Home Brew ballancer (4 AMP)

[nebriancent]

i have been working on getting a 4 amp ballancer running stabaly over the last couple of months in my spare time and wanted to know what you guys think about it
***EDIT (you should be able to charge at well over 4 amps but that should be the most it will actively balance at)
***EDIT no cut-out for overcharged cells (shrugs) should be fine at 4.1 anyway as it gives you some headway to let other cells catch up
***EDIT can be easily adjusted to charge lifpo4 just swap out the voltage divider feeding the tl431 ref pin with values that will turn on at required voltage (tl431 wants to see 2.5ish volts to turn on)

its based on the tl431 adjustable shunt regulator

schematic

ballancer.png

(86.6 KiB) Downloaded 2 times

prototype 1

prototype 1.jpg

(249.96 KiB) Downloaded 2 times

put the 3906 transistors in backwards (still worked for the led for some strange reason c-e swaped?)
proto 1 was only good for 1 amp charging w/o overheating the un-heatsinked power transistors (mounted under board)

proto 2 (working right)

prototype 2.jpg (230.85 KiB) Viewed 1317 times

powered up shunting 4 amps of current

shunting 4 amps.jpg

(105.27 KiB) Downloaded 2 times

and yess that is a cut down pentium 1 cpu heat-sink held in place with a hair elastic on my aluminium angle heat bar (its temporary)


BOM per series + mouser part numbers

cant remember the part number off hand for the insulators for the power transistors

IC

TIP147T
1: $0.775
25: $0.673
100: $0.59


2N3906TAR
1: $0.076
25: $0.066
100: $0.045


TL431BCLP
1: $0.666
25: $0.524
100: $0.46



(CAPASITORS)

REA101M1ABK-0511P (10v 100uf)
1: $0.066
10: $0.06
100: $0.058


C320C104Z5U5TA (0.1uf ceramic)
1: $0.131
100: $0.087


B32529C103J (0.01uf polly)
1: $0.12
50: $0.112
100: $0.106

(RESISTORS)

280-CR25-1.0-RC (1 OHM resistor 25w)
1: $1.19
10: $1.08
100: $0.895

291-120-RC (120 OHM 1/4w)
1: $0.109
10: $0.059
200: $0.021


ERD-S2TJ331V (330 OHM 1/4w)
1: $0.098
25: $0.038
100: $0.029


PV36Y202C01B00 (2ktrim)
1: $1.49
50: $1.01
100: $0.876


291-1K-RC (1k 1/4w)
1: $0.109
10: $0.059
200: $0.021


291-5.6K-RC (5.6k 1/4w)
1: $0.109
10: $0.059
200: $0.021


291-8.2K-RC (8.2k 1/4w)
1: $0.109
10: $0.059
200: $0.021

(OTHERS)
m3 screws and tap
aluminum angle
old cpu heatsink
insolating heat conductive to220 pads
insolating washers for mounting the transistors to the angle
red led 3-5mm
proto board

[Samson]

This is a nice simple circuit but could be even simpler.
The 2N3906 is not really needed. The Output of the TL431 can provide enough current (100 + mA) to turn on the TIP127, though I might include a small value resistor between the base of the TIP127 and the output of the TL431 to limit base current. For continuous operation you may need a heat sink on the TIP127 as it is rated at 2W at 25 deg C. but is running at over 3+ watts when full on. Also C1 and C2 are not needed.

[nebriancent]

please explane why you think the caps are not needed...
I used them to try to smooth out the ripple as the tl431 switches on and off as I was worried it may be hard on charging equipment and batteries .... of course not knowing 100% if it would even become a problem

the 3906 is not to switch the other transtsitor but instead to drive the led as a ballancing indicator

so in fact both pnp transistors are being switched by the tl431



the darlington pair transistors are clamped to a piece of aluminium angle to act as a way to grander heat to the case (once built)

I kept the 3906 in as a way to later switch a optocupler to drive a fan circuit if cooling became a issue

plus it is nice to see what batteries hit terminal voltage first

things still left to work on would be
1 a aluminium case (to help shed heat)
2 a fan circuit using optocuplers to drive the base of a transistor feeding a lm7812 so I could pull off of the hole pack evenly wild activley ballancing .... perhapse I would need a freewheel diode on the fan (if cooling is a problem after making said heatshedding case

[Samson]

The cells act like a capacitor in effect. If filtering was needed a small value cap from the pot wiper to the negative buss would be more effective at reducing signal noise\ripple. I was under the impression that the 3906 was the driver for the op transistor but I see that the TIP127 is a darlington so disregard my comment re not needed (though a darlington is not needed, it cannot hurt).

[nebriancent]

I didn't even think about the battery acting as a cap...(palm face)
if I used a regular pnp power transistor instead of a darlington I was worried about pulling too hard on the tl431 as it is rated at 100ma current... but yes now that I think about it wile connected to a battery the electolitic caps are just extra parts... thankyou for pointing that out to me it drops Tue cost per serries connection

the reason I built it in the first place was I was tired of having a max balance current of 500ma like on some of the comercial units ... I don't need a current limit on a bms now I onley need a cell level Low voltage indication and I have found I way to do this using the tl431 aswell (loves this chip... so versitile)

[fechter]

Yes, you probably don't need caps on the front end at all.
C3 could possibly cause oscillation in the output.
I'd recommend using a 47uF from base to collector on U2 to kill any possibility of oscillation. That would really be the only cap needed.

[nebriancent]

just so I undrstand you right you suggest a 47uf cap in parrilell with the resistor that holds the base of the pnp transistors high
and that would kill any posible oscilation on the output.. would I the not have to worry about overdriving the tl431 wild pulling down the base and charging said cap... if so what value resistor do you think would be needed to avoid said overdriving (in sierries with thw hast of the darlington)

[fechter]

No, cap should go across base to collector on the big darlington. There won't be any problem overdriving the 431. The capacitor could be about anything over 10uF. I found that 47uF ones were the same size/price as 10uF so went with those. With this cap in place, you shouldn't need caps anywhere else in the circuit.

An interesting thing I discovered about the 431 is the output becomes 'noisy' around the trigger point. The noise gets amplified by the gain of U2, which could be a lot. If you have an oscilloscope, you can see this. The cap on U2 acts like an integrator and slows the output swing. This will totally take out the noise.

I found that without the proper filtering, if you tried powering several cell circuits in series without the cells connected, it would go into a full swing oscillation.

[nebriancent]

wow I hadn't realised that the noise would be that bas from the 431

I thought THE FILTER cap I had on the output would help kill the noise

I havnt seen any significant noise through the spice app I use.... but that dosnt mean its not there...

I realty need to buy a scope:(

[methods]

I realty need to buy a scope:(

Don't talk about it - just do it.
You can save money by going with a 2ch - 4ch is nice but rarely needed.
You can also save quite a bit by giving up some resolution... it does not have to be 400Mhz+ - even 100Mhz is great - or 20Mhz to start with.

Remember to budget for a probe that is rated for at least as fast as the scope - most people underestimate the importance of the probe. They can cost you $100+ if you buy them so start looking out now. Dont pass by the crusty looking one on the bottom of the drawer... there are some really nice *really old* probes around that kick the crap out of an ebay probe.

There are some things an engineer just has to have
If you want to be successful you have to have a scope.
Do it bro - I promise you will look back and wonder how you ever designed with out one.

Spice can never simulate the parasitic realities of your prototype. I dont even simulate anymore since all it ever does is disappoint me. Easier and faster to just build it an measure... and some times I find solutions by mistake in just doing it.

-methods

[nebriancent]

i have since abandond this project as i found it not a good bang for buck idea
i have learned alot from building this ballancer and it is in my buddies house in daily use bringing cells within 5mv of ballance (i could have fine tuned a bit better but w/e)
i recently have lost my main job, started building a headway pack(juust befor i lost my job) and can not longer afford to fund this project

but thats not to say that i wont pick it up in the future as it is a easy to modify base for a full size car ballancer that i found verry stable with wide flucuations in heat

it is also easy to modify target voltage cutout by changin 2 resistors
and i could also up the current handaling by building a small mosfet driver and replacing the pnp darlington transistor with a pchannal mosfet

but i have to for now dump the project untill my funds come back in order

if anyone wants to continue this project on their own feel free so long as you give credit on the base to myself and the forum

cheers for now

[Jeremy Harris]

I built a near-identical circuit a couple of years ago, but I limited the shunt current to about an amp to minimise the heat dissipation. I used much higher resistor values in the voltage set potential divider, and a TIP125, rather than a TIP127, plus I didn't bother with the capacitor. The circuit's been working fine for around 2 years or so now on my electric boat.

TBH, I can't claim credit for the idea, as I first saw this circuit here on ES around 4 years or so ago, as a joint development between fechter, bobmcree and ggoodrum. The circuit details are almost certainly buried in the archives here somewhere.

[nebriancent]

i was unaware that somone had a near identical circuit running already

i fount heat stooped beeing a issue after the transistors were heatsinked and the resistor leads were beefed up

the resistor seems to pull heat off of the transistors as it is rated for 25w and is only disapating about 16w across the whole curcuit