due to electricity was in the bedroom with arcing between the hot/neutral when people were asleep (and could be overwhelmed by the smoke before they could get out of the house). Another addition to the code a few years ago was what I call "child proofing" the outlets(*). You will see all new (but not existing old stock) outlets having a (usually) mechanical cover for the slots which requires a plug to be pushed in (only the pressure from both prongs will open the cover) to protect against the inquisitive fork or finger problem. NFPA 70 does take into account industry recommendations (for the conspiracy theorists), and the perceived return on the costs (something that saves 1 life over 10 years but costs billions is not likely to make it into "code"). Gary (*) Technically, I think these are called Tamper-Resistant Receptacles, and are required in all new work as of NFPA 70-2008.
On Fri, Dec 3, 2010 at 04:02, John van Oppen <jvanoppen@spectrumnet.us> wrote: ...
GFCI breakers are often required on large services, most large (new) 480v services I have seen (1000A and larger) a have Ground fault breakers,
Actually, my recollection is that large new services include arc suppression rather than ground fault (480V service may be floating in any case, since it would depend on delta-wye distribution). There has been strong efforts to protect the low voltage electricians (in common power distribution speak, 12K+ voltage is high voltage, less is considered low voltage; yes, this is a different point of view). Even with a 100Cal suit on, you really want arc suppression at those high joule ratings to protect a life (every master electrician has a story about arc flashes, and some stories include the outline of the ex-individual on the opposite wall). It is now common when doing work on downstream devices to reduce the arc limits so that ones life has increased protection. A protective trip is better than the alternative.
in fact I have seen some bad outages on entire datacenters where the main breakers had a lower ground-fault current setting (for tripping) than a branch circuit that had a phase-to-ground fault resulting in the main breakers tripping instead of the branch circuit.
*Proper* engineering is more than just putting in a breaker with a high enough rating. The days of nice resistive (think incandescent light bulbs) or inductive (motor/transformer) loads are long gone. Switching power supplies (or large pulse rectifiers) require a more careful analysis. I have seen too many upstream breakers being set at the wrong trip values (the larger breakers have internal adjustments), and trip first. Gary
In a message written on Fri, Dec 03, 2010 at 04:57:03PM +0000, Gary Buhrmaster wrote:
limits so that ones life has increased protection. A protective trip is better than the alternative.
Not always. I worked in a data center with something I thought was very, very cool. http://www.hilkar.com/highresistance.htm The concept, at a high level, is rather than tie the (service, not signal) ground back to grounding rods directly you run it through a large resistor. Now when a phase is "grounded" it runs through the resistor, allowing a small but safe current to flow. Why is this cool? Well, say you have a power strip running at 10A with a bunch of servers on it. If you took a paperclip and inserted it in an empty plug connecting hot to ground with a normal system (simulating a faulty bit of gear) the breaker would trip, all your servers would go off. If you did this with a high resistance setup the paperclip would conduct about 0.5A, maybe less. An alarm, dectecting current, at the resistor would go off to say there was a fault. Your circuit would draw 10.5 amps and everything would stay up and running. That faulty bit of gear didn't take down your entire power strip. This totally eliminates arc faults, and there isn't enough current to ground to arc. I think GFCI's are also unnecessary, as the fault can't conduct enough current to be harmful. Not there are a TON of other details to building such a system. Cost is a factor why more folks don't do it, and it takes a lot of pencil scratching by your EE types. Still, one of the coolest things I've ever seen, and I wish more data centers were built this way. -- Leo Bicknell - bicknell@ufp.org - CCIE 3440 PGP keys at http://www.ufp.org/~bicknell/
Dear Leo,
I worked in a data center with something I thought was very, very cool.
http://www.hilkar.com/highresistance.htm
The concept, at a high level, is rather than tie the (service, not signal) ground back to grounding rods directly you run it through a large resistor. Now when a phase is "grounded" it runs through the resistor, allowing a small but safe current to flow.
currents above 1mA and 50V are dangerous. also the net-frequency of 50hz/60hz cause troubles for the heart (Ventricular fibrillation). If a really fail-tolerant system is needed, that the only solution if to have a ground-free system. the incomming power is transformed (1:1 for example) and not earthed. a special device monitors the voltage between earth and power and do an alarm if one of the power-lines connects to earth - but do no shutdown. the fault can then be repaired without shutdowns. only when 2 faults occur the breakers trip. usually hospitals use such a configuration. probably hilkar system is similar to this one. Kind regards, Ingo Flaschberger
GFCI breakers are often required on large services, most large (new) 480v services I have seen (1000A and larger) a have Ground fault breakers,
Actually, my recollection is that large new services include arc suppression rather than ground fault (480V service may be floating in any case, since it would depend on delta-wye distribution). There has been strong efforts to protect the low voltage electricians (in common power distribution speak, 12K+ voltage is high voltage, less is considered low voltage; yes, this is a different point of view). Even with a 100Cal suit on, you really want arc suppression at those high joule ratings to protect a life (every master electrician has a story about arc flashes, and some stories include the outline of the ex-individual on the opposite wall). It is now common when doing work on downstream devices to reduce the arc limits so that ones life has increased protection. A protective trip is better than the alternative.
Don't confused arc-flash protection with arc-flash circuit breakers. Doesn't sound like you did, but I said it anyway. As far as ground fault protection, in the 2008 NEC Code, it is required on any service 600 volts or less, 1000 amps or more, per 230.95. As an aside, generally it is accepted that 600v and less is 'low voltage' (not to be confused with telecom/datcomm low voltage); 1kv to about 35kv or so is medium voltage, and above that is high voltage. I think IEEE or ANSI or someone defines this. Google around. Arc flash is a whole new requirement, generally for the life safety aspect of the operators of the electrical equipment. I love putting on a arc flash suit to close a 800 amp breaker, when in the old days we'd do it barefooted on a damp floor. Maybe it wasn't smart, but some of the new arc flash stuff is a bit ridiculous.
On Thu, Dec 2, 2010 at 22:07, Ricky Beam <jfbeam@gmail.com> wrote: ...
I think they are now a violation of the NEC. And they were delisted by UL years ago. They pose a hazard as they will not react fast enough to prevent a fatal shock. (and the only one's I've ever seen were outlawed as the breaker itself was a fire hazard.)
While I do not have a copy of NFPA 70-2011 (the latest latest, released a few months ago), my reading of NFPA 70-2008 still allows GFCI breakers (NFPA 70 is the official name for NEC). Personally, I prefer to specify and use GFCI outlets (and I tend to not daisy chain) so that the the fault is next to the use (and no collateral outages occur). Of course, specific breakers may not meet the newest requirements.
It is probably worth nothing that a 3-phase input in Europe is actually 240/415 volt Y (for every panel I have seen in Germany at least, even the places I have lived there had 240/415 three phase). The normal 240v single phase outlet circuits were the phase to neutral voltage. Obviously Europe also runs at 50 hz vs 60 in the US as well but the three phase still works the same way. A Europe 64 amp 240/415 circuit is pretty close to equivalent in to a 277/480 Y configured 60 amp circuit in the US. The biggest notable difference is that equipment that runs on two different service voltage ranges where Europe has far less need for in-building step-down transformers since even small loads work on the phase-to-neutral voltage of the big services. I always find it interesting in the US to note how many 480v to 120/208Y step-down transformers one can find in a big building or datacenter. John -----Original Message----- From: Ingo Flaschberger [mailto:if@xip.at] Sent: Thursday, December 02, 2010 10:08 AM To: Kevin Day Cc: NANOG list Subject: Re: Want to move to all 208V for server racks
I was just recently trying to explain this to a European friend who thought I was hallucinating this system, so I took a picture.
http://dl.dropbox.com/u/230717/temp/208YPanel.jpg
That's a picture of one of the breaker boxes in our office, showing what you described. There are 3 phases coming into the panel, each a different coil off a Y transformer, as well as a "neutral". Those are the 4 black wires you see at the bottom. You can see how the three hot phases are staggered as they go up the breaker rails.
For standard 110V service, you use a single-wide breaker and send one hot phase + neutral and you get 110V. The difference between two phases is 208 volts though, so you use a double wide breaker and can send to device without using a neutral wire. Just 2 hots and a ground. If that's all you're doing (you don't need legacy 110V service anywhere) you skip the ground wire going into the panel entirely.
that one looks dangerous. In europe: http://img406.imageshack.us/i/verteilerkasten.jpg/ 64A 240V 3-Phase input. Out to Servers single phase, output to airconditioners with 3 phase (not at this picture). Kind regards, Ingo Flaschberger
It is probably worth nothing that a 3-phase input in Europe is actually 240/415 volt Y (for every panel I have seen in Germany at least, even the places I have lived there had 240/415 three phase). The normal 240v single phase outlet circuits were the phase to neutral voltage. Obviously Europe also runs at 50 hz vs 60 in the US as well but the three phase still works the same way.
A Europe 64 amp 240/415 circuit is pretty close to equivalent in to a 277/480 Y configured 60 amp circuit in the US. The biggest notable difference is that equipment that runs on two different service voltage ranges where Europe has far less need for in-building step-down transformers since even small loads work on the phase-to-neutral voltage of the big services. I always find it interesting in the US to note how many 480v to 120/208Y step-down transformers one can find in a big building or datacenter.
aeh.. 230V / 400V is right voltage in technical terms in most european copuntries. (years ago it was 220V / 380V, before it was decided to go up with the voltage) and in bigger datacenters there are also step down transformers from 10kV down. Kind regards, Ingo Flaschberger
On Thursday, December 02, 2010 12:51:42 pm Kevin Day wrote:
For standard 110V service, you use a single-wide breaker and send one hot phase + neutral and you get 110V. The difference between two phases is 208 volts though, so you use a double wide breaker and can send to device without using a neutral wire. Just 2 hots and a ground. If that's all you're doing (you don't need legacy 110V service anywhere) you skip the ground wire going into the panel entirely.
The photo of the Square-D QO plugin breaker panel was nice; thanks. Our Liebert 'precision power' PDU's have what appear to be GE panelboards on the 120/208 side, 42 position, two panelboards per PDU. They are wired like the Square-D in you picture. However, there is one thing in your reply that needs clarification. A wye connected branch circuit panelboard must always be equipped with a grounded conductor (neutral) AND a grounding conductor (ground) to meet code. That is unless the panel is listed for delta use in a 'neutral-less' arrangement and no single-pole breakers are present in the panel. The grounding conductor must never be used as the grounded conductor; the ground is only for fault currents, never for load currents, and the grounded conductor must be bonded to the grounding conductor only at the service disconnect. That's in a simple single grounding conductor system; many datacenters are wired with separate safety and signal grounding conductors (ours is) that require special precautions to be taken. The signal ground is often called the technical ground in the industry, and is, per NEC, always bonded to the safety (or NEC) ground at the service disconnect. That makes it real fun.
On Thursday, December 02, 2010 11:46:28 am Leo Bicknell wrote:
You may also find this arrangement in larger multi-tennent buildings where they are fed with 3-phase power.
There are two other 3 phase setups that are somewhat common. 120/240V delta (has the third leg 'wild' at 208V to neutral, with the neutral (grounded conductor in NEC parlance) connected to a centertap on one transformer of the three required). This one has special labeling requirements and a prohibition on single-phase loads being connected to the 208V leg. This one is sometimes provided in a two-transformer 'open delta' arrangement (instead of the correct three-transformer 'closed delta') and is, you might say, 'three phase lite' in practice. 480V delta corner ground. This one is 480V three phase on three wires; one phase is grounded at the service entrance, and the other two phases are 480V to ground. This one also has special labeling requirements and no 'neutral' in the conventional sense.
On Thu, 2 Dec 2010, Jay Ashworth wrote:
No, I'm pretty sure he means "across the 2 high legs of a 120/208 3ph Wye service", and I'd never heard that idea suggested before. I can see why it reduces the amount of copper you need to run, but it seems as if it would have compensating disadvantages, though I can't think precisely what they might be at the moment.
The only ones I can think of are relatively modest, such as needing 2-pole breakers or a pair of ganged single-pole breakers for each circuit, so a panelboard would only be able to support half as many 208V circuits as 120V circuits. That could translate into needing more panelboards, more/larger switchgear to feed those panelboards, etc, but you can plan for this up-front easily enough if this new construction or a re-fit of an existing space. The panelboards we put in our DR site last year are quite large, so we have some room to grow, and we also used 3-phase PDUs with both 120V and 208V receptacles, there are fewer individual circuits going out to each cabinet. jms
----- Original Message -----
From: "Ingo Flaschberger" <if@xip.at> To: "Jay Nakamura" <zeusdadog@gmail.com> Cc: "NANOG" <nanog@nanog.org> Sent: Thursday, December 2, 2010 11:22:32 AM Subject: Re: Want to move to all 208V for server racks Dear Jay,
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
you mean 240V AC 50HZ and move from 120V 60Hz? (or also 50Hz)
you will need to check each device if it supports 240V, commonly the specified power ratings are printed at a stricker on the device itself.
Kind regards, Ingo Flaschberger
On Thu, 2 Dec 2010, Ingo Flaschberger wrote:
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
you mean 240V AC 50HZ and move from 120V 60Hz? (or also 50Hz)
Probably not; 208V AC here in the US comes from 3-phase distribution systems and is relatively common in datacenters, as well as other commerical and industrial settings. What we've done is to install one 120V, 15A circuit per rack along with the 2x or 4x 208V 30A circuits. There are too many oddball and/or just plain old devices out there to go totally without. Like another commenter mentioned, the prime offender these days are devices with external power bricks or wall-warts; often times they only have NEMA 5-15 plugs so at least there won't be temptation to stick them in 208V receptacles. Assuming you go with IEC C-13 or C-19 receptacles for those 208V circuits, that is. Just be careful on older non-autosensing power supplies where you have to flip a switch to go from 100-120V to 200-240V input, in that you make sure to flip them to begin with, and that you flip them back should you ever mover them back to a 120V circuit. -- Jameel Akari
On 12/2/10 8:35 AM, Jameel Akari wrote:
Just be careful on older non-autosensing power supplies where you have to flip a switch to go from 100-120V to 200-240V input, in that you make sure to flip them to begin with, and that you flip them back should you ever mover them back to a 120V circuit.
Been there, done that with my nagios box when I had to replace a fan years ago. The build table was 120V so I flipped the switch and forgot to flip it back. It actually booted for about 5 seconds before things inside the PSU started exploding and spewing magic smoke. Scared the daylights out of me. No damage other than requiring a new PSU. ~Seth
On Thu, Dec 2, 2010 at 11:22 AM, Ingo Flaschberger <if@xip.at> wrote:
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
you mean 240V AC 50HZ and move from 120V 60Hz? (or also 50Hz)
you will need to check each device if it supports 240V, commonly the specified power ratings are printed at a stricker on the device itself.
Hi Ingo, 208 and 480 both at 60 hz are common three-phase voltages available in commercial buildings in North America. 208 is three hot conductors 120 degrees out of phase with each other, each 120 volts to common. 480 is the same but with 277 volts to common. 208 is often used with higher-wattage computing equipment while 480 is usually used for distribution on the input side of a large UPS and for lighting. Another thought for you Jay - if you deliver L21 series receptacles to the cabinet (5 wires) the customer can employ it as 120vac, 208vac or a mix as they choose, though you will have to facilitate plug converters for their PDUs. Also mixing on the same circuit complicates amperage estimating something fierce unless the use at one of the voltages is trivial. Regards, Bill Herrin -- William D. Herrin ................ herrin@dirtside.com bill@herrin.us 3005 Crane Dr. ...................... Web: <http://bill.herrin.us/> Falls Church, VA 22042-3004
On Dec 2, 2010, at 7:58 AM, Jay Nakamura wrote:
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
I once had a cage that was all 220v. Wasn't an issue at all. Had two devices that required some effort to work around... 1. A small media converter. This was powered by a wall-wart style power supply that fed it 12v DC, but, the wall wart that shipped with it did not handle 220v. No regret, but, a quick trip to Fry's to buy a suitable universal wall-wart with 220v capability and problem solved. 2. Our cordless screwdriver charger would not accept 220v. Initially, we just plugged it into an outlet in the customer work area whenever we were at the datacenter. Long term solution, we bought one of those international transformers and hooked it up that way. I think we might have built some of our own adapter cables to deal with plug issues, but, it's pretty easy to build 3-wire pigtail converters and Home Despot has all the necessary supplies. Owen
On Thursday, December 02, 2010 10:58:52 am Jay Nakamura wrote:
Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v?
Wall wart supplies will need changing, more than likely. I have a few racks with 208V distribution (EMC 40U racks are built this way), and haven't run into many issues. But you do need to watch carefully, and whatever you do do not wire a 5-15R or 5-20R (or any other '5-' receptable or 'L5-' receptacle) to 208; use the proper '6-' receptacles or IEC receptacles (as mentioned) for all 208 power. This is typically mandated by NEC in new installations, and the electrician doing the distribution should be familiar with the NEMA connector chart. See https://secure.wikimedia.org/wikipedia/en/wiki/NEMA_connector for more information on those connectors. Also, if your customers provide their own UPS equipment that could be an issue, as very few UPS I'm aware of are multi-voltage input (APC SmartUPS 3000 is what I use typically, and that by default needs an L5-30R). But if you provide a neutral with your 208 (using an L14 connector) you then can have mixed distribution with 120V available on 5-15R's but 208 receptacles for major power consumers in the rack. That's what I've done in most of my racks that need 208 (like for 7609's and Cisco 12K), other than the EMC's, which use L6-30's for the rack input, and IEC receptacles for the devices in the rack.
On 12/2/2010 9:58 AM, Jay Nakamura wrote:
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
Biggest issue we see with people still needing 120V outlets is external modems for out of band access. Most of the time these modems are attached to the console of carrier managed routers. Or as others in the thread have mentioned, wall-warts for things like USB hard drives, low-end KVMs, etc often are NEMA 5-15P plugs hardwired to them. ASA5505s have this problem with the cable as well, but their power supplies will work on 208V with the necessary adapter. Jeremy
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v?
I can tell you, that from a collocation operator prospective, we want you to do 208v. I'd love to require it in our facilities, but sales people won't let me go that far. Why? A couple of reasons.. Neutral current, more power delivered using less copper, etc. Personally, I like delivering two L21-30's per rack and call it day - allows for a comfortable 8kw per rack in 2N+1 redundancy. And, it still has a neutral if it's needed, which we hope it isn't. We rarely run into things that require 120v, but it's usually older equipment. Or, most notably, the 'call home' modems that EMC uses on their SANs.
On Thu, Dec 02, 2010 at 01:59:33PM -0500, Alex Rubenstein wrote:
A couple of reasons.. Neutral current, more power delivered using less copper, etc. Personally, I like delivering two L21-30's per rack and call it day - allows for a comfortable 8kw per rack in 2N+1 redundancy. And, it still has a neutral if it's needed, which we hope it isn't.
Here's a question for you. How do you calculate the total current & power capacity of a L21-20 or L21-30, and how do you do the calculations in order to balance the load between the phase legs? This seems like it would be a trivial thing to do, but given that the three legs are 120 degrees out of phase with each other, I don't think you can just do normal addition. For example, I have APC AP7961 3-phase PDUs with L21-20 plugs. The management interface claims a maximum load per phase of 16A (which I believe is the 80% derating of 20A required by NEC). Does this mean I can draw 16A * 3, or 48A total if I have a perfectly equal balance? Also, how does this relate to power, i.e. how many kVA or kW does this provide? 16A * 208V * 3 phases ~= 10 kVA? On another note, how do you calculate N+1 power feeds in your racks? If you have 2 PDUs fed from two different branch circuits/UPSes/etc. do you just set your PDU load alarm thresholds at 50% of the max rating of each PDU and never load them beyond that point, so that if you lose one PDU/branch circuit/UPS and the dual-power servers transfer their load over to the other side, it doesn't get overloaded? That would be 8A on each phase in the case of my AP7961's. Of course, things get complicated when you have a mix of single- and dual-power servers, especially if you have server admins who don't keep you apprised as to the types of equipment that are installed there as things change over time...
On Fri, Dec 3, 2010 at 07:54, Chuck Anderson <cra@wpi.edu> wrote: ....
On another note, how do you calculate N+1 power feeds in your racks? If you have 2 PDUs fed from two different branch circuits/UPSes/etc. do you just set your PDU load alarm thresholds at 50% of the max rating of each PDU and never load them beyond that point, so that if you lose one PDU/branch circuit/UPS and the dual-power servers transfer their load over to the other side, it doesn't get overloaded?
That would be around 40%, not 50% (80% of 50%). Note that there are some caveats. Some power supplies are more or less efficient at different (low vs. high) utilizations, and depending on the design, you are running (with 2 power supplies) either each at (around) 50% of load, or 1 at 100% and the other at 0%. It is *possible* to be able to run near 60% on two UPS circuits if the power supplies are inefficient at 50%. But this requires a lot more design and evaluation work than the (easy to calculate) 40% target. Also note that *your* electrical engineer may de-rate the circuits capacity due to the fact that switching power supplies generate numerous artifacts on the lines. These are all advanced (electrical) engineering topics. Gary
Also note that *your* electrical engineer may de-rate the circuits capacity due to the fact that switching power supplies generate numerous artifacts on the lines. These are all advanced (electrical) engineering topics.
From a practical, real-world standpoint, these are not concerns today.
On Friday, December 03, 2010 02:54:37 am Chuck Anderson wrote:
Here's a question for you. How do you calculate the total current & power capacity of a L21-20 or L21-30, and how do you do the calculations in order to balance the load between the phase legs? This seems like it would be a trivial thing to do, but given that the three legs are 120 degrees out of phase with each other, I don't think you can just do normal addition.
You would be correct. A pretty good three-phase calculation reference can be found at http://www.3phasepower.org/3phasepowercalculation.htm It is vector addition, and with the angles involved you end up multiplying and dividing by the square root of 3 a lot. Also see the wikipedia article, and here's a few others: http://www.gavinelectrical.com/content/threephase.htm http://www.servertech.com/uploads/documents/0000/0236/3-Phase_Power_in_the_D... That last PDF is directly related to this discussion, and a good read.
On 12/02/2010 09:58 AM, Jay Nakamura wrote:
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
We run our datacenters with mostly 208V power because it lets us get more power-hungry equipment in a single cabinet. With the exception of very old servers, pretty much all standard power supplies are auto-sensing across the 110 - 240 range voltages and will work fine as long as you use an IEC C13 to C14 cable. Most of the older power supplies have a manual switch you must switch if you don't want to blow the power supply. All network equipment that uses a standard IEC C13 cable that I've seen is auto-sensing, but you should certainly check the documentation. I've seen recent and old Dell, Cisco, HP and Netgear switches that work fine with 208V. For anything with a AC adapter, we check the transformers and find most of those are auto-sensing too. The trick is either the customer has to know in advance and pick up an AC adapter with a C14 connector (which is fairly rare since they all use different polarization, voltage and connector sizes), or to stock some NEMA 5-15 to C14 converters. For a Cisco ASA, which we see a lot, you need a C5 cable. The standard cable is a C5 to NEMA 5-15. We picked up some adapters from C5 to C14 standard pretty cheap to make these work. It is very good practice to check EVERYTHING before plugging it in because if it can't handle 208V, you will hear a pop and it will be dead before you can realize your error. Pretty much anything that uses power has a label on it somewhere describing its supported input voltage. -- Kevin Stange Chief Technology Officer Steadfast Networks http://steadfast.net Phone: 312-602-2689 ext. 203 | Fax: 312-602-2688 | Cell: 312-320-5867
On Thu, Dec 2, 2010 at 7:58 AM, Jay Nakamura <zeusdadog@gmail.com> wrote:
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
I got burned when I tried to install a DSL modem for OOB access, in our 208v network racks. The modem only accepted 120v. My choices were too either run a power cable across the cage to a 120v rack or install a 120v circuit in the network rack. Chris.
Hi, On Thu, 2010-12-02 at 10:58 -0500, Jay Nakamura wrote:
I really want to move all newly installed internal and customer racks over to all 208v power instead of 120v. As far as I can remember, I can't remember any server/switch/router or any other equipment that didn't run on 208v AC. (Other than you may need a different cable) Anyone have any experience where some oddball equipment that couldn't do 208v and regret going 208v? We won't have any TDM or SONET equipment, all Ethernet switches, routers and servers. I have control over internal equipment but sometimes customers surprises you.
In one colo I helped manage, we had some crappy netgear switches which couldn't handle 208v. Provided you have proper equipment, you should be fine though. This was a non-profit though, so we were trying to get by with whatever was the most cost-efficient option. William
btw, one thing I do not recall seeing on this thread is that 208v avoids one of the common problems with 120v, which is the third harmonic issue. With the cheaper switching power supplies, one will often see significant 3rd harmonics in the waveforms(*). The 3rd harmonic, across a 3 phase circuit, are additive on the neutral. In worst case, your (common) neutral current may exceed the line currents. Proper engineering for significant 120v distribution in new DC construction often requires double sized neutrals to mitigate against this. Using 208v mitigates this particular issue. Gary (*) There are also other harmonics, but for this discussion, 3rd is what matters.
btw, one thing I do not recall seeing on this thread is that 208v avoids one of the common problems with 120v, which is the third harmonic issue.
With the cheaper switching power supplies, one will often see significant 3rd harmonics in the waveforms(*). The 3rd harmonic, across a 3 phase circuit, are additive on the neutral. In worst case, your (common) neutral current may exceed the line currents. Proper engineering for significant 120v distribution in new DC construction often requires double sized neutrals to mitigate against this. Using 208v mitigates this particular issue.
I should have been clearer, when I originally said, "it lessens neutral currents." I meant that to mean, as well, less harmonics.
participants (34)
-
Adrian Chadd -
Alex Rubenstein -
Antonio Querubin -
Chris Adams -
Christopher Phillips -
Chuck Anderson -
Darren Bolding -
Gary Buhrmaster -
Ingo Flaschberger -
Jameel Akari -
Jay Ashworth -
Jay Nakamura -
Jeremy Bresley -
Jima -
Joel Jaeggli -
John van Oppen -
Justin M. Streiner -
Kevin Day -
Kevin Oberman -
Kevin Stange -
Lamar Owen -
Leo Bicknell -
Matthew Petach -
Michael Loftis -
Michael Painter -
mikea -
Nathan Stratton -
Owen DeLong -
Ricky Beam -
Seth Mattinen -
Steven Bellovin -
Valdis.Kletnieks＠vt.edu -
William Herrin -
William Pitcock