This is probably not what you want but for a truly tiny DIY "UPS" you can use a USB-C PD compatible power pack / bank with an external mini PCB trigger board that lets you manually select output voltage. Normally with PD (Power Delivery) spec power packs the device and power pack "negotiate" proper voltage and current for fast charging depending on multiple factors. These small PCB trigger boards instead let you manually select output voltage (you probably are gonna want 12V I'm guessing for most devices). You can then wire this to a barrel connector or whatever an individual device uses (probably would want one battery pack and board per device). This is, of course, assuming you can power your devices directly off direct current (DC) power. If you need AC I have also had luck experimenting with the same setup outputting 12v DC to small tiny fanless AC inverters. By using an existing USB-C PD power bank you can size it appropriately for the device and time needed and also be reasonably confident that charging and discharging of the internal cells is done safely (assuming you choose a reputable power bank brand). To properly do this you would probably want a power bank with two ports so you could use one for input (battery charging) and the other for output (to the control board and then the final device). Usually these are called "USB-C PD Trigger Boards". I couldn't find the exact one I have but it has a button to select output voltage (5v, 12v, 18v, etc). No idea how efficient these are - some may use buck converters or cheaper parts or vary in quality. Others can be permanently set to a specific voltage via jumper or soldering and appear to be simpler/ smaller in design. I also found this project (I have never used it) which looks to be of possibly of higher quality (and significantly higher price): https://lectronz.com/products/pocketpd This is something similar to what I have used (I have not used this specific board - so buyer beware of course): https://a.co/d/i1mtXJw AITRIP Type-C USB-C Fast Charging Trigger Module (ZY12PDN with Screw Terminal) The people I know doing the vanlife thing also have amazing large and portable battery setups - would be worth looking at forum posts on their setups. For portable prebuilt setups (not custom LiFePO4 installs integrated into the van) I see the "EcoFlow" brand recommended a lot. I also don't know if this was specifically discussed in this thread but have you looked at the smaller Cyberpower "pure sine wave" UPS models ? They are not designed to be rack mounted but are quite compact. I have an older model and just running some network equipment at home it seems to last a few hours. I have not measured the efficiency under low loads like this though - it could be similarly terrible. Ideal world you would avoid the use of an AC inverter entirely (by using network equipment that can run off DC power directly). Does anyone know if the efficiency drops off at lower draws is because the AC inverter being less efficient at low power draw? Or is it somewhere else in the power chain ? On Fri, 11 Apr 2025 at 15:28, Javier J via NANOG <nanog@lists.nanog.org> wrote:
where do you source those batteries?
The cheapest ones that will fit in the ups from amazon with a decent rating.
I literally use a tape measure to measure the old dead lead acid batteries to see what will fit. I have found that I can get batteries with more amp hours than the lead acid replacements.
I figure even if the manufacturer is crap will outlast LA battery technology.
- J
On Fri, Apr 11, 2025 at 4:52 PM Mark Tinka via NANOG < nanog@lists.nanog.org> wrote:
On 4/11/25 17:40, Gary Sparkes wrote:
30% being a reasonable floor absolutely is true.
I didn't say 30% DoD was unreasonable. I said that claiming that going below it to 20% is dangerous is not true.
But, happy to agree to disagree.
Far less stress to go 100 to 30 and back to 100, then 90 to 20 and back to 90, etc. Keeping 30 as an operational floor lets you use full capacity as needed and remain at full functional charge with maximum lifespan retention/recovery.
That can be said of any charge/discharge window if you are not consistently discharging below a given threshold, whatever that is to
you.
EV’s charge/operate that way to extend pack lifespan primarily.
Right, but the key motivation for that is for the car manufacturers to meet warranty claims, normally at least 10 years. And they will remotely manage those charge/discharge profiles to put warranty objectives over range maximization.
4.2v float is fine if you aren’t routinely low end stressing it.
4.2V float is for NMC. LFP is usually around 3.5V float.
Yes, the internal resistance bit is true, but that really does start to kick in around 25-30. A lot of datasheets I’ve worked with talk about 30% and stress zones.
Internal resistance in Li-Ion cells is highest as the battery approaches a fully-charged or fully-discharged state. In other words, internal resistance is highest at 100% SoC and below 20% SoC. But since we know that Li-Ion batteries have a non-linear voltage curve until about 10% SoC, internal resistance is most dangerous below this SoC value.
Far better to remain at 100% float for battery lifespan than to routinely dip below 30%.
In my experience, not going below 20% SoC will be better than holding a 100% charge for an extended period of time, especially if you are not actively controlling ambient temperature.
Li-Ion batteries really do not like holding a full voltage for too long, although, for me, that would not be as bad as routinely running an SoC below 20%.
Calendar aging is not as cut and dried as it may seem. This is blatantly obvious with cellphones, of course, but holds true for lots of other implementations too.
Li-Ion batteries have a completely different use-case for cellphones than for home backup, because we prioritize capacity and peak performance for cellphones vs. home backup. This is why it is quite normal for people to expect their phone battery to be pretty "useless" after an average of 3 years.
I expect 90% capacity on 10-year-old batteries stored properly almost always, usually – at a minimum.
Again, stored batteries have no value to anyone :-).
The better metric is how to maintain working batteries for 10 years and see how much capacity you've retained by that time.
But cycle durability is what truly matters in the long run for lifespan, not calendar aging, for batteries that often see use. And 30% is a sweet spot between usable capacity and lifespan extension to often double the manufacturer’s rated cycle count.
Yes, we all want cycle durability, but calendar aging is unavoidable. And since the biggest contributor to calendar aging is ambient temperature, most owners will lose capacity due to that because they do not have active cooling for their batteries.
I disagree with 30% being a recommended DoD floor (most OEM's do not suggest that), but that's okay :-).
Remember, cycle count means you’re actually using it – so I’m not charging or discharging any less, only doing so to specific levels.
Li-Ion batteries are not cheap. People will always prioritize capacity, and the money to buy a larger pack just to save 10% from your discharge cycle does not justify the extra longevity in environments where most people will let batteries overheat.
I’d also point out a lot of research is also indicative of low discharge levels being the leading factor to degradation, not capacity float charge status. Heat is the the number one factor (outside of or at high end of design spec heat, as often seen in consumer devices). Low discharge is factor #2. High / full charge stress comes in around #3. Better to engage #3 than #2 or #1 for lifespan retention.
Maintaining a long term 100% SoC state is not problematic if you can actively cool your battery. Most owners will not, which is why this can be more damaging to your battery than a low SoC.
Remember, most batteries will spend the majority of their life closer to 100% SoC than at 1% SoC.
Mark. _______________________________________________ NANOG mailing list
https://lists.nanog.org/archives/list/nanog@lists.nanog.org/message/S4CZUXPX... _______________________________________________ NANOG mailing list
https://lists.nanog.org/archives/list/nanog@lists.nanog.org/message/AY2K4WBT...