Old BMS General Thread

[dillond666]

Awesome job Peter, fascinating stuff

In the end I wimped out and decided to order boards without the soldermask, does make them look lovely though.

[retepsnikrep]

Good lad on ordering, you may regret not getting the soldermask though It looks good but also stops hamfisted soldering like mine causing problems with bridging tracks etc etc.

I've done about 20 Slave boards now, might finish tomorrow if I have a major push.

[cianof]

Fair play to you Peter, been following this thread for a while now. Those boards look excellent. Great work!!

[retepsnikrep]

Thanks.

Done 25 now (50%) and I'm knackered Sadly back to work at 07:00hrs tomorrow but I shall try and do five a day after work for next few days to finish it off.

I now need to sort out my battery mounting in the rear of the car before I can wire up all the slaves and test them, as I only wan't to fit them to the cells once when the cell layout and securing to the removable battery trays is complete.

I have split the pack in half so I can lift it in/out in 2 blocks of 25 cells. It's pretty heavy each half weighs about 35kg.

I will wire up one half on the bench, test it all and then it can drop in, and be secured with four bolts. Then repeat for other half.

[GregsGarage]

Looks like I am going to be busy as well. Just got some circuit boards from Peter today, so a weekend soldering is in order. Thanks Peter.

[retepsnikrep]

Done 45 boards now, only 5 to go On the home straight at last.

OK That's not strictly true, I've done 46 so far but mucked one up by soldering the V Ref in the wrong way round, I noticed and fixed it but the boards a bit messy now so I'll use that as a spare

The old adage, "if it doesn't work in simple mode, it def won't work in complicated mode" applies.

I'll revert to/add the enhanced Digital Slave functions back in again as I get the associated Master functions up and running.

www.solarvan.co.uk/bms/SlaveSimpleDigit ... _v0_02.txt

[GregsGarage]

I have built one of Peters analogue slave boards and bench tested it, and so far it works as intended.

I have adjusted the clock speed from its initial setting of 31kHz to 250kHz. At the slowest speed the response was very slow, maybe 2-3 seconds. Seemed worse at the low voltage cutout, maybe the lower voltage is reducing clock speed even more! The next step up, 125kHz, was better but still a noticeable lag. Current draw at 3 volts is around 300-400ua (according to a cheap amp meter), at 250kHz.

I have also adjusted the MinCellV to 220 (2.2 volts) from the original value of 200. The reason I have done this is that while turning the voltage down towards the 2 volt cutoff I found that the small delay meant that the voltage might get lower than 2 volts before the LowOpto gets turned on. The extra 0.2 volts should give a buffer.

Now I am going to see if this works with my EV managment module (must come up with a better name).

[retepsnikrep]

Greg, just some thought's, one of the things that takes the time is the 10x over sampling on the adc input, you could reduce that to 5 or even remove oversampling altogether, and get an immediate speed increase of 2-10x enabling you to then select the slowest cpu clock speed.

The LM385 V ref is fed via a 22k resistor which equates to a permanent load of approx 130ua. It needs a certain amount of power to produce a reliable voltage but the resistor could be increased in value.

100k would give 30ua but may be too low for reliablility, something in between say 47k may be better and combined with the slowest cpu speed may make a big difference to current consumption.

I'm also intending to adjust the low V setting to about 2.5v as the amount of available power is very low once the cell V drops down to these levels.

Although the cpu will function to below 2V, it's pointless pushing it until it locks up and has to be hard reset, some V overhead is a good idea.

To keep it all in perspective, your 400ua drain is probably below the self discharge rate of the 100ah cells anyway

It will run for over 80days for 1ah consumed

[GregsGarage]

Peter,

Thanks for the suggestions regarding the sampling and possible resistor changes but I am happy with the current drain as it is, at least for now. I am using 100ah cells, but as you say, the self discharge rate is probably similar to the drain these boards draw. Maybe after this is proven in a vehicle, I might try some of those changes.

I'll go with your suggestion and raise the low voltage point to around 2.5 volts. While seeing how it would behave with a nearly discharged batttery, I lowered the voltage down to just under 1 volt, at this point the low voltage opto stopped conducting but when the voltage got back to 2 volts it started conducting again. As long as I didn't go below 1 volt it seemed to work fine. So that is very promising, still working at voltages below spec and no reset needed.

Here is a pic of the testing, showing the slave board connected to my module, and yes that is the kitchen table!

In this picture you can see the load led on the slave board is on, and the led on the top right represents the battery charger. turning up the voltage another 0.1v turns on the high voltage opto and my board shuts off the charger. I am running my board on 24 volts, which is what my controller uses, but this board will also run at 12 volts. Had a moment last night where I wondered if the optos would be alright at 24 volts, they seem fine.

[retepsnikrep]

Greg's going to be first with them running in his car I think

The optocouplers for info are rated at

Voltage, isolation:5000V ac
Output type:NPN Phototransistor
Current, input:60mA
Voltage, output max:70V

https://uk.farnell.com/jsp/search/produ ... KU=1328382

http://www.farnell.com/datasheets/93351.pdf

I'm glad it works so far