24V to 500-1000V, 20W floating DC-DC converter

It's still advisable for any npn running over it's Vce range at the temperature limit.

Germaniums were both leaky and more temperature-sensitive - typical BE resistor

Yes, i will use a 5 ohm emitter resistor to limit both base and collector current.

It will be much less than 11V for the TL431. I think it can handle 100mA.

Check your battery spec. A float voltage will seldom be 12V.

You're looking for a series cell ballancing circuit . . ?

RL

Maximum is 12.6V for lithium battery.

Yes. each 12V module (3x 4.2V) has it's own BMS ballancing circuit. I just need to limit external voltage to 12V. 4 modules for 48V. 8 modules for 96V.

sure you need the 12V limit? wouldn't it already be limited by the individual BMSs?

Yes, the BMS will cut-off over 12.6V, which would break the serial chain. I need to shunt it slightly below. This is only for charging.

On Wed, 09 Sep 2020 11:08:49 -0700, John Larkin

Because.

I should have said ANY bipolar transistor, used in a power circuit. Leakage current is process controlled and pretty stable; that is to say - it is generally untested.

Iceo is usually worse than Icbo, because it is generated by concurrent Icbo/beta effects. The emitter-base resistor avoids the effect on Icbo, by permitting a path for Icb that is not amplified - as I understand it.

As with all silicon device leakage, it gets worse non- linearly, with temperature. In doing so - it adds to self heating which will aggravate a supposedly static condition. Local hot-spots on a wafer can develop.

Most nondestructive Ice/Vce plots will show this effect as a 'negative resistance' slope at lower currents. This reduces between the stimulated conditions as Icbo becomes Icbx - from no bias to reverse bias - as will occur if some unaccounted for current is forced through an emitter-base resistor that is missing or inappropriately large.

If this were a thyristor structure . . .

Under transient conditions, the emitter-base resistor also aids in reducing miller charge effects, through the same mechanism, though saturation effects would dominate in a non-linear switching circuit.

As the OP is anticipating a battery protection circuit, leakage current may only be a concern if connections remain, on the shelf.

In a simple circuits like the OP's, where the TL431 can't be guaranteed 'off' above 1Vak, a path for it's stray current, to bypass the bipolar's base terminal, is only prudent.

RL

The emitter resistor is an ideal place to swot excess power in both the bipolar transistor and the TL431.

RL

TL431 wants about 1mA to function correctly, (*1)

So something less than 600 ohms, eg 470 or 560 to allow the transotor to switch off.

At half a watt dissipation in the TL431 (*2), max current at 12V is about 40mA, so you need a beta of over 120 in the transistor, which is getting expensive at 5A, so use a Sziklai pair. (Darlington would work too, but NPNs are generally preferable to PNPs)

Consider using a power resistor (2 ohms, 60W) on the collector side of the transistor (*3), that way you can use a smaller heatsink. (you only need to dissipate about 15W now (*4))

There's probably a way to use more resistors and transistors and dissipate even less heat in silicon

1: "RECOMMENDED OPERATING CONDITIONS" bottom of page 2 2: "MAXIMUM RATINGS" page 2 3: collector as drawn in the schematic, this would be the emitter lead of the large NPN in the sziklai pair. 4: peak heating of the transistor is at about 2.5A

--
  Jasen.

when the BMS goes open-circuit to protect the battery you still need a way to continue to charge the other batteries in the series string.

--
  Jasen.

Seeing as how you're using this circuit for balancing (which it doesn't actually do - it's just overcharge limiting), you might be interested in the 'non-digital', 'non-comparator' balance method used here -

RL

search of TL431 came up with this thread recently. So, i am high-jacking th is for a quick question:

e only

s transistor (TO-220 PNP)?

Yes, that's pretty close to the balancing part of my batteries. However, t he constantly on buffers/drivers drain out the batteries. I am in fact try ing to figure out how to window out the middle part. When the batteries ar e fairly balanced, i want to turn off the buffers/drivers.

A search of TL431 came up with this thread recently. So, i am high-jacking this for a quick question:

ide only

ass transistor (TO-220 PNP)?

the constantly on buffers/drivers drain out the batteries. I am in fact try ing to figure out how to window out the middle part. When the batteries are fairly balanced, i want to turn off the buffers/drivers.

The shunt regulators are in addition to the BMS balancers. Both circuits a re necessary.

. A search of TL431 came up with this thread recently. So, i am high-jackin g this for a quick question:

side only

7

pass transistor (TO-220 PNP)?

, the constantly on buffers/drivers drain out the batteries. I am in fact t rying to figure out how to window out the middle part. When the batteries a re fairly balanced, i want to turn off the buffers/drivers.

are necessary.

Also, the LT1782 only goes up to 18V. I am using LM358 on 24V, and up to 3

6V.

Run the simulation; balance is windowed naturally, so balanced consumption is a few uA.

Comparators and hysterisis are for nincompoops. Present company excluded.

RL

+------------------+------+ | | | BAT1 | 1M 12V | | | | | +--------+-[22]-balancer--+ | | | | BAT2 .--(---------' 1M 12V | `---------. | | | | | +-----+--+-[22]-balancer--+ | | | | BAT3 .--(---------' 1M 12V | `---------. | | | | | +-----+----[22]-balancer--+ | | | BAT4 | 1M 12V | | | | | +------------------+------+

each balancer (which is basically an op-amp driiving a complementary emitter follower) only sees about 24V but yeah 1mA to run an LM358 could hurt. but your TL431 circuit uses about the same amount of power.

a low power op-amp like TBS611 at 100uA is about 10 times better but at $1.33 about 10 times the price

--
  Jasen.

TL431 only needed during charging. Would be nice to use balancer during discharging.

Can't find that part. Who makes it?

Op amp is your choice.

LT6000 LT6001 LT1672 LT1494 AD8657 AD8659 LT6023

These have Iq of as low as 1.6uA, and run on voltages as high as 36V.

The balance circuit, itself, is never meant to see

2 x cell voltages.

Run the simulation.

RL

:

oup. A search of TL431 came up with this thread recently. So, i am high-jac king this for a quick question:

ary side only

ge 7

the pass transistor (TO-220 PNP)?

ver, the constantly on buffers/drivers drain out the batteries. I am in fac t trying to figure out how to window out the middle part. When the batterie s are fairly balanced, i want to turn off the buffers/drivers.

ts are necessary.

36V.

OK, thanks. The LT17s and LT60s sound good, but a bit pricey. Is the circ uit your own design, or pull from somewhere? Just wondering if there is pa ttern or copyright issue for production.