Low cost BMS

How do you store and manage your electricity?
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Low cost BMS

Post by Nevilleh »

Thanks for the suggestion, but the .jpg image was 130 Kb and the .gif one is 2 Mb which is too large for this system. Beaten every which way! I'll have to try something else.
Printed it, re-scanned it at 100 dpi resolution and saved it as a gif. Let's see what we have :

No, same error:
Error uploading file!!.
The file failed the security scan and has been deleted as it may contain malicous code
Last edited by Nevilleh on Sat, 19 Jan 2013, 05:57, edited 1 time in total.
User avatar
woody
Senior Member
Posts: 1716
Joined: Sat, 21 Jun 2008, 02:03
Real Name: Anthony Wood
Location: Mt Colah

Low cost BMS

Post by woody »

Png is a good format.
You can take a screenshot with the print screen button, paste that into ms paint, crop, save as png. (Windows)
Ubuntu can do it too
Planned EV: '63 Cortina using AC and LiFePO4 Battery Pack
Renard
Groupie
Posts: 219
Joined: Sun, 29 Aug 2010, 18:55
Real Name: Robert Fox
Location: Cobargo NSW

Low cost BMS

Post by Renard »

My experience with the cell-top modules:

The micros came pre-programmed from Microchip Direct. But they charge US$29 for a set-up fee and US$60 minimum charge for programming. Decide for yourself if it's better to buy the PIC programmer for some $50 or so.

The resistors came mostly from Rockby who often have some special deals. Also the 2-pin headers.

The less standard parts came from Mouser -- the LM385M3X-1.2NOPB, the 6R8 (RK73BATTE6R8J), the DNBT8105-7. Cheaper to use their site in US$ not A$.

The optocoupler VO615A-3X019T and the thermistor NB12K00103KBB came from element14. There may be a cheaper substitutes, but I was using Neville's choice as I knew he had got them to work.

If you want to work on getting the price down, concentrate on the micro, the optocoupler, the thermistor and the LM385.

As for the oven, I don't think my experience was a happy one, but then my oven was old and had only one element.

The master unit Neville sent me is programmed for bottom balancing.
Renard
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Low cost BMS

Post by Nevilleh »

I bought all my parts from Element14 and RS Components. I find the shipping charges from Mouser are quite exorbitant.
The PicKit 3 programmer is good value direct from Microchip and its easy to use. You plug it and click on a couple of buttons. That lets you experiment with different settings eg balancing. And you don't need to do any programming if you don't want to, I'm happy to send .hex files!

Here's the data I tried unsuccessfully to upload the other day:

Image

Data output from bms to netbook, processed and Excel'd to make graphs. Shows a short drive, backing out of the drive, accelerating gently then some full throttle acceleration. You can see cell #9 is sagging a bit!
Last edited by Nevilleh on Sun, 27 Jan 2013, 07:19, edited 1 time in total.
User avatar
4Springs
Site Admin
Posts: 674
Joined: Thu, 23 Dec 2010, 01:14
Real Name: Christopher Walkden
Location: Selbourne, TAS

Low cost BMS

Post by 4Springs »

Hello Neville,
I've downloaded DesignSpark, and had a go at re-designing the Cell Top circuit board to suit the SE130AHA cells. These cells use M8 bolts, 106mm apart. So my board has these changes:
1. Moved the + terminal to the left. Changed the terminal sizes to 13mm OD, 8mm ID.
2. Moved the balancing resistors to the left to get them a bit further away from the other components.
3. Moved a track on the thermistor, to give it better thermal conductivity to the negative terminal.

How hot do your resistors get with top balancing? There is enough room to increase the number of resistors, or the spacing if required. I will have very little airflow around my cells, and the BMS will be at the top, where it is hottest.
Because I don't have any of the cells to test with, I am looking intently at pictures. There is a lump on the cells between the terminals - I can see it on yours as well. Does this get in the way at all? Should I attempt to put the balancing resistors away from it?

Image

If anyone can see any further improvements, or problems with my improvements, please let me know! Now is the time to make changes...

DesignSpark is reasonably easy to learn. And it runs quite well on my Puppy Linux machine under Wine.
Renard
Groupie
Posts: 219
Joined: Sun, 29 Aug 2010, 18:55
Real Name: Robert Fox
Location: Cobargo NSW

Low cost BMS

Post by Renard »

I presume the lump you refer to is the vent plug. It doesn't impinge on the board.

I suppose that more balancing resistors would be good, but with more resistors you may have to consider upgrading the transistor which switches them.

I note that Neville's existing board produces about 2W when top balancing.
Renard
User avatar
4Springs
Site Admin
Posts: 674
Joined: Thu, 23 Dec 2010, 01:14
Real Name: Christopher Walkden
Location: Selbourne, TAS

Low cost BMS

Post by 4Springs »

Renard wrote:I suppose that more balancing resistors would be good, but with more resistors you may have to consider upgrading the transistor which switches them.

I note that Neville's existing board produces about 2W when top balancing.

I was thinking more along the lines of the same resistance over more resistors. This would keep the same current, but would be dissipated over a wider area, meaning each resistor would be cooler. Only if heat is a problem though, I seem to remember Neville saying that they got quite hot, and that was with bottom balancing, which would be a lower voltage.
arber333
Groupie
Posts: 56
Joined: Mon, 11 Mar 2013, 02:29
Real Name: Arber
Location: Slovenia
Contact:

Low cost BMS

Post by arber333 »

Hi,

I am currently experimenting with Nevilles BMS, the low balancing version. I have two issues now. I have 40 cells and 130VDC system.
System works very well and shows all parameters untill i press the pedal and drive. At that time master shows over 300V and over 60 cells!!!
I have three banks of cells and like 3m of 7wire shielded cable between them. I didnt put the shielding mesh to ground, should i? Would this be the problem here? Clearly EMI is causing comms go wild. BUT comms do not drop out!

Another capital problem is when i try to balance cells in bulk. Display normally shows "balancing 40" and then after a minute or so comms drop out! I mean i get "no comms" sign. I use 3R3 3W wirewound resistors with ceramic core. Should shunt cca 1A from 3V.
After i disengage balancing switch the comms return after a minute or so! Weird no? Maybe master optocoupler is giving trouble...hm?

Link to my build: mazdamx3ev.wordpress.com/tag/bms/
This site is in Slovenian, i intend to translate in the near future... as the time will allow.

Does anybody have any experience with this or suggestion what could cause this? I consider second problem more critical, as it means i cant even reliably balance my cells.

edit1: When i connected one cell to master it was balancing ok for cca 1hour, hm... i have to try individual banks and determine where the error is.
edit2: I had master on for some 2 hours with one cell module on a bench power supply and it shows normal balancing. Resistor is hot but not too hot. Transistor is very hot though. Must be shunting more than 1A.


A
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Low cost BMS

Post by Nevilleh »

The pcb tracks are not designed to carry 1 amp. Where did you fit the 3R3 resistor? The ones you describe certainly wouldn't fit on the pcb.

If the shunt transistor is getting hot, it probably doesn't have enough base drive. Measure the Vce while its on and measure the voltage drop across the base resistor, R5. You might have to reduce R5 down from 330R to get a bit more current.

As for the comms, try connecting only one bank of cells and use twisted pair - NOT shielded and certainly not "screened" cable with the screen floating. If that works OK, add a second bank similarly connected and so on.

I have 45 cells in 3 banks, all connected with twisted pair and there is no sign of EMI even at 500 A. The bank of cells in the boot has connections about 2m long.

Edit: I just looked up the data sheet for the transistor (DNBT8105) and it has a max Ic of 1A, a dissipation of 600 mW and it requires 100 mA of base drive to saturate it with 1 A Ic, so you are trying to use it at the limit of its ratings. Vce (sat) at 1 A is .5 V so you are dissipating 500 mW plus the base drive which should be 100 mA at 1.1V ie a total of 610 mW.
If you want to use such a high shunt current, you'd be better off with a mosfet as the micro port pin max sink or source current is 25 mA and that is not enough base drive for the bipolar transistor.
Last edited by Nevilleh on Mon, 11 Mar 2013, 01:46, edited 1 time in total.
User avatar
coulomb
Site Admin
Posts: 4324
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Low cost BMS

Post by coulomb »

Nevilleh wrote: ... use twisted pair - NOT shielded and certainly not "screened" cable with the screen floating.

Just curious Neville, what's the issue with shielded / screened cable? I'm guessing extra capacitance from one driven line to the other.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Low cost BMS

Post by Nevilleh »

I'm told that the screen can act as an antenna and actually increase the coupling of EMI into the signal wires, especially if they are not a twisted pair. Another thing people do is ground both ends and then the earth current flowing induces interference into the signal wires.
arber333
Groupie
Posts: 56
Joined: Mon, 11 Mar 2013, 02:29
Real Name: Arber
Location: Slovenia
Contact:

Low cost BMS

Post by arber333 »

Attached is my cell PCB. I connected 3R3 between fat tracks on the + side. Underneath i kept copper plating to dissipate heat.

So far with single and 4 pack modules there werent any problems. Cells were shunting happily at 1A. The trouble started with current master build. This is why i suspected optos in the start. Time will tell...
Todaj i will first connect shield to gnd and then went on to test separate pack shunting.

AImage
Last edited by arber333 on Mon, 11 Mar 2013, 17:45, edited 1 time in total.
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Low cost BMS

Post by Nevilleh »

Well, it looks like it should be all right. As I mentioned earlier, the transistor is probably not getting enough base drive to fully saturate at 1 A and the voltage drop down that track from the emitter to -ve is reducing it still further. You can reduce R5 to 160 ohms and that will give you 25 mA which is the max the port pin can source.
I used a bipolar transistor originally because of the low Vce(sat) but a mosfet might be a better choice, especially with some of the low gate drive ones that are available now. Certainly if you want to up the shunt current as you have. A bit late for you if you have built and installed 40 though!
Can't see any reason why shunting would affect the comms - if it was a low voltage thing I'd expect a "No Comms" message, although that is generated only if no serial port interrupt is seen for a second or so and it may be that you are still getting corrupted stuff that is enough to trigger the interrupt but can't be decoded into sensible signals.
arber333
Groupie
Posts: 56
Joined: Mon, 11 Mar 2013, 02:29
Real Name: Arber
Location: Slovenia
Contact:

Low cost BMS

Post by arber333 »

I tested individual banks for balancing and the cells were balancing fine for 1/2 hour or so. No trouble here. However i must have connected the comms wrong after final assembly and in the process 3 modules got zapped. I mean pic12F was blown off in smoke. But heres the kicker: every burnt module was the final module in its box. This is not coincidence, its the second time this happened. I will have to remove batteries anyway. Might as well resolder R5s.

However as i said before, individual box was quite able to sustain balancing with 12 or so shunts on. It must be something else....
bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

Low cost BMS

Post by bga »

Hi Arber333,

Some more thoughts:

Layout:
=======
Hot side of the opto-coupler is very close the cold side tracks.
Consider revising to maintain 5mm clearance.

Maybe more leadins off the bolt holes so the washer can't cut the track so easily.


Bypass shunt:
============
Use a SOT23 FET?, say NXP's PMV31XN (20V, 5A @ Vgs = 3V)
Should sort the bypass shunt. (and no gate resistor needed)

More than one battery size?

data comms connectors:
====================
Polarised connectors may be a good idea.
Making all the cell interconnect cables may be a chore!
Maybe consider an easy to crimp connector, say a 0.156 inch Molex type. The 0.l inch crimps are fiddly and easily botched when made.

I have been leaning towards modular plug-in types lately, such as Altronics 0.15 inch P2692 (miniature boxed PCB) and P2652 screw clamp terminal to suit the header.
This allows the comms cables to be made on the bench and installed safely into the battery pack.

Last edited by bga on Wed, 13 Mar 2013, 10:19, edited 1 time in total.
It's not the end of the world, but I can see it from here.
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Low cost BMS

Post by Tritium_James »

I'd avoid using screw terminals in a car - they tend to vibrate loose after a while.
arber333
Groupie
Posts: 56
Joined: Mon, 11 Mar 2013, 02:29
Real Name: Arber
Location: Slovenia
Contact:

Low cost BMS

Post by arber333 »

@ bga:

By close you mean R8 proximity to opto mounts? Actually i just copied that :).

The skinny track has auxiliary function as a fuse in case something would happen on the Pic supply line. Washer is actualy the M8 normal one, i use it with thin springy lockwasher. It covers most of the plating, i think it leaves some 2mm of tin uncovered. So there is no damage.
I dont like the fat lockwashers (the cut version). This also frees up 1,5mm of height clearance. This is important because i have only 20mm between terminals and front battery box casing.
Yes MX3 is one tightly packed car.

How do you mean bypass shunt FET and no resistor needed? Does the fet heat up for 3W?

I glue leads with hot glue. This should cover for vibrations. I dont mind cell to cell wires but between boxes wires are a problem since i have 2way connectors that look the same :(. I have to connect them using different connectors or something.
If someone has some idea which terminals to use, please...

A
arber333
Groupie
Posts: 56
Joined: Mon, 11 Mar 2013, 02:29
Real Name: Arber
Location: Slovenia
Contact:

Low cost BMS

Post by arber333 »

Ok

I solved my problem. Since some of you are working on your BMS i guess it is good to explain.
I changed comm lines to soft UTP computer cable. The thing is thin! But it works very good. I now have comms stable through all RPM and up to 200A (havent tried it faster :)). However when i start to take off from standstill and just after i stop the comms fall out for a second and then resume. Basically i dont mind this as long as the system restarts. However it shows that some powerful EMI is going on at start and stop.
When i read the ACe5 inverter documentation i found out it injects stoping (countering) current into motor just before 0RPM. Hm...

The other problem just went away by itself :). After i changed comm cables all three banks are balancing normaly. I tested for 1/2 hour, each module is stabilised at cca 800mA current, but transistor is still hot. I/O port disconnects charger at 3,6V as it should, so bottom balancing works.

I noticed something else though. If i have master connected to some cells and i add another bank to the comms in hot swap, the first modules start to balance without indication. I noticed this by smell of hot resistors. However if i restart (unplug 12V) master, everything shows normal, no balancing. A glitch in software maybe?

A
Last edited by arber333 on Fri, 05 Apr 2013, 13:12, edited 1 time in total.
wheresmyEV
Noobie
Posts: 25
Joined: Mon, 28 Mar 2011, 02:24
Real Name: kent Martin
Location: Brisbane

Low cost BMS

Post by wheresmyEV »

Hi, sorry if am repeating anything already said.
Just want to add a few general tips for PCB's. Add a polygon around the whole board top and bottom layer if it is 2 sided. Don't forget to set clearances for the max voltage you expect to see. Add vias wherever possible approx 5 -10mm apart. Add vias near component grounds (again assuming double sided). Tear drop large connectors, i.e. battery terminals. Large pad to tiny wire, consider tear dropping these to stop mechanical stress on the bolt breaking the trace. I haven't read all the comments or looked into the design details so please excuse any oversight on my part. Good luck.
Of all the things I've lost I miss my mind the most
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Low cost BMS

Post by Nevilleh »

arber333 wrote: Ok

I solved my problem. Since some of you are working on your BMS i guess it is good to explain.
I changed comm lines to soft UTP computer cable. The thing is thin! But it works very good. I now have comms stable through all RPM and up to 200A (havent tried it faster :)). However when i start to take off from standstill and just after i stop the comms fall out for a second and then resume. Basically i dont mind this as long as the system restarts. However it shows that some powerful EMI is going on at start and stop.
When i read the ACe5 inverter documentation i found out it injects stoping (countering) current into motor just before 0RPM. Hm...

The other problem just went away by itself :). After i changed comm cables all three banks are balancing normaly. I tested for 1/2 hour, each module is stabilised at cca 800mA current, but transistor is still hot. I/O port disconnects charger at 3,6V as it should, so bottom balancing works.

I noticed something else though. If i have master connected to some cells and i add another bank to the comms in hot swap, the first modules start to balance without indication. I noticed this by smell of hot resistors. However if i restart (unplug 12V) master, everything shows normal, no balancing. A glitch in software maybe?

A


Yesw, I remember telling you to use UTP right from the start!!

A glitch in the software? Surely such a thing is quite impossible Image

Haven't been able to reproduce that here, can you offer any more info?
arber333
Groupie
Posts: 56
Joined: Mon, 11 Mar 2013, 02:29
Real Name: Arber
Location: Slovenia
Contact:

Low cost BMS

Post by arber333 »

yes, you told me, i repent :).

I am using 1W dcdc galvanicaly isolated module 10-20V in and 12v out. All comms are utp but the short lines directly from master that i use to connect switches to.

When i have the comms up and running 39cells i then take the last wire (comm in) and disconnect from module. comms fall out. i then reconnect one more module and comms return. But this time some of the other cell shunts are active. I can smell. and also when i start balancing i get something like "ballancing 36" and i know that 4 cells are ballancing involuntary. Or the sytem doesnt see those modules. after i switch power off and on everything returns to normal.

i also noted that while driving over cca 200A some cells report 3,8v for few moments then back to 3,2.
Also under heavy acc and regen comms fall out and led is on for a sec or two. should i put master in some alu box?

A
arber333
Groupie
Posts: 56
Joined: Mon, 11 Mar 2013, 02:29
Real Name: Arber
Location: Slovenia
Contact:

Low cost BMS

Post by arber333 »

I figured my motor is throwing heavy EMI, but also the comm lines are going along the power cables.
I had to reduce EMI pickup so the BMS could process voltages correctly.
First thing i did was put 12V 1W isolated power supply on the input side. Didnt work. Image
Secondly i put the master circuit inside alu box. Still no joy!
Yesterday i took M10 nuts and comm cable and made some (5) turns round the nut. For more turns i would need more cable.
Huh, mixed results, but better i guess. When driving steady at 130A the pickup is minimal, but when accelerating heavily (500A) or regen disturbance remains. I will probably remove the cables and put them inside cabin. I hope mr. Faraday can help me limit EMI, since power cabled go on the outside.

Any other suggestion?
arber333
Groupie
Posts: 56
Joined: Mon, 11 Mar 2013, 02:29
Real Name: Arber
Location: Slovenia
Contact:

Low cost BMS

Post by arber333 »

Hi guys

I solved my problem of the jumpy comms. After msg with NevilleH about his input RC filter i tried as he did.

I got one 0,01uF(103) ceramic cap and soldered it across input resistor R1 (2,2K) pads with R1 in the middle. The first try was a dud. The disturbances were even higher. After brief msg with NevilleH he reccommended using scopemeter to identify the cap value where comm signal tops start to get rounded. That got me thinking; i was actually using different opto than he was. He used Vishay VO615A series but i got SFH618A. However the first couple cell boards i made according to original BOM. I went and cannibalised those boards and switched optos. Hurah! Success the first time. Comms are stable and glitches are gone. It looks like it wasnt the lines that were saturated, but master had its filtering set too wide.

I guess i will stop working on optical comms for now and focus on bluetooth serial connection and android app to show every cells voltages for the full pack. Finaly i can drive with confidence!

A

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Low cost BMS

Post by Nevilleh »

I am pleased that you solved your problem!
I guess if the parts list says to use a 6mm bolt, you better not substitute it with a 1/4" one.
I hope you used the correct optos on the cell modules.
Edit:
I had a look at the specs for both optos and your SFH one works with a much lower current - you didn't say what the CTR was though. It could probably be made to work by fiddling with the drive resistor (current) but better to use the tried-and-true VO one.
Last edited by Nevilleh on Wed, 05 Jun 2013, 06:09, edited 1 time in total.
arber333
Groupie
Posts: 56
Joined: Mon, 11 Mar 2013, 02:29
Real Name: Arber
Location: Slovenia
Contact:

Low cost BMS

Post by arber333 »

Yes!!! I am pleased too Image.

I expected my optos to work BECAUSE of the lower current demand. I didnt count on the interference though. The optos are SFH6186-4 model and have ctr to 300% so the ones on the cellPCBs should be good. You think the PCBs may have a problem with the input resistor? Actually i do have a +1,4V discrepancy measuring the whole pack with PCBs vs the multimeter. Hm.

I dont agree with analogy Image, i think using fine threaded bolt on normal thread nut would be better. You can still get it in, not the whole way though....

A
Last edited by arber333 on Wed, 05 Jun 2013, 20:29, edited 1 time in total.
Post Reply