Low-voltage shunt references

The margin at the low end may actually be better.

I'm surprised that it costs that much to build.

I've told people "give me 1% of what I will save you" and they won't do it.

I'm having reasonably good luck with that sort of approach just now, perhaps because it's easy to quantify a design-out.

Cheers

Phil Hobbs

Den onsdag den 6. december 2017 kl. 04.08.28 UTC+1 skrev John Larkin:

the numbers I've seen talked about is less than $300

I'd guess $150, $200 max. ...Jim Thompson

Well, then don't make the client try to think that way they don't like it. Get as much intel as you can (buying rounds of drinks usually helps) and then make your best stab at doing the math yourself, and then ask for a specific dollar value that's say 1.6% of what you will save them. Then let them "negotiate you down" to 1% but don't make it too easy.

Me: Just pay me 1% of what I save you.

Them: No, but how would you like to be a consultant?

Well, I'm still in business, and they're not.

The conventional circuit with emitter resistor and (current limited) base reference voltage exhibits this property for Vref = 1.0 + Vbe and Re = 1.0 ohm (actually slightly less because Rce(sat), but who's counting).

DN3540s have less resistance, and there are still others out there?

Tim

any company that refuses to let someone save them money deserves to go

NT

Alas, the Schottky metallization is laid on the Si surface, not buried safely underneath. Long-term reference drift would be an issue. So would buffer amplification.

A SiGe alloy has lower bandgap (1.12 down to 0.86V). It's not much, but it's something. Maybe a tunnel diode variant?

If you're protecting for shorts over peltier or either wire to ground, would it be enough to limit the supply that feeds the H-bridge ?

If the current limits were symmetrical or I had split high-current supplies , yes. However the spec limits often aren't even close to symmetrical--a cu rrent that would be quite safe in one direction (cooling) can destroy stuff if it switches direction. Since I also don't want to blow anything up in c ase of a short or wiring error, I really want a current limit on each corne r of the bridge.

Cheers

Phil Hobbs

and I am looking forward to the results....Hope they are positive, price wise

Cheers

Klaus

Yup. My other main LOB is proof-of-concept demos, and I don't worry as much about parts cost there. Assuming the demo succeeds (which nearly all of them do), I will definitely get asked what it would cost in production, so I don't go hog wild even there.

The less imagination you require of them, the more likely "money people" are to unbelt.

Cheers

Phil Hobbs

Also the whiz-bang-do-everything-for-a-premium-price parts selection, outside of the high-performance analog stuff, is very much geared to the disposable consumer phone/internet gizmo market; better think carefully about what you use if your customer expects you to be shipping them something with a lifespan of longer than 18 months

The LM339 scheme would be to make a bridge circuit, with the 339 sensing when the voltage drop in the emitter resistor exceeded the set point, and then stealing the base drive. It would oscillate like a SOB unless I compensated it like crazy, which might make it too slow. So maybe a couple of SS dual op amps such as MC33078DGKs (25 cents each) and four three-cent BAV99Ws to make clamps. That's more or less what I'm doing on the diode laser driver portion.

I'm trying fairly hard to avoid putting a switcher on this board for noise reasons, so I want to run the TEC off the 5-V supply.

One of the selling points of this gizmo is that the laser supply is shot-noise limited out to 100 mA and close to it up to 250 mA. (Yes, you have to drop a lot of voltage across the sense resistor do do that, and use a very quiet sense amp besides, but it helps the laser noise a lot and so is a good selling point in a very crowded niche.)

Cheers

Phil Hobbs

A current source looks like 1 ohm? ;)

Cheers

Phil Hobbs

Interesting--I hadn't heard of Schottkys aging like that. Do you have a reference for it, or are you going off general principles?

Cheers

Phil Hobbs