Low-voltage shunt references

I make references all the time that are = GOZOUTA ...Jim Thompson

On Tuesday, December 5, 2017 at 11:34:51 AM UTC+11, snipped-for-privacy@gmail.com wrote :

at really need _independent_ four-quadrant current limits. The simple way t o do that is to use an asymmetric complementary H-bridge--one side is a bip olar current source (collectors common) and the other is a voltage source w ith a gain of -1 (emitters common).

imit. I'm using the classical two-BJT, two-terminal current limiter, except with shunt references doing the limit function. Conveniently, TLV431s wor k like NPNs and LM4041s work like PNPs, so it's just one IC, one transistor , and two resistors per corner.

ncreases the power dissipated in normal operation.

Have you looked at the LM10?

Element 14 have got 2500 of them in stock (their number 1468909 for the SOI C part LM10CWM/NOPB) but they aren't all that cheap, at a couple of dollars each.

One of Bob Widlar's odder parts, but very handy in some applications.

I was thinking of makine a Peltier driver with a couple of LTM8052A modules. It won't be as quiet as yours but might be ok for most things. It also won't be the cheapest way to do it, but might be simple and small.

By the way, read and follow the moisture sensitivity warnings for the LT modules. They do mean it. I was caught out because I was fitting a non-LT part to a blank footprint on an assembled PCB that had been sitting around for a while, but the PCB already had LT modules on it when I put it through reflow again to add the other part. I then had to replace several failed LTM parts. There were no more problems if the boards/parts were baked before reflow.

Linear Tech offers a similar part - LT1635.

Not cheaper, or smaller.

RL

Several people make this part,

which doesn't quite do what you want, but it's interesting.

It needs a 2.2V supply though. Unfortunately there seems not to be low-cost shunt reference below 1V. Phil, if you don't want to use a boutique part you might want to see if the client can donate a "free" uC pin to your cause. I know, they always razz us about such requests "And next thing you know, he also wants one of the timers!". Then send a xxx kHz signal to that pin, add a Cockroft-Walton multiplier to it and that makes the supply voltage. If they won't give you a pin then a single logic Schmitt inverter would do. With supply voltages this low you'll need Schottky diode in it but those are cheap and come in tiny multi-packs. Not pretty but it works.

Den mandag den 4. december 2017 kl. 22.42.02 UTC+1 skrev Phil Hobbs:

first hit on google,

Note the 0.9V minimum. That's equivalent to a cold vBE, the _absolute_ minimum that can be accomplished until someone comes up with another, smaller, "bandgap" structure. ...Jim Thompson

Germanium, Lead-sulfide :^) (ducks)

George H.

I won't hold my breath. ...Jim Thompson

I believe that had been obsoleted.

I'll actually be doing the digital side of this one as well, using my usual standard LPC17xx Cortex M3, so I could do that.

I still have to do the thermal budget for this gizmo, but I may well wind up using an LM339 and a few transistors to do all four corners, assuming that I have a higher VCC available for the 339. I really should be able to do this using ~100 mV of sense resistor drop, but I'd really rather not redesign the gizmo that extensively. (I obviously don't sell the same design to more than one customer, but do mix-and-match design features when making new ones.)

Cheers

Phil Hobbs

I saw that one, thanks. Obsolete, unfortunately.

Cheers

Phil Hobbs

FWIW, Rochester Electronics has 716,451 of the NCP100ALPRPG in stock at

A Cockroft-Walton off of a port pin could also supply a LM339 and if thermally critical use a low-power version.

Still not a big attraction for new designs, especially since it only saves me a quarter of the lost headroom.

When I get a bit further down the road with the customer I'll revisit this--a 339, two PNPs, and a dozen or so resistors isn't horrible for making four independent current limiters. Besides, it would be cheaper, probably under 25 cents for all four. Maybe I'll use a pair of LM393s instead, for a bit of redundancy at a dime extra.

Cheers

Phil Hobbs

Or use shunt resistors and diffamps to pick off the current, and compute the current limits.

There are lots of current sensor chips around nowadays.