# Zener diode reverse characteristic curves.

Discussion in 'Electronic Design' started by James Meyer, Jul 21, 2003.

1. ### James MeyerGuest

I was told that there are no knees associated with zener diodes if the
characteristic curves were plotted with the "correct" scales on the V-I axis.
Take a look at figure 5 on page 3 for the CZRA series diodes whose datasheets
can be found at http://www.comchiptech.com/spec/CZRA1110_1350.pdf and tell me
what sorts of axis ate needed to reduce the curve to a straight line.

Jim

2. ### Jim ThompsonGuest

A French curve ? Sent by "couriel" ?

...Jim Thompson

3. ### George R. GonzalezGuest

There are a lot of problems with this datasheet. One sees this a lot
these days. Why when I was a boy, a datasheet was a datasheet.
I still remember the one for then 2N107........ Oh, what? Where am I?

First of all, these are in no way ZENER diodes. The actual zener
phenomenon only happens below 5 volts or so. Anything over that is

Secondly, these don't seem to be specified as (non-zener) REGULATOR diodes.
They seem to conduct somewhere around 120% of rating, so they're
more likely to be snubber diodes, which you might expect to have much
looser and variable curves than a true REGULATOR diode that you plan
to use to generate a REGULATED voltage.

Now back to your question, If I Recall Correctly, a diode's forward
conduction
curve is an exponential one, so if you plot the forward voltage linearly on
the X axis
and plot the LOG of conducted current on the Y axis, you get pretty
straight line.
You can see a bunch of these in one place in the back of Bob Pease's
excellent
book, "Troubleshooting Analog circuits". This exponential curve goes all
the way down,
way down down down below what we usually think of the lower limit of
conduction,
arounf 0.6 volts for silicon diodes.

confusingly
called "forward" or "normal" conduction for Zener (and Zener-workalike)
diodes.
It's probably not far off an exponential curve, with perhaps the x intercept
shifted by
the Zener voltage.

Maybe you could look at some Zener SPICE models and see what they use?

4. ### James MeyerGuest

Whenever I see a post that uses any of the words "no, never, all, and
always", I'm almost 99 percent sure that an exception exists somewhere.

That's why I usually try to include a few "weasle-words" in my posts.
Most of the time.

Jim

5. ### James MeyerGuest

Most of the Google hits I got had to do with SPICE models. I was wary
of quoting them because I had the feeling that they were developed to show the
designer what he felt he *wanted* to see rather than what actually existed.

My response was to a previous message that stated categorically that
there were *NO* knees associated with diodes, not even with the reverse
characteristics of "zener" diodes.

That's clearly a specious statement.

Jim

6. ### Winfield HillGuest

George wrote...
Actually, from measurements and studies I have made, the
common "avalanche" zener diodes (all those above say 8V)
are generally VERY far off from any type of exponential
characteristic.

For example at low currents where the zener knee is first
encountered (yes there is a definite knee for most types)
the zener voltage may actually drop as the zener current
is increased. Furthermore, after the conduction current
is well established, the voltage drop at higher currents
can look more like a fixed unvarying zener breakdown in
series with a fixed internal series resistance. This is
clearly seen at least in pulsed high-current testing.

And to further complicate matters, this "fixed" breakdown
voltage has a positive temperature coefficient (which can
be quite large at high zener voltages), so along with the
series resistance this changes the measured voltage as the
current and the dissipated power is increaseed, and adds
time course and package mounting issues to the situation.

Whew, all this talk of breakdown makes me hungry, I'm off
to grab some lunch.

Thanks,
- Win

7. ### Jon ElsonGuest

Figure 5 is the voltage drift with respect to temperature.

Figure 4 shows the current vs. reverse voltage, and the knee on this
curve is quite obvious. Do you mean you need a linearized temperature
coefficient, or linearized voltage/current? Although you can't get it in
a 100+ V range, there are some great reference 'diodes', like the LM399,
LM4040, etc. that have much improved characteristics.

Jon

8. ### John LarkinGuest

Wake up, George. It's 2003. Nobody proof-reads data sheets any more.

John

9. ### Jim MeyerGuest

Here's why I can't take SPICE models and simulations seriously....

In the listing below, change the two lines as indicated, run the
simulation again, and then explain why there is a *negative* 390 mV
potential across D1 at low current levels.

to
SYMATTR Value 1N750
to
TEXT -320 80 Left 0 !.dc dec I1 .01e-9 200e-3

10. ### Jim ThompsonGuest

I don't have time to load your listing right now... I'm scurrying to
finish a job before my chassis fan dies... the monitor pops up every
thirty minutes or so when the speed drops seriously below spec :-(

But my guess... Mikey uses behavioral diode models which are a wee bit
hokey... they're a voltage source/ideal diode gimmick.

...Jim Thompson

11. ### Jim ThompsonGuest

I think you'll find that the "true-Spice" models are pretty reliable
if you choose IS, RS, BV and IBV judiciously.

Those models which use a DC source plus an ideal diode can do weird
things.

...Jim Thompson

The proper terminology about something which there may be more than
significant doubt is "take it with a grain silo of salt." At least
that's they way I like to say it.

13. ### Robert StrandGuest

My tests have shown fairly close to exponential behaviour for zeners with
nominal voltages say below 4.7V, above that voltage they are like you describe.

Rob

14. ### Robert StrandGuest

This was a reply to Win's post. It doesn't look that way on my browser!

15. ### Winfield HillGuest

Robert Strand wrote...
If you search the s.e.d. archives on Google using the words zener
oscillation, you should uncover the half-dozen threads exploring
this subject. Several of us spent far too much time on the issue,
but the interesting physics we found makes for interesting reading.
Some of the posts feature my attempts at ascii oscilloscope traces.

Thanks,
- Win

16. ### Robert StrandGuest

If you search the s.e.d. archives on Google using the words zener
Win, thanks for the tip. I've have read some of those over the years,
probably not all of them though.

Regards
Rob