Low cost BMS

[SpmP]

Thanks astro123, foursprings for putting the infos on Sourceforge.

My friend has the car I believe these were developed for, and we have found a few hardware mods necessary... Mostly the resistors are inadequate and need to be 1W+ and off the board 8) (probably add an LED too)

Anyway, I am going to need 28+spares of these soon, and looking at getting a digikey order over the threshold, but before I do, I figure I should ask if anyone is needing any boards fabbed, or has experience with who is good for doing it. Maybe a bulk order is better.
Also any feedback as to which parts are better etc, or maybe has an old order in csv that may speed things up 8)
I would much rather work with AVR due to my experience, for those who have played with the code does it look very PIC specific? Would it be a hard port to AVR C?

Does anyone have an approximate cost breakdown for master and cell

My friend apparently has the source for both top and bottom balancing if this is of any use.
Sorry if there are RTFM questions here 8)

[4Springs]

Originally Posted by SpmP

My friend has the car I believe these were developed for, and we have found a few hardware mods necessary... Mostly the resistors are inadequate and need to be 1W+ and off the board 8) (probably add an LED too)

Interesting!
I actually reduced the value of the resistors (300mW) and it has been working very well. But this may be because I reduce the current from the charger as soon as one cell starts to balance. If your charger can't do that then it changes things. Still, have you tried the software options? - turn on balancing resisters at a lower voltage and then keep balancing once the charger has turned off (I can't remember if harlequin's code did that or not?).

I'll post my views on your other questions (prices, sourcing etc.) in another post. Good to hear from you!

[4Springs]

Originally Posted by SpmP

Anyway, I am going to need 28+spares of these soon, and looking at getting a digikey order over the threshold, but before I do, I figure I should ask if anyone is needing any boards fabbed, or has experience with who is good for doing it. Maybe a bulk order is better.
Also any feedback as to which parts are better etc, or maybe has an old order in csv that may speed things up 8)

I have posted details of my component purchases on the Sourceforge website (look under Files -> Components). Modules were about $4 each, Master about $50. These prices are parts only without the board itself and without the PIC microprocessors.
Have a look at the AEVA forum thread (low cost BMS) if you haven't already - there is a lot of activity over there. In particular Astro has found that the Master board can use the 16F1847 PIC microprocessor chip. This is a drop-in replacement for the 16F1827 listed in the bill of materials, but it has more memory if you think you might need it.

I had my boards made by PCB Train in the UK. They were very fast, knowledgeable and patient (I had never done this before and they helped me figure out what file formats worked best). For some reason I can't find an invoice right now to see how much they cost...
I will be wanting some Master boards soon, but I'm not sure if the economics would work out in our favor considering that we don't live in the same country. It may cost more to send mine over to me than it would have to get them done separately! Worth looking into though.

Originally Posted by SpmP

I would much rather work with AVR due to my experience, for those who have played with the code does it look very PIC specific? Would it be a hard port to AVR C?

I have not used AVR but I can make some general statements...
There are two versions of the code now, both written for the PICs. The original is for the Hi-Tech compiler and the later versions are for the XC8 compiler. My understanding is that both compilers are/were available from Microchip, but that XC8 is their current version. For some reason XC8 is quite different from Hi-Tech though - Hi-Tech has lots of useful library functions that are not included in XC8. XC8 has a free version, Hi-Tech does not. Hence the port to XC8!

It took a fair bit of effort to convert the code from Hi-Tech to XC8. It meant that I had to learn PIC in great detail before I could work out exactly what the Hi-Tech code was doing. I would suggest that you would have the same problem porting to AVR? You may need to learn PIC to work out what the code is doing, so might as well have used PIC anyway! Having said that, I did try to write the code in a portable way, but I am not the best coder in the world (or street), so there are probably lots of PIC specific things scattered around.

[Astro]

Originally Posted by 4Springs

I have posted details of my component purchases on the Sourceforge website (look under Files -> Components). Modules were about $4 each, Master about $50. These prices are parts only without the board itself and without the PIC microprocessors.

I managed to do my cell top modules for AUD$3.66 each (including PCB, components, PIC micro and all shipping to Australia).
I used Diygsm they made very nice boards and shipped via DHL with a tracking number. They double checked the files i sent them and let me know when something wasn't quite right.

I don't have the figures for my master boards but i would estimate less than AUD$10 each. The masters PIC chip was USD$1.72 compared to the cell tops AUD$1.32 PIC chip.
The master PCBs were AUD$2.40 compared to cell tops AUD$0.98 per module.
The only other significant difference would be the voltage conversion chip and the LCD.
I can't remember what the voltage conversion chip cost but it wasn't much and the 16x2 line LCD was AUD$2.25 from eBay. I upgraded my LCD to the 20x4 line version which costs AUD$7.23.

There are other expenses like solder paste that i haven't included.
Mainly because i made over 50 cell top modules and 3 master modules and still have heaps of paste left in the syringe. So it will be a shared expense over many projects.

So i think the BMS really does live up to the threads title of "Low cost BMS".
When i was shopping around for a BMS system i was seeing commercial cell top prices around $14 with the master module with display around $250.

Commercial 48 cell system = $922
Low cost BMS 48 cell systems = $185

It all depends on how you value your time, if you are comfortable with DIY and how much satisfaction you get from knowing you did it better and cheaper yourself.

[thingstodo]

Originally Posted by arber333

However system is designed to monitor cells even if transistors burn out. I would get plenty of notice.

This sounds like a good feature. Monitoring only.

I'm a fan of bottom balancing, but I would feel better if the cell voltages were monitored.

I am not confident enough in my DIY fabrication skills to put a board on the batteries. I would likely INCREASE the chance of a battery fire instead of DECREASE.

Instead of replaceable fuses, would *REALLY* thin wire be better protection? It would take some experimenting to determine how small it would need to be. AWG #40 is the smallest that is listed in WIKIPEDIA here https://en.wikipedia.org/wiki/American_wire_gauge

That way if the insulation rubs off, or anything goes wrong in a vibrating, moving vehicle, the thin wire just melts and the short circuit clears. A small amount of damage but no battery fire! AWG #40 lists 8.5 amps for 32 ms ... I assume longer would melt it - which is the idea.

[thingstodo]

Originally Posted by arber333

What kind of cells you use? I still have spare PCBs for LiFe.

I'm using Leaf cells - lithium polymer - so the boards likely won't fit without some modifications.

[thingstodo]

Originally Posted by arber333

OH, how did you get them? Were they new or used?

Used - I got one of the Renault/Better place packs (same as the leaf cells) from the Better Place bankruptcy, through EVTV ... 2 years ago?

I can send you PCBs. You populate sensor side and connect wire link. You can just omit transistor and everything will still work. It just wont shunt...

Great - less work for me! PM me how much you are willing to sell them for. I have 48 cans with 3 terminals each. 0V, 4V and 8V.

Oh and idea about fuse wire isnt new.

Very little that I write about is new. I read about most stuff and forget where I got the info from ... makes it hard to credit the originator! But I should make more of an effort to.

Original plans have + sensor supply wire in case chip would melt. That wire burnt out in number of ocasions when i experimented/tortured it, so it works!

So what gauge wire did you use for the 'fuse', and was it copper, aluminum, or some alloy?

I find that sort of practical stuff very valuable

[thingstodo]

Originally Posted by arber333

Huh i will have to reflect on that. My BMS is made as 8S that means 9 wires. If you connect modules so each 8V terminal is connected to next 0V terminal you need 4 modules for 9wire link yes? Since you have equivalent of 96cells that means you will need 12 PCB plates. That is some volts - 360VDC! Do you intend to connect them all in series? This BMS can support 255 cells in series

The original idea was to have all in series. I had a Netgain Industrial controller. Up to 450VDC in, programmable voltage out (was set to 190V) and max out at 1200 amps. Liquid cooled ... and broken! I killed it by not connecting the liquid cooling before doing my testing.

Now I'm looking at using an older Cougar (P&S) so it tops out at 144V and 500 amps. If I use 3 strings of 128V, I won't stress the batteries much. 32 cells is divisible by 8, so that works out nicely.

You are in luck since i also need at least 3 more PCBs. I will order PCBs and when i get them i will resend them to you OK? Ill PM you for your address and price...

I'll check out the link and see what these boards look like ...

Its just PCB copper thin enough to only carry an amp or so. No need to be precise. When module shorts it draws plenty more than an amp!

Traces .. that works!

[thingstodo]

Originally Posted by arber333

AH! I recommend you repair that ontroller since usually its just the IGBT and driver chip that are destroyed. Its like $100 and some skill.

It would be GREAT if I could fix the controller. I just need to reverse engineer the circuit and figure out what part of it stopped - the transistor stage, the optocoupler, or the components in the big black block that I can't see into.

I made my 3phase controller and i had water cooling working, but i burned IGBT still when i reved motor and released clutch, bam!
I simply replaced IGBT and affected driver chip together with its DCDC supply brick. Took like 2 days since i had everything on stock.

Were there scorch marks? How did you know it was the IGBT? Is there a simple test I can do to test the IGBT?

[thingstodo]

Originally Posted by arber333

Oh, you have a special driver board? And the way you say it, driver chip is glued in some resin? Can you send a photo of the driver board? The one that is connected to IGBT.

The DC controller is a commercial Product - Netgain Industrial Controller NetGain WarP-Drive Classic Controller - there is potting compound on some of the circuit, or parts of the circuit.

I have no schematics. I'll get the pictures posted tonight.

Yes when i heard bang i knew one of the motor winding coils backfired and IGBT got it all. Poor thing exploaded. Even the top blew off. Yes scorch marks were visible, but i took all 3 IGBTs off and measured them just in case.

I had a big bang as well. No scorch marks visible. It may be possible that the BANG was the fuse element blowing apart? I expected there to be scorch marks, shrapnel, etc but the inside of the controller is quite clean?

Usually only the top transistor in IGBT is toast.
Drivers i tried with my PSU and if the chips draw more than couple mA something was wrong. With the chips i use sometimes even the DCDC converter is toast, but not allways. So whatever you find you check the pricing and decide.

I need to make an extension cable so that the controller is connected to external signals - contactor outputs, throttle signals, precharge voltage, etc - while the controller is in pieces. That way I can see how far the signal makes it before it stops - transistor, optocoupler, or blob.