Home grown BMS ideas !

Post by **acmotor** » Fri, 16 Jan 2009, 21:48

Johny said to start this topic....

It started as a stain on the page in.....

hyjacked topic

Go for it...

Here is my circuit again just to start the topic here.

NOTE. Transistor used in final version was KSH127

Post by **jackhyq** » Fri, 16 Jan 2009, 22:05

just circut or can do that by yourself ?

Post by **zeva** » Fri, 16 Jan 2009, 22:19

Ah what a coincidence, I've started playing with BMS design myself this week..

As many of you have heard, I'll be building a new pack for the MX5 over the coming months (based around K2 26650EV cells) and will be doing a homebaked BMS for it myself. Personally I plan to construct 4S30P modules, and will have a single BMS per module with a microprocessor monitoring the four cell voltages and triggering shunt transistors/resistors through optoisolators. And probably communicating with some kind of master unit (for monitoring voltages, controlling a charger or warning about low cells etc) over I2C serial. Well that's the theory anyway, time will tell if it actually works out! Will post more info as it happens.

I'm certainly interested to hear what others may have done or are planning to do for their BMS too, in case I may want to steal their ideas

Post by **acmotor** » Fri, 16 Jan 2009, 22:28

Jackhyq,

I can't claim all the fame.
The circuit is mine (well based on the TI app. note for the TLV431) but a friend layed out the PCB. (master of the art)
The aim was to have no adjustments required as the zener are very accurate 1.250V. This has worked.
I've tested 4 out on TS 40Ah and now assembling the others.
Only 216 to go ! edit: 52 to go now
My early BMS were pic-axe micro based, but worked out too expensive.
This one should come in under $10 all up.

Hoops is probably on the track by using 1 micro to control / monitor multiple cells.

I've run a module of 4 cells with voltage monitoring via resistor tree (and software subtraction). That works fine but the idea needs to be done over 20 cells.

Post by **Johny** » Fri, 16 Jan 2009, 22:29

OK. I stayed away from prog.zeners because I thought that dual or quad comparator would be cheaper, but I see they are $UA0.12 on Octopart so your system looks pretty inexpensive in parts.

I'll look at a quad comparator and prog.zener and see if I can get the parts count down. A single 14 pin DIP is simpler to load if they are to be DIY built.

A minor variation would also work for lead-acids (my back up plan with the Oz$ going dooowwwnnn). Reason being that I see that "smart" lead-acid chargers uses low current 0.02C to sense if the battery has fully charged then switch to float. My concern is that the shunt zener-resistor system may prevent that switch. We would need an opto input bus as well to switch the eq. off for charging. Do eCrazymans's chargers work OK with the zener-resistor system?

Edit: Added 'a'

Post by **acmotor** » Fri, 16 Jan 2009, 22:50

Johny,

Do go for surface mount as any through hole is potentially a mechanical issue on the underside (cell side) of the PCB near cell terminals.

I started with comps.and zener but the prog zener has the comparator built in.TLV431

Re the SMPS chargers on red suzi SLA and zener/res eq.
See red suzi about the float charge.

Post by **jackhyq** » Fri, 16 Jan 2009, 22:53

Acmotor, That's great !

Post by **acmotor** » Fri, 16 Jan 2009, 22:59

The best thing about building something and posting it is that others will know what not to do !

... or at least have fun flaming it !

Post by **Johny** » Fri, 16 Jan 2009, 23:11

OK on surface mount - I guess I'm a Luddite. Im a bit used to doing DIY PCBs in single so bias toward DIP - I must change.

I see the float reference but my point was that some chargers wait at 14.2->14.5 Volts for the current to drop to less than 100mA which isn't going to happen with a 200mA drain via the zeners.

Post by **woody** » Fri, 16 Jan 2009, 23:19

LED + Darlington Clamper would have to be the minimalist approach.

They also mention an alternative: Zener Shunt Regulator

Using LEDs is attractive to me, easy to see which cells are charged

Also could use a similar system to limit under-voltage I guess.

Can you get opto-isolators with a 2.5V cutout to use for a similar low-voltage cut-out circuit?

Post by **acmotor** » Fri, 16 Jan 2009, 23:20

I'm not a fan of the overcharge to 14.5V.
I just give it another 8 hours at 13.6 to 13.8 and it is charged anyway.
Chill out, be patient !

The simple zener/res. bms is not for a 'smart' charger.

Post by **acmotor** » Fri, 16 Jan 2009, 23:31

The problem I found with normal zeners on SLA 12V level was the need to 'select' the voltages out of a batch of zeners. These varied from 6.4 to 6.9 V on '6.8V' zeners.
The prog. zeners are very accurate up front.

There are many minimal systems for lead acid (I use one) however I am not content with minimal on the TS !

Post by **woody** » Fri, 16 Jan 2009, 23:42

What are the important/unimportant BMS features?

These are the ones I can think of:
0. Safe
1. Overvoltage protection while charging
2. Overvoltage protection while regenerative brakeing
3. Undervoltage protection while driving (buzzer or cut throttle after 20 seconds or what?)
4. cheap (seems silly to spend more on the BMS than the cell)
5. efficient
6. turn off charger when charged
7. optimal charge curve for long cell life
8. identify faulty cell
9. bypass faulty cell / cell group

For 9, if you have 1 reversing contactor per cell block of 48V (60-72V acmotor?), the reversing contactor connects the 48V block either to the 48V bus for charging, or in series with the rest of the blocks to make the 600V or whatever. In case of failed cell in the block, the BMS could take that block out of the series, giving you 200 cells instead of 220 for your VFD, still very drivable.

Post by **acmotor** » Fri, 16 Jan 2009, 23:58

All important.

0 Safe... excludes bipolar /fet junction directly across cell ! and resistor whose power rating is exceeded by full on pwm from micro.

7 optimal charge curve. how simple is Constant voltage current limited ?

8 can be just visual at the cell ?

9 interesting... are you thinking of this while pack powered down only ? Just use a mechanical SPDT switch ? can be AC rated as it will not be operated with live DC.
Change-over 600VDC relays may be expensive.

Post by **woody** » Sat, 17 Jan 2009, 00:13

Same site has links to about 15 differnet BMSes available, in progress.

Post by **Johny** » Sat, 17 Jan 2009, 00:18

acmotor. What charger(s) are you using for Red Suzi's lead-acid now? I though they were eCrazyman. Aren't his "smart"?

Post by **woody** » Sat, 17 Jan 2009, 00:50

acmotor wrote: All important.

0 Safe... excludes bipolar /fet junction directly across cell ! and resistor whose power rating is exceeded by full on pwm from micro.

acmotor wrote: 8 can be just visual at the cell ?

I think is both simplest and best. (hard to stuff up)

acmotor wrote: 9 interesting... are you thinking of this while pack powered down only ? Just use a mechanical SPDT switch ? can be AC rated as it will not be operated with live DC.
Change-over 600VDC relays may be expensive.

I saw a4x4kiwi contactors which split his pack into 12 x 48V packs and a 24V pack for charging. It made sense for me to join them all in parallel to 1 48V pack for use with a big charger.

The way to do this was DPDT contactors:

    48V Bus + ---O  O--- + 48v block - ---O  O--- - 48V Bus
                   /                        \
prev 48V block ---O  O-----(bypass)-------O  O---  - next 48 V block

When the contactor was energised (show), the blocks join together to make the 600V chain, when unenergised, they join onto a common 48V bus for charging.

The reversing contactors have 4 terminals on each end. I thought they were double pole double throw, but they aren't are they, a reversing contactor is pretty much quad pole single throw, which you can wire as double pole double throw, I don't think I can do the bypass thing anyway unless I use even more hardware.

I think your manual cutover is a better option.

Post by **woody** » Sat, 17 Jan 2009, 01:10

Johny wrote:Do eCrazymans's chargers work OK with the zener-resistor system?

I think a4x4kiwi is testing that for us at the moment!

I think his first full charge never went from CC to CV, but I can't be sure.

Also one charger let out the magic smoke...

Post by **woody** » Sat, 17 Jan 2009, 01:14

Same site's take on requirements.

Post by **acmotor** » Sat, 17 Jan 2009, 01:33

Woody, if the DPDT contactors were rated at 600V I would feel at ease with the module parallel switching / bypassing idea. Otherwise I would be taking great care with the operation !
At zero current you may well get away with it.

The chargers I have ended up with are fixed voltage, current limited (actually ecrazyman power supplies). I would need a more effective BMS than simple zener if the charger were taking V up to 14.5V for a faster charge.

Post by **Mesuge** » Sat, 17 Jan 2009, 03:07

woody wrote: Same site has links to about 15 differnet BMSes available, in progress.

Apart from BMS and his MiniEV build, there are couple of interesting CANbus links as well (lower folder level).
http://carrott.org/cgi-bin/twiki/view/BMS/WebHome

Post by **a4x4kiwi** » Sat, 17 Jan 2009, 04:00

@Woody, All chargers are smart chargers and have worked successfully with the Zener Resistor balancer.

The Zeners and resistors do get quite hot to touch before the charger switches to float. I have an extractor fan on the trey that is powered during charging.

Post by **Johny** » Sat, 17 Jan 2009, 16:53

Are there any specs on the chargers - like current required for float switch? I'll assume not so how hard would it be to gently test a couple of them to determine what the current at switch point is? I suspect this might have to involve a longish time (a few seconds at each current) unless they don't bother to "smooth" the current reading.
My concern is that the zener-resistor system is near the threshold and some switch but some don't - result, unbalanced 600V pack.

Post by **jackhyq** » Sun, 18 Jan 2009, 01:57

Johny wrote: Are there any specs on the chargers - like current required for float switch? I'll assume not so how hard would it be to gently test a couple of them to determine what the current at switch point is? I suspect this might have to involve a longish time (a few seconds at each current) unless they don't bother to "smooth" the current reading.
My concern is that the zener-resistor system is near the threshold and some switch but some don't - result, unbalanced 600V pack.

Hi Johny,I have specs of our charger,do you still need ?

Post by **a4x4kiwi** » Sun, 18 Jan 2009, 02:03

FYI here are my charger specs. They are from e-crazyman on ebay.

I was concerned that they would not turn off with the zener / resistor across them, but all have been fine.

input voltage: 98~245V AC 50/60Hz
power efficiency: 90% +/- 2%
size: 17.8*10.4*7.4 CM
indicator method: LED
working temperature: -5~+40 Degree Celsius
cooling method: FAN
output voltage:55.6 +/- 0.4V
output current:DC 2.5A +/- 0.1A
battery fits: 17-20AH
AC wire length: 71.5 CM (i will changing to AU wall plug)
DC wire length: 83 CM
13 units to AU price is 306 usd included shipping