Even Newer 4 to 24-cell Battery Management System (BMS)

[AndyH]

Has anyone built a V2.6 board yet? I'd like to compare notes with someone that has one working.

Thanks,
Andy

 

[fechter]

Hello all of you,

This is my first post at this amazing forum, after many hours of reading from the side. It is a pleasure to see the large activity in the DIY EV world. There is not that much going on it the area here in Denmark yet, so forums as information rich as this one are gold mines. Thanks to all the great contributors.
-Niels

With a name like ComponentKiller, you'll fit right in here.

It seems your chargers have a built-in shutoff feature that is being tripped by the throttling action. The BMS circuit was designed to work with a CC, CV power supply with no brains. Your chargers have a feature that does not interface nicely with the throttling circuit.

With very large cells, a 10hz switching speed would be within the 'normal' range, but most run a bit faster. You could try removing C3 to see if that makes it run faster. If you can get the frequency higher, it may work better with your charger. If C3 doesn't speed it up enough, try removing C2. If you can get the frequency higher, it may solve both problems (it might not also).

Since your charger has an auto shutoff feature, it may be better to disable the auto shutoff on the BMS board. This could be done by disconnecting D1 on the control circuit. If you try this, you'll have to adjust the charger voltage so the current drops off just as the last cells light up.

If this approach still does not work, it may be necessary to interface BMS directly to the charger circuit, which could be a bit complicated. This is really the best approach for very large cells, as it eliminates the FET on the control circuit and allows for unlimited charging current. It also eliminates the need for any oscillation in the control circuit, since you use the PWM that's part of the charger.

The first time you try to balance a pack, it can take a long time (days). It may help to use a single cell charger on the low cells to speed up the process. After you get the pack balanced, the balancing time should be much shorter after that.

The issue with the first cell results from the voltage drop in the wire going from the BMS board to the cell. The voltage drop in the wire adds to the cell voltage, causing it to get more charge. You could minimize this by using heavier wire for that connection. Normally this amount of voltage difference is not a real problem, but it does result in odd behavior. You may notice a similar effect on the highest cell.

 

[steveo]

Hey Fechter

I have measured the ground and all shunt & shunt per channel .... all read .5v ...
..........
I put a multimeter to the d1 .... and i get a reading both ways i put the multimeter... where as on working board .. i only get a reading in one direction...

the main led sit in an orangish red colour... and not pure red .. when its below the HVC mark ... something is odd here ..

-steveo

Yep, something is odd here.
I would try removing D1 and see if you can get the main LED to go solid red when its below the voltage set point.

With D1 removed, you could use the diode check.

Installing D1 backward should not blow anything.

.5v on the all shunts line is low enough to trigger the SCR.

You could first try measuring voltage from the control ground to both sides of D1. When the SCR trips, the side going to the SCR should pull low. When the board first comes on, that spot should be near 12v.

Hey Fechter,

I have measured the control ground & both side of the d1 diode.. The top side of the diode measures 1.4v ... the bottom of the diode measured .75v - this done with the bms below the HVC mark

With the Bms at the HCV point .. i measure .53v on the top of d1 .. and .74v at the bottom of d1

with d1 removed ....the light still stays orange below the HVC mark

thanks
-steveo

Bump ... .. Fechter ?

 

[fechter]

Bump ... .. Fechter ?

Sorry, it's hard to keep half a dozen different troubleshooting threads organized.

Try powering the board and short pins 7-8 on any optocoupler. This should make the main LED turn green when the pins are shorted and go back to whatever it was when you remove the short. This tests the gate driver chip and some of the related circuitry. The voltage on D1 seems too low.

Double check the polarity of the capacitor on the gate driver input (C2 on ver. 2.4 and 2.5 boards). If this is backward, it could drag the line down and prevent the output from going on fully.

It seems the 'Any Shunt' line is being held down by something. You could also try powering the board and measure the voltage on each cell circuit at the junction of D101 and D102 against control ground. When the cells are below HVC, this junction should be near 11-12v. If any cell circuit measures low at this point, it indicates a problem in that circuit. As the voltage increases and the cell LED lights up, then this junction should pull down to 0.5v or so.

 

[steveo]

Bump ... .. Fechter ?

Sorry, it's hard to keep half a dozen different troubleshooting threads organized.

Try powering the board and short pins 7-8 on any optocoupler. This should make the main LED turn green when the pins are shorted and go back to whatever it was when you remove the short. This tests the gate driver chip and some of the related circuitry. The voltage on D1 seems too low.

Double check the polarity of the capacitor on the gate driver input (C2 on ver. 2.4 and 2.5 boards). If this is backward, it could drag the line down and prevent the output from going on fully.

It seems the 'Any Shunt' line is being held down by something. You could also try powering the board and measure the voltage on each cell circuit at the junction of D101 and D102 against control ground. When the cells are below HVC, this junction should be near 11-12v. If any cell circuit measures low at this point, it indicates a problem in that circuit. As the voltage increases and the cell LED lights up, then this junction should pull down to 0.5v or so.

Hey Fechter,

You do an awsome job at maintianing the site nevermind the threads!!!

Here is the update.

If i short pins 7 & 8 for each optocoupler the light turn redish-orange colour, but not red ..
I checked c2 to make sure its installed correctly ( this is the tiny capasitor beside the chip " IXYS ").. I even changed this !
I checked voltages of d101 & d102 ..against ground .. The voltages read 8.25v Below & at the HVC...
If the main led is red when i plug in the power (rarley occurs) .. When i check to voltages agains d101 & d102 ... they sit around 10.25v .. but soon as i crank the voltage to hit the HVC .. again it drops to 8.25v

hope this helps

thanks
-steveo

 

[fechter]

If i short pins 7 & 8 for each optocoupler the light turn redish-orange colour, but not red ..

thanks
-steveo

Hmm.... when you short pins 7-8 on any opto, it should make the main LED go green. I think you need to solve that problem first.

Measure the voltage on pin 4 of the gate driver before and during shorting pins 7-8 on an opto.
It should be around 3v to start, and less than 0.5v when pins 7-8 are shorted.

The voltages on the junctions of D101-102 sound correct when its below HVC, but it should drop more when you hit HVC. Getting an accurate reading on these may be tough if it's oscillating, since most multimeters don't behave well with a square wave. An oscilloscope would be much better.

You could try measuring the junctions of D101-102 on each cell while shorting pins7-8 on the opto for that cell. Shorting the opto will get rid of any oscillation on that cell. When you do this, it should drop down to around 0.5v.

It almost sounds like something is holding the any shunts line high.

 

[Doctorbass]

Hey Fechter,

I just wanted to thanks you very much in the name of everyone here! :wink:

You work so hard to help us here and I want you to know we see that and appreciate that.

Doc

 

[steveo]

Hey Fechter,

I just wanted to thanks you very much in the name of everyone here! :wink:

You work so hard to help us here and I want you to know we see that and appreciate that.

Doc

I agree with doc ... how would are live be without the Es Forums ..

can you imagine??

Fechter kind of remind me of bill nye the science guy LOL

http://www.youtube.com/watch?v=WDW0IqwMK00

-steve

 

[fechter]

Thanks for the support guys, but you do most of the testing work for me.

Maybe if the design of my circuit was better, there wouldn't be as much need for troubleshooting

This circuit is far from ideal in many ways, but does the job for now.

I am currently looking at improvements that should make it better without adding too much cost.

The auto cutoff part of it is the most problematic, and I'm looking at some new approaches to implement it. The throttling thing kind of sucks too, but I think that can be made much better without too much work. I just wish I had more time to work on it.

 

[steveo]

Thanks for the support guys, but you do most of the testing work for me.

Maybe if the design of my circuit was better, there wouldn't be as much need for troubleshooting

This circuit is far from ideal in many ways, but does the job for now.

I am currently looking at improvements that should make it better without adding too much cost.

The auto cutoff part of it is the most problematic, and I'm looking at some new approaches to implement it. The throttling thing kind of sucks too, but I think that can be made much better without too much work. I just wish I had more time to work on it.

Hey Fechter

I want to make a suggestion if i may.

Would it be possible to eliminate the possibility of blowing a channel if one battery wire disconnects?... Alot of use use connections between the bms & battery balancing leads..

I've blown a few channels by this happening .. and i've been discouraged to contiue using it ..

-steveo

 

[AndyH]

I've built a pair of V2.6 boards so far and cannot get the auto-shut off to work in it's specified configuration. Until it's figured out, I recommend installing the capacitors for the control section (C2 and C3) in the V2.5 configuration.

The caps are easily added to the bottom of the board, but can be added to the top if you'd prefer.

The control circuit also works with just the addition of the .1uF cap across pins 4+5 of the IXDF404 gate driver and without the additional 10uF cap (C2 in the V2.5 circuit).

Not elegent, but very functional. Andy

 

[fechter]

Yes, Andy has discovered some problems with the changes in the ver. 2.6 boards. I'm hoping to find an easy solution to that so they can be used without needing to rewire them. Worst case, rewiring them like the 2.5 boards will work.

Ultimately, I'm hoping to come out with a whole new charge control circuit that should (in theory) work better than the existing 'hack' and address some of the issues that persist with the design.

I wish I had more time (and money) to work on it....

 

[brainzel]

Yes, Andy has discovered some problems with the changes in the ver. 2.6 boards.

Worst case, rewiring them like the 2.5 boards will work.

Oh I began to solder the 2.6 boards yesterday. Should I rather wait untill you matched and fixed the problem?

 

[ejonesss]

you mean like a quick disconnect?

and how does disconnecting a wire blow a channel?

Thanks for the support guys, but you do most of the testing work for me.

Maybe if the design of my circuit was better, there wouldn't be as much need for troubleshooting

This circuit is far from ideal in many ways, but does the job for now.

I am currently looking at improvements that should make it better without adding too much cost.

The auto cutoff part of it is the most problematic, and I'm looking at some new approaches to implement it. The throttling thing kind of sucks too, but I think that can be made much better without too much work. I just wish I had more time to work on it.

Hey Fechter

I want to make a suggestion if i may.

Would it be possible to eliminate the possibility of blowing a channel if one battery wire disconnects?... Alot of use use connections between the bms & battery balancing leads..

I've blown a few channels by this happening .. and i've been discouraged to contiue using it ..

-steveo

 

[fechter]

Oh I began to solder the 2.6 boards yesterday. Should I rather wait untill you matched and fixed the problem?

Yes. You can build all the cell circuits and the voltage regulator, but I would try and hold off on the part around the SCR and gate driver until we figure out what's needed.

 

[AndyH]

Oh I began to solder the 2.6 boards yesterday. Should I rather wait untill you matched and fixed the problem?

Yes. You can build all the cell circuits and the voltage regulator, but I would try and hold off on the part around the SCR and gate driver until we figure out what's needed.

I can confirm that the two V2.6 boards that I've built work perfectly when assembled with the V2.5 capacitor configuration. My third V2.6 board is going together today and I'll try a couple of Richard's suggestions and report back.

Regardless of which version one might be building, I highly recommend using the V2.6 RX01/RX02 resistors - the 120K/330K combination does a beautiful job of reducing cell shunt levels below 3.7V and reducing the overall voltage required from the charger to get all the shunts on. Earlier boards required a charger capable of 3.7V per cell plus about .5V as earlier reported. I'm seeing latching with as little as 3.65V per cell. Very nice!

 

[fechter]

Good work Andy. I still have a pile of parts and bare board.

I really think changing the feedback capacitor (C2 I think) to something way smaller might do it.

It would be great if you could try some of those things I mentioned and let me know what happens.

 

[ejonesss]

i dont know if this is the right place to mention this so you mods and admins are welcome to move it to the right place.

while doing my rounds on various web sites i ran across a new way to make circuit boards (home etching)

http://hackaday.com/2009/09/02/direct-to-pcb-inkjet-printing/

the above method uses a normal printer and looks like a cd/dvd label printer kit.

the trick is getting it lined up.

this would be useful for prototyping where there would be lots of bad boards due to mistakes and such.

 

[pm_dawn]

Hi !

I really would like to use the Anyshunt and Allshunt lines to control parts of the charging in my car.
How can I tap into these lines, I mean what components should I use to for example pull a small relay.

Best REgards
/Per

 

[fechter]

Hi !

I really would like to use the Anyshunt and Allshunt lines to control parts of the charging in my car.
How can I tap into these lines, I mean what components should I use to for example pull a small relay.

Best REgards
/Per

The any shunts line runs at a very high frequency, so I don't think you could use it to drive a mechanical relay.

Both lines are normally high (12v with respect to charger negative) and get pulled low when active. You could tie in an opto coupler between the 12v line and the any shunts line (with a 1k resistor in series) to drive an isolated output at high frequency.

You could also get the any shunts signal off the drive for the green LED. If you want a logical opposite, use the red LED. This may be better than tying directly into the lines as the gate driver will prevent the load from affecting the circuit.

The all shunts line could trigger a mechanical relay by placing the coil across R5, the resistor that connects to the anode of the SCR. When the SCR triggers, it could also operate a small 12v relay coil. The relay coil needs a diode across it in the normally non-conducting direction to absorb the spike when the coil turns off.
In order to do this, I think the voltage regulator would need a higher current rating or you'd need to use a separate 12v supply to power the control circuit. If it was a very small relay, the existing regulator might be OK. The SCR is rated for 500ma. The on board regulator is rated for 100ma, but the circuit takes 30ma or so. If you know the relay coil resistance, you can calculate the required current.

If you want to run a large relay, you'd need a separate transistor driver for the coil.

 

[pm_dawn]

Ok Thanks alot fechter!

Another question.
I was benchtesting the BMS yesterday And I could not get latch up.
I had no cells and powered only the first four cells. I had like 22V as max voltage, I could see that the LED got greener and greener but would not latch.
Any ideas ?

Also number two:
Can the LVC be used directly on a hall sensor type throttle?
Grounding the output signal from it?

Best Regards
/Per Eklund

 

[fechter]

Ok Thanks alot fechter!

Another question.
I was benchtesting the BMS yesterday And I could not get latch up.
I had no cells and powered only the first four cells. I had like 22V as max voltage, I could see that the LED got greener and greener but would not latch.
Any ideas ?

Also number two:
Can the LVC be used directly on a hall sensor type throttle?
Grounding the output signal from it?

Best Regards
/Per Eklund

It won't latch until all the cell LEDs are lit. You can test by grounding the all shunts line with a jumper.

Yes, you should be able to tie the LVC line directly across a hall throttle in most cases. Observe proper polarity. To test the LVC function, you can attach a 1.5v battery to any cell circuit and verify it kills the throttle. It would be safer to insert a 1k resistor in series with the throttle signal line and tie the LVC wire into the controller side of the line. The resistor will limit the current to a safe value for all the components.

 

[brainzel]

First a great thank to all and especially to Fechter and Gary for supporting!

I have too questions:
1) I have 38 cells and two 2.6 boards; should I solder 19 channels on each board, or should I rather solder one complete board (24) and the rest (14) channels on the other board?
(I changeed the 6.8Ohm/5W to an 3.9Ohm/5W for my 120Ah cells, as Gary told me.)

2) You wrote, that changing the C2 capacitor would fix the Problem, which one should I take instead of the .1uF from the original 2.6 board layout?

Thanks

 

[fechter]

First a great thank to all and especially to Fechter and Gary for supporting!

I have too questions:
1) I have 38 cells and two 2.6 boards; should I solder 19 channels on each board, or should I rather solder one complete board (24) and the rest (12) channels on the other board?
(I changeed the 6.8Ohm/5W to an 3.9Ohm/5W for my 120Ah cells, as Gary told me.)

2) You wrote, that changing the C2 capacitor would fix the Problem, which one should I take instead of the .1uF from the original 2.6 board layout?

Thanks

You could configure the boards either way, but I would recommend the 24/12 arrangement since it will allow you to use complete sections.

We're still working on the capacitor issue. I'd suggest 0.01uf at this point. R3 on the control circuit should also be changed from 10k to 4.7k (you can use two 10k in parallel).

 

[brainzel]

OK thank you, Fechter.
I will order 0.01uf capacitors and 4,7k Ohm resistors and take a full 24 and a 14 channel board.