Zener Q

IIRC traditionally that is 6.2V or 6.8V. Check the datasheet to make sure.

from what I've read zero is supposed to be around 5V

the slightly higher voltages might be a zener and diode in series so the positive and negative coefficient balance out

-Lasse

ON the very first day of my very first engineering job, my mentor took me aside and informed me that I could trust what it said on the data sheet some of the time. Selected parts were a no-no.

I once had to clean up after an engineer who'd used a selected part. It was part numbered. There were a kagillion of 'em in stock. And it was dirt cheap compared to the alternatives.

Nobody bothered to check and learn that the project that had specified the part had bought a lifetime supply and that the company was out of business. We learned the hard way when both production lines shut down.

I think you're mixing standard versus compensated.

Thinking back to my days at Dickson Electronics, I made hybrid circuits, the other half of the building made TC'd zeners... if I'm remembering correctly 5.6V + a diode drop = 6.2V, physically a chip-stack the way they made them, had virtually zero TC.

But I could be remembering wrong... that was 40 years ago :-) ...Jim Thompson

I really like LM329s, but they don't come in SMT and there's some worry that they may go away. I usually put in a population option, so that the 329 can be replaced with something like an LT1021-7. Those are a lot more money, and no better for most of my uses, but if the 329 goes away, no board spin is required.

Cheers

Phil Hobbs

Just also remembering that, on a single bipolar process, that 5.6V + TWO Vbe's, made a nearly flat 7V.... that's what is in most of my alternator regulator patents. ...Jim Thompson

Depends on the operating current. 5.6 may be better.

Use a bandgap!

--

John Larkin Highland Technology, Inc

jlarkin at highlandtechnology dot com

Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation

Why not use an LM4040/4041 or some such?

Motorola made a lot of money selling the the 1N823 - 1N829 series

which is a 5.6V zener in series with a regular diode.

If they could have done as well with a 5.1V zener on its own, I don't think that they would have bothered inventing the 1N823 series.

-- Bill Sloman, Nijmegen

MANY years ago I did this exercise with through-hole zeners and came up with a 1N751A (5.1V nominal) at some particular operating current (can't remember now, but NOT particularly low).

You'd be better off using the cheapest regulator you can find- most of them use bandgap references internally, even if they don't advertise it. Or a shunt reference like a TL431.

Best regards, Spehro Pefhany

Buried zeners generally have much better noise and long term stability than bandgaps, AFAIK. So that could be it.

There's also those floating gate types, if that concept doesn't keep you awake at night.

Best regards, Spehro Pefhany

(a) Bandgaps are a good 20 dB noisier than buried zeners. The LM4040's

(b) The 4040's initial accuracy is good, but its tempco is poor. TI doesn't make the good grade of LM329 anymore, but they're generally much better than the quoted typical 50 ppm/K. LTC still makes the good ones for occasions when that matters (6 ppm typical).

(c) I prefer through-hole for voltage references because it reduces die stress, which can make the voltage walk all over the place. Routing a C shape in the board around the reference works OK too, or so I'm told.

Cheers

Phil Hobbs

See e.g. AN82 page 7

to

Yeah, like that. Getting good low-frequency stability from a reference can be a challenge. I usually don't care that much about absolute accuracy, but some folks use my gizmos to make measurements lasting many hours, so stability is really important.

Cheers

Phil Hobbs

We like ADR420-series (jfet things) parts for serious references. The B grade is 1 ppm/degC typ. But we are more concerned with TC and stability than noise, although their noise is pretty good too.

to

Those are pretty nice parts, but I bet they don't cost $0.30 like an LM329! ;)

Cheers

Phil Hobbs

ry

at

I'm going with the LTC6655-5 in my low distortion oscillator. I was thinking of the Analog Devices jFET things, but one of my Dutch contacts had had trouble with bad 1/f noise, and recommended the LTC6655 as a whole lot better.

-- Bill Sloman, Nijmegen