DIY Lithium Packs, Proposal and Discussion

[luthj]

I got torque setup on my phone, and an OBDII to bluetooth adapter connected to the BMS.

I roughed out a few gauges for the various critical PIDs. Looks like I can get all the important data. Even better, with Torque pro, I can log data directly to my phone for later review. The BMS also supports logging data, but I need to have my laptop connected via serial. You will notice the cell voltage variance. This was mid balance.



Here are the balance resistors I made up.



All organized, just about ready for close up.

 

[john61ct]

Your correct, the Orion Jr gives me full control over balancing. Right now it set to balance when any cell is over 3.25V, and bring all cells within 5mv. It will do this regardless of charge/discharge.

Since my rig will be occupied full time, time above 3.4Vpc is very minor. I may go with a lower charge voltage in the 14.0 range, and do a weekly/monthly run to 14.2V to complete a balance. The BMS can only balance at 150mA, so it needs a bit of time as the cells age.

Not to be argumentative but IMO

3.55V is climbing the curve a fair bit, monthly would be better than weekly

Just know, sitting holding CV stage at 13.8 - 14.0V or so for the time it takes to "finish" balancing gets the job done just as well as going higher

In fact you could decide to balance at 3.25Vpc, does not matter a bit so long in that case as **both** HVC and LVC are based on first-cell-to-hit the setpoint, not pack level voltage.

As before, not my intention telling you what to do, just putting it out there

 

[luthj]

I was just able to cram the current sensor into the box. An interesting note about the current and voltage measurements from the BMS. The raw values are not software filtered. They vary significantly when running from the inverter, or charging from a mains charger. For example my AC unit pulls about 48A, but when viewing the live data from the BMS, the raw current varies from 35-52A. Now there is an "Average current" option, which is averaged over 200ms. Interestingly the average current is very close to what my outback battery monitor shows, which is good. Cell voltage does vary plus/minus 25mv when charging from the inverter/charger at higher currents. So there must be significant noise, or ripple current.

IMG_20200831_155740924

 

[luthj]

A few notes as I wind down the testing.

The Orion BMS requires some cell taps to be attached to a specific cell when not in use. Failing to do so can cause the ASIC which reads cell voltages to pick up noise, creating a random fault code, which made me think the BMS was failing.

The Jr 2 also uses the current convention where positive current is discharge, and negative current is charge. I got them backwards, and it caused me an hour of fiddling.

For SOC calibration, I have decided to use only upwards drifts. Specifically at around 3.5Vpc it resets to full. Though I may add a down drift at the lower knee if I find myself spending a lot of time there.

I am only seeing about 30mv drift at 200A discharge, so the cells seem well matched. I will do a full discharge capacity test tomorrow, and log it all for a nice graph.

 

[luthj]

I got a weak cell fault when testing, so I checked the BMS internal resistance measurements for the first time. They were pretty varied, with one cell over 1mohm, and others near 0.5mohm. I should be seeing around 0.2 max (at mid range SOC, discharge), and about 0.12 nominal. Though it will vary with SOC and temperature.

I bit the bullet and opened the cover. Nothing seemed out of place. So I took all the busbars off, and polished them with 1,000 grit paper. Easiest method was to oil my belt sander with a 1000 grit belt. I should have paid closer attention, but the 200 grit I used before was not up to the task, and I saw the scratches when I got a few passes in.

Reassembled and torqued to 35 in-lb. Results shown below at nearly full charge. Much better. The values do fluctuate a bit, but should be fine for my application.

 

[luthj]

Did a full discharge test. Called it done at 520AH. Looks like all 560AH are available, but I going a bit conservative. I will post up a graph when I have a chance.

Tested the alternator charging and control. For the naysayers about "alternator meltdowns", I had my alternator pegged at ~100-120A for an hour in 85F temps low idle only. The main case never got over 150, and the outer windings maxed out around 180F. This was measured with my IR thermometer, which is plus/minus 10F accuracy at those temps. With the pack at around 20% SOC actual, the alternator was outputting 13.46V, and the pack was at 13.17V. With no charging source the pack was at about 13.1V. My pack is around 0.5-1mohm at this SOC/temp.

 

[luthj]

Here is the voltage, I tried to smooth out some of the noise, but this about as good as its going to get. Discharge was at about 140A. Charge was at 110A.



Here is the tail end of the charge cycle. The discontinuity in the middle is when my laptop went to sleep for a bit.

 

[john61ct]

I imagine everyone knows by now, my thoughts on the right side of that last graph?

 

[luthj]

Relax, it's qualification testing. It will get adjusted down when I put it into service.