Using Lithium Iron Phosphate batteries with APC Smart UPS

[rsaeon]

@ze_cook

The graphs tell me that my UPS charges the battery up till 29.2 volts, which is exactly right for LFP full charge voltage, 3.65 * 8. It then idles the battery at 28V, which should be fine for LFP.

That’s so very, very close to the ideal voltages for 8s LFP battery pack. It’s really helpful to know that this UPS is about as good as it gets for a LiFePO4 conversion with minimal modifications.

These are the ideal voltages:

24V 8S LiFePO4 Battery w/ BMS

• Absorption: 29V

• Float: 27.2V

• Inverter Cut-off: 21.4V-24V

And if you want a 5000+ cycle lifespan (13+ years at daily full discharge):

If you want your LiFePO4 cells to last a long time, you can set your absorption to Victron’s custom LiFePO4 charge profile recommendation:

• 12V Battery: 14.1V

• 24V Battery: 28.2V

• 48V Battery: 56.4V

You can pull full capacity with the absorption figures above, but the charge rate will be reduced at high SOC.

Float should be 13.6V/27.2V/54.4V

Low voltage disconnect should be 12V/24V/48V

From:

https://diysolarforum.com/threads/recommended-charge-profile-for-diy-lifepo4-batteries-sticky-post.5101/

You might be able to tweak the charging/float voltage of this UPS to bring it close to these values, but it’s probably not necessary since this isn’t a solar setup where you’d be charging/discharging daily.

Which leads to the question — why a 100Ah batttery pack? This UPS tops out at 600W, so if you’re drawing the maximum capacity, you’ll see a run time of 24x100/600 = 4 hours. At your current load of 70w, it’s over 34 hours of runtime. A 6Ah battery pack of 32650’s from quartzcomponents would’ve costed around Rs 4k and give you two hours of run time. Larger LiFePO4 cells are useful when you have a high current draw, since most cells are limited to 1C discharge. A 32650 based pack (6/12/18/24Ah) would be limited to a 24x6 = 144W max load. Your 100Ah pack can easily handle over 2000W constant load.

@n1r0

• Putting LiFe on trickle charge would result in overcharging, which is BAD for any Lithium battery, which might result in thermal runaway. Ever seen EV’s catch fire? This is the primary cause

This UPS float/trickle charges at 28V, which works out to be 3.5V per cell, which is less than the 3.65Vpc safe maximum, so there shouldn’t be any issue with overcharging. Not all UPS/Inverters do this, but it looks like this specific APC model is uniquely suited for an easy LFP swap. Not NMC/Lithium Ion, but LiFePO4/LFP.

I left my 100Ah cells connected to 40A charger at 3.65V for over a day and there was no temperature rise over ambient, or any current draw, it flatlined to zero after full charge was achieved.

@n1r0

• BMS can’t disconnect the battery AND keep the battery ready for use by the UPS in case of power cut.

If the UPS floats/charges at less than the BMS cut-off, then there shouldn’t be any battery disconnection issues. Which again, this UPS appears to be just about perfect for a LFP conversion, it’s very exciting to know!

The biggest issue/drawback with LFP battery swaps is that the BMS would cut off the battery during a discharge before the UPS triggers a shutdown or low battery alarm. So you’d need to put in a system that prevents this scenario from happening (turning off load before the BMS’s low voltage disconnect kicks in).

I have a similar issue with LA batteries in inverters, my Luminous inverters signal a low voltage alarm at 11.5V per battery and cut off load abruptly at 11.4V. I’ve setup battery monitors that trigger shutdowns before this voltage is reached. LA voltage levels also “bounce back” a little bit after the load is decreased so the inverter never actually shuts off. If a system fails to shutdown, then yeah, the inverter cuts off power abruptly. But I do not know if LFP’s have a similar “bounce back” behaviour.

Arrow points to the moment where shutdown is wrapping up. The reading after that is the “bounce back”, this is for a two LA battery 24V system. If it drops to 22.8V, the Luminous inverter would cut off power.

But in this case, using 100Ah battery pack with a 70W load, this will probably never happen (BMS low voltage disconnect). You might even be able to move the entire setup to a cart/trolley, and keep your homelab online while switching houses — if you’re chasing uptime stats like me:

@n1r0

• Putting LiFe on trickle charge would result in overcharging, which is BAD for any Lithium battery, which might result in thermal runaway. Ever seen EV’s catch fire? This is the primary cause

From what I’ve been able to read about (I’m not an expert) but EV fires are usually due to the dynamic nature of the system (bumps and knocks) which loosen up connections over time and cause imbalanced charging and/or arcing inside the battery pack which may not be built to the best standard, or when a less than ideal charger is used. In a static system like this, these issues shouldn’t occur with a LFP pack.

@n1r0

I got a LiFe battery last month and experimented with it: once fully charged, the BMS disconnects the battery for a few seconds. A proper LiFe charger will then stop charging. But a trickle charger will try to resume charging, and trip the protection circuit again, and this loop keeps repeating.

Did you monitor the voltages? If there’s a way to adjust the charging voltage, change it to a value less than the BMS cut off. I’m considering swapping out the LA batteries in a 750VA ups with a small LFP pack but that’s because my APC can be reprogrammed to adjust the charging voltage through a serial cable and an obscure set of commands.

@n1r0

An easy way (at least in my mind) for retrofitting LiFe batteries in Pb systems is to add an extra cell in series. Since the total voltage required for 9 cells in series to reach 100% SoC will never be reached by a 24V Pb charger’s 29.2V, there won’t be any risk of overcharging. Downside is you will get less usable capacity, but this will increase your battery life

That sounds like an inefficient but simple way of adding LFP packs to an unmodifiable inverter/UPS but it would not be easy to find a LFP BMS that can handle an odd number of cells (5S/9S/17S for 12V/24V/48V systems).

@ze_cook

A note on UPS, they are not meant to be running for longer durations, at full loads, and are specced accordingly.

Your particular model should be fine, so long as it has a fan.