Mixed Li-Ion and LiFePo4 parallel discharge test

[NPhil]

OK I try to outline what will happen
- charge all up to 29,2V (3,65 x 8 = 4,17 x 7)
- Relaxing cells. LFP drops to 3,4-3,45V/cell = 27,2V
- NMC relax V I am not sure, lets say 4,1V/cell = 28,7V
- There is already 1,5V difference, NMC will overcharge LFP

- start to discharge it.
- LFP drops fast (like 2-5% discharge) to 3,3V/cell = 26,4V
- NMC does not drop, lets say 4V = 28V
- Here is again a 1,6V difference. Both will discharge, but NMC will discharge faster

- 95% NMC 4V -> 28V
- 95% LFP 3,3V -> 26,4V ... NMC discharges faster

- 50% NMC 3,8V -> 26,6V
- 50% LFP 3,25V -> 26V ... they are coming closer

- 10% NMC 3,5V -> 24,5V
- 10% LFP 3,15V -> 25,2V ... now it reversed and LFP discharging faster

- 5% NMC 3,2V -> 22,4V
- 5% LFP 2,9V -> 23,2V ... LFP discharging faster

- If no load and stays here then LFP charging NMC

(I am too tired now to check the charge diagrams too )

I see it working only if you remain in the 20-80% SoC part. That is only 60% usable capacity.
First NMC will discharge faster than LFP will discharge faster ... if constant load. If not they start to charge each other (causing extra wear)

I don't think most of that stuff will happen the way you think it will happen. Some of it, I have observed not happening. For example, now that I have a BMS on the LiFePO4 battery, it's held to about 27V by the BMS. BTW, your speculation is not, in itself, objective evidence.

 

[DThames]

I created this post and the attached test to establish some facts (at least in this example). It is true that the load shifts back and forth (between the two types) over the discharge curve. But if the total load is low C, for each of the batteries, it shouldn't be a big deal that one battery shoulders most of the load at times. I would adjust the charge to voltage to support the battery type that had the greater amount of "settle to" voltage range so one battery would not be trying to hold the other one at a voltage higher than its settle volts. I mean, charge to a level that the LiFePo4 could settle to 3.4v per cell without the Li-ion trying to "charge" it. You would be less than 100% charge on the Li-ion, but fairly good.

 

[ddalfons]

FYI...Individual boards were actively balanced and charged to 29.20 volts. The boards were then connected and allowed to absorb for a few minutes at 29.20 volts (no current flowing). Once removed from the charger the boards rested until a 3.00 amp load was switched on at 28.90 volts. Attached are the test results for the individual boards as well as the paralleled boards, the details are in the image. More to follow, preliminary current sampling during discharge shows no circulation at all, only a slight balancing current after cutoff. Will post final current sharing results once hardware for greater dynamic range sampling is complete.

 

[crossy]

@ddalfons Any update on this?

 

[NPhil]

@ddalfons Any update on this?

I have no detail testing results (too many other things have been happening), but I have run 8S 24V 55AH of LiFePO4 in parallel with 4 7S 5AH Tesla 2170 packs for a couple of years now in my mobile solar power testbed. It's still running well enough to keep two 65W mobile refrigerators (one as a +4F freezer) and an assortment of smaller stuff running overnight, and a pot of tea and a couple of small meals when there is some sun. I still haven't done a capacity test. I can't remember why I said 27V two years ago, that was an error, apparently. I did set the charge controller to about 27V (because I was _afraid_), but, after a couple of years of daily testing/use, it seems clear that I should have stuck with the Epever Tracer AN defaults for LiFePO4, because I doubt I'm getting a full charge at my location, where some trees create a shadow which cuts off late afternoon solar charging.

 

[JulianGoesPro]

Thank you guys for all that info… I am using two 24V LiFe4Po 100Ah packs and had some 18650 laying around so I made a 7S7P pack out of 3000mAh cells. So obviously the LiOnPack is way smaller, but I have an inverter that shuts off on 22V supplementing my home electricity, so it’s no big deal when voltage sags super low, but I have a Pi connected to my Victron (and Sun1000 + Trucky + Shelly) setup which needs power or I lose connection / data. Would love to hear whether a 8P lion might hit a better “optimal” range looking at those curves… I am not sure…

Thanks!