Commanda's BMS

commanda

100 W
Joined
May 7, 2007
Messages
108
Location
Sydney Australia
This is my BMS. It started out based on the original McRee/Goodrum circuit.

The criteria I wanted over and above the original design was:
Use any voltage source, not a specific SLA charger.
Unit was to shut down when charging was complete.

To this end, I used 2 constant current sources. These have a negative temperature co-efficient, so the current decreases as the heat rises.

Initially, in bulk charge mode, both current sources are on. Any Shunt Low turns off the higher current source. Charging continues on the second, lower current source.

Once all cells are full, All Shunts Low triggers a time delay before switching the R/S flip flop, which shuts down both current sources. And lights the Charge Complete led.

Bulk charge starts at 4 amps, and settles down to 2.7 amps as it warms up. Trickle charge is about 400mA.

The charger comprises 3 x 24 volt, 6.5 amp Meanwell power supplies. These have a voltage adjustment pot, and the DC supply voltage is about 62 volts. The charger has a diode in series with the output, so the batteries don't power up the supplies, and also a 12 volt regulator to power the BMS logic.
The charger is housed in a recycled server computer power supply case. The BMS also has a 12 volt input from the scooter, to power the LVC to ebrake circuit. This has not been tested yet.

The Mark One version looks a bit prototypish. Mark Two will have a better case, better pcb layout, and I'm working on a switchmode current source, which should allow more current and less heat. Works great in LTSpice, but not built yet.

First photo is the battery pack, 16 x TS LFP 40 AHA.

Second photo is the underside of one pcb, containing the per-cell circuitry for 5 cells. Mostly SMD.

Third photo is the BMS enclosure. 2 x 10 led dip bar graph for the per-cell indication, and a hard disk drive cooling fan for, well, cooling.

Fourth photo is the whole lot on my workbench.

Fifth photo is inside the BMS.

Sixth photo is a closeup of the current sources, showing the little TO92's superglued to the fets (thermal feedback).

And last, but not least, the circuit diagram in pdf format.

Now to install it in my EVT scooter.

Amanda.

_DSCN0812.JPG
_DSCN0821.JPG
View attachment 4
View attachment _DSCN0823.JPG
View attachment _DSCN0824.JPG
_DSCN0826.JPG
View attachment BMS-circuit.pdf
 
Thank you for sharing!

Would you please tell me the values of R1 and R2 on the second page of the pdf? These are the resistors connected to the bases of Q2 and Q4. I can't quite make out their values; the pdf is a bit blurry.

I like how you implemented automatic charge shut-off. This, I believe, is very necessary. It is all too easy to forget you have a lithium battery charging, and leave it on too long, increasing the risk of a dangerous fire.
 
R1 = 0R15 (nought point one five ohms)
R2 = 1R2 (one point two ohms)

These set the current for each of the current sources. When enough current flows to turn on the NPN, it starts to pull the gate of the FET low, tending to turn the current off. The Vbe of the NPN decreases with increasing temperature. So the current decreases with temperature.

Amanda.
 
Excellent. Thank you for posting.

I'm eyeballing TS for a small car (72V 150AH)... wasn't sure how to address charging, but this looks promising.
 
Just a quick update. I added a small circuit into the power supply box, which detects current flow to the batteries. Once the batteries are full, and the BMS shuts off the current, the power supply now switches off as well.

The circuit uses a 0.1 ohm, 5 watt resistor for current sensing. An LMC7111 cmos rail-rail op-amp, a fet, and a relay. Relay contacts (NO) switch mains power to the supplies. Push-button switch across the relay contacts to start charging. Now I can plug it in when I get home, and not have to worry about going outside later to switch off the power supply.

Didn't take any photos,sorry. I can scan the circuit and post it if anybody is really interested.

Lessons learnt:
0.4VA sub-miniature push-button switches will not switch 3 x 150 watt power supplies.
Allow for the overhead current of 16 leds, 16 opto's, and 2 cooling fans when choosing the set-point.
Make sure the power supply cooling fan current doesn't go through the sense resistor.
Make sure you use the Normally Open relay contacts, not the Normally Closed.

Amanda
 
In the Australian vernacular................................Bugga.

Let the magic smoke escape.

Got home last night, plugged the charger in, about an hour later couldn't hear the cooling fans on the power supply. Investigated, power supply had shut down, like it does at end of charge. It normally takes closer to 4 hours, so something was wrong. Lifted the seat, found the BMS was way warmer than normal. Turned the power supply back on, 14 of the shunt indicator leds came on, but no fan. There was also a hint of burnt silicon smell.

Cracked open the BMS, to reveal the photo below. It appears the cascade failure was initiated by failure of the fan. One lead had come adrift. It had been tacked to the leg of one of the power resistors for the current source. I would guess that heat had fatigued the solder joint, then vibration did the rest. Also found one of the little TO92 transistors glued to the fet in the current source had also come adrift. This would compromise the negative thermal feedback controlling the charge current. I've also been a little slack the last few days since implementing the auto shutoff of the power supply, in that I haven't been lifting the seat to eyeball the BMS whilst charging. If I had, I may have caught the non-functioning fan before the rest of the damage was done.

Anyway, the Mk2 version is already under construction, so the long-weekend this weekend has come at just the right time.

View attachment DSCN0832a.JPG
Amanda
 
Hi Amanda,

Nice work, thanks for sharing, and sorry about the smoke.

You said you were using Meanwell PSUs - I think they are pretty much the same as the ones I have.

Here http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=6749 is a trick for controlling the current from them. With a scheme like that you don't need your FETs Q1 and Q3. You could eliminate the heat dissipation from them and probably not need a fan.

HTH

Nick
 
Nick,

Saw that post. What I'm trying to achieve is a BMS that just needs almost any dc input voltage. Ultimately I'll probably play with a Sepic buck/boost converter. But the development plan is linear current source, switchmode current source, then develop from there. What I had only failed due to physical implementation. Mk2 will be better.

With your add-on circuit, the under-voltage cutout may catch you out. The Meanwell supplies I have, I adjusted the voltage trim circuit on one to try and bring the voltage down, but it went into a hiccup protection mode because of low output voltage. Didn't pursue the matter any further. Of course, YMMV.

Amanda
 
An update after 21 months. Can't believe it's been so long.
My EVT 4000 gets ridden to work 5 days a week, 10Km round trip using about 10AHr out of my 40 AHr pack.
My rebuilt BMS has been working just fine. Last week I noticed one of my 2 cooling fans has deceased, but since it's the middle of winter right now I'm not too worried.

This is the dual linear current source, constant voltage power supply charger, with low current detect auto shutoff. Essentially the circuit I posted in the first post of this thread way back when.

Amanda
 
Back
Top