Avalanche diodes?

Sep 07, 2011 32 Replies

Ah. After a little clarification, it seems that what you're actually looking for is a Transzorb:

formatting link

I once worked at a place where they had MOVs all over the place - MOVs have a soft breakdown curve, and they deteriorate a little bit with each transient they suppress. When the company muckety-mucks discovered Transzorbs, they issued ECOs to throw away all the MOVs and replace them with Transzorbs.

It worked.

Good Luck! Rich

"hrh1818"

What else is different and makes one device an avalanche diode and the other a controlled avalanche rectifier?

** Forget badly worded definitions - look at the specs.

There is a huge difference in the energy that can be absorbed during reverse conduction.

.... Phil

If a diode has a tightly specified reverse breakdown voltage, like 5 or 10 per cent, and is intended to be used as such, we call it a zener diode, even though the mechanism may technically be avalanche.

A "controlled avalanche rectifier" has a specified minimum reverse breakdown voltage but usually no max. It's a rectifier that's designed to dissipate some amount of reverse-bias power without being damaged.

The rectifier diode is, well, intended to be used as a rectifier. The zener is not. [1]

John

[1] I screwed up and used an LM1117 to regulate 3.3 volts down to 2.5 for an FPGA Vcc_aux supply. My bad. Turns out that a 7.5 volt MELF zener, use in the forward direction, drops almost exactly 0.8 volts at our operating current, and solders beautifully into the 1117 footprint. So that's what we do now.

ftp://jjlarkin.lmi.net/Diode.jpg

On a sunny day (Thu, 08 Sep 2011 22:22:56 -0700) it happened John Larkin wrote in :

It will only work if the current is sort of constant.

Sure. All regulators need their loads to be sort of constant.

The Vcc_aux doesn't use a lot of current and the voltage limits are pretty wide.

John

On a sunny day (Fri, 09 Sep 2011 06:34:56 -0700) it happened John Larkin wrote in :

? Not the one I am working on.

Yes OK. But I did consider that solution in the past and rejected it. Why save on a 1$ compnent in a 1000 $ piece of equipment?

Greed .. I discovered some really funny market working todsy, but am under NDA.

Greed? Yes! We're saving 35 cents in a $5000 piece of equipment. Plus, the 1117 needs two resistors to program it and a tantalum output cap for stability. All that junk wastes PCB surface area and costs more to kit and place and inspect.

I really like the MELF backwards zener thing. We do it on purpose now.

Incidentally, as the forward drop of a diode goes up, its tempco goes down. Somewhere around a volt roughly, it's zero. Happens at a lower voltage for schottkies, of course.

John

Wouldn't it be safer to place a TL431 on the output to insure it does not move above that 2.5V ? I think the little extra drift can be suppressed with the 431 with out any issues.

Jamie

If I was going to add a bunch of new parts, I could just use a real LDO and a couple of programming resistors.

The FPGA isn't very picky about the Vcc_aux voltage, so the funny zener trick is safe. And it's sure not going to oscillate.

John

Makes sense. The silicon resistance tempco balances out the diode drop tempco.

Cheers

Phil Hobbs

Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net

...or a fixed LDO, and forget the programming resistors. ;-)

Always a tradeoff but there is usually plenty of margin in the LPUL (as long as you're not trying to drive 3.3V logic or something equally dumb).

On a sunny day (Fri, 09 Sep 2011 08:11:01 -0700) it happened John Larkin wrote in :

It is interesting, I had to do some soul searching this week, trying to design something for production in larger quantities. The desire to cut on components was sooooo strong. I finally decided to throw in some more chips to make it 100% what a customer would expect (and then buy more), then do the 'close enough' one which would be usable, but not that perfect. So if you look at the diagram you see a lot of things, and may wonder - could that not be done with a micro, and there is a micro on board too. But it could be done with the micro, but not perfectly - clocks etc.. Better would be a FPGA and less glue logic, but that would probably price me out of the market..

you

ode

lled

e
&

Two similar posts in a week. It is september

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required