Paralleling LiFe with Li-Ion - How to match the discharge curves?

[thunderstorm80]

Hi,
My main battery is A123 79V 20Ah.
On long distances, I add 72V 23Ah Li-Ion battery with GA cells which on the paper is excellent in capacity and supplement the more powerful (but heavier) A123.
On practice, the A123 has a flat discharge curve (~3.3V through most of the SOC) while a standard Li-Ion declines almost linearly from 4.2V to 3V.
The result is that when fully charged, the Li-Ion does it job well while the A123 with it's lower impedance takes most of the load.
But when the Li-Ion fell below 79V, it barely do anything, "waits", and only "kick in" again when the A123 is empty. The problem then, is that the Li-Ion takes the entire load on itself which GA cells don't like on the long term.

I wonder how is it possible to combat this?
I could use DC/DC boost converter, but it will have significant loses and it would be very heavy by weight.
A resistor in series with the A123 will cause even bigger loses, and diodes in series to the A123 will only solve the problem very temporary. (and will have some loses too)

If I were to build a PWM switching circuit in series with the A123, in order just to limit it when desired (without closed loop for voltage monitoring - similar to opening/closing relay but very fast), will it still be heavy by weight, for currents up to 40-50A?

 

[flippy]

the only way to combat this behaviour is to ditch the A123 and increase the GA battery size to get the proper capacity.
or ditch the GA for better cells with more lifespan and a flatter curve.

your ideas with converters is a no go. you are creating more problems instead of solving the base problem.

if you dont want to lose speed or much power you need to crank up the battery voltage as high as the controller can take and dial back the max rpm and amps to limit the power so you dont destroy the motor.

 

[john61ct]

Yes either live with it, or go to all one chemistry, LFP likely not, due to lower density.

 

[dogman dan]

You really can't match those discharge curves. But if you must run that way, run till voltages match, then connect. Run till the A123's are empty, Then STOP.

I seriously doubt your GA's are waiting. But current will flow both ways, depending on the needs. Its near impossible to see what is really happening, unless you have dual watt meters.

On a burst, the current will flow from the lowest resistance pack the most. Then it will flow back the other way, which can be bad for some BMS. ( if above regen capability limits for the bms). If the GA's are getting hit hard during the bursts, that's not good. If they seem to be doing nothing, then they are not getting stressed much.

Sounds to me like you are ok, the GA's don't show obvious strain, till they are alone. Clearly, once they are alone, they are also pretty empty, so that is why I say just stop when the Lifepo4 needs to.

If you must, limp home on the GA's alone. But I mean limp. keep it 300w or less for that last mile.

 

[thunderstorm80]

You really can't match those discharge curves. But if you must run that way, run till voltages match, then connect. Run till the A123's are empty, Then STOP.

I seriously doubt your GA's are waiting. But current will flow both ways, depending on the needs. Its near impossible to see what is really happening, unless you have dual watt meters.

On a burst, the current will flow from the lowest resistance pack the most. Then it will flow back the other way, which can be bad for some BMS. ( if above regen capability limits for the bms). If the GA's are getting hit hard during the bursts, that's not good. If they seem to be doing nothing, then they are not getting stressed much.

Sounds to me like you are ok, the GA's don't show obvious strain, till they are alone. Clearly, once they are alone, they are also pretty empty, so that is why I say just stop when the Lifepo4 needs to.

If you must, limp home on the GA's alone. But I mean limp. keep it 300w or less for that last mile.

I installed a second Cycle Analyst so I can diagnose the performance.
The behavior is as I wrote - First 2/3 A123 and 1/3 GA's, then for most of the discharge it's 5/6 A123 and 1/6 GA's, and then when the A123's are empty the GA takes the entire load. At that point I "limp", as you say - limiting myself to no more than 800W (The pack has 1600Wh so I limit myself to 0.5C). The GA's are still 70% full when the A123's are empty....
A good thing at that "bottom" point, is that whenever I am just coasting or idling, the GA will charge the A123's a bit (and at a moderate rate), so then when I resume throttling, that extra energy immediately coming to use is from the A123's.