zener diode drift????

[Brian Marshall]

I've been working on a project that requires a constant reference voltage.

I have been using a similar set up to an MXR phase 90.  10K resistor from V+ in series with a reverse biased zener and large cap to ground.  The zener I'm using is a 4.7V 1n5230b, not that i think that would matter in preticular compared to any similar device.

While on breadboard, I was using the voltage divider directly as a reference for the circuit.  After a few minutes things seemed to slip out of bias so i adjusted the trim pots by ear, and it worked again.  I thought, well that was wiered, and checked the voltage divider, and got a reading of 3.5 volts.  I tried a new diode, and it instantly slipped to 3.5 volts.  I thought maybe i was loading it down, so i buffered it with an opamp. The voltage held around 4.7 volts for a few minutes, and then fell to 3.5v again.  I double checked the resistor, thinking maybe it was not resisting enough, and overheating the diode.  Nope.  The resistor read 9.9K on the ohm meter.

anyone have any insight here in to what might be going on.  Given that I am buffering the voltage, I know that it is not being loaded by anything other than a TL072's input impedance which is extreemly high.

[George Giblet]

Zener voltage varies with zener current.   Your current is slow which makes the voltage low.

Zener voltage does vary with temperature but that's not your problem.  If you run a 5.1V or sometimes a 5.6V zener at a magic current you can get very low shifts with temperature - it only works with that voltage zener.

[Paul Perry (Frostwave)]

I'd try a 10k load on the zener, to stabilise the current through it somewhat.
But if you ever need a GOOD reference voltage, there are always voltage reference diodes... about 50c, and that's what they are for! I use a LM4040 www.national.com/pf/LM/LM4040.html device in some stuff, but it's overkill for this application.
A curious fault I have seen more than once, is a zener going open circuit when it heats up - exactly what you DON'T want, because it means full volts is suddenly applied to whatever you wanted the stabilised voltage on - usually with catastrophic results. And then the zener cools down & looks 'normal' again. 

[R.G.]

While on breadboard, I was using the voltage divider directly as a reference for the circuit.  After a few minutes things seemed to slip out of bias so i adjusted the trim pots by ear, and it worked again.  I thought, well that was wiered, and checked the voltage divider, and got a reading of 3.5 volts.  I tried a new diode, and it instantly slipped to 3.5 volts.  I thought maybe i was loading it down, so i buffered it with an opamp. The voltage held around 4.7 volts for a few minutes, and then fell to 3.5v again.  I double checked the resistor, thinking maybe it was not resisting enough, and overheating the diode.  Nope.  The resistor read 9.9K on the ohm meter.

Zeners drift, but they're not that bad.

I suspect that you have something else going on, like perhaps your +v source is dropping under load as something else - or the +V source - heats up. The zener is then starved for voltage and drops. It can also happen that you are overloading the zener some way. The buffering says no, but that's a good one to check for.

If it were mine, I'd try that again and when the voltage on the zener drops to 3.5V, I'd measure the voltage on both ends of the 10K feeding it.

[Brian Marshall]

While on breadboard, I was using the voltage divider directly as a reference for the circuit.  After a few minutes things seemed to slip out of bias so i adjusted the trim pots by ear, and it worked again.  I thought, well that was wiered, and checked the voltage divider, and got a reading of 3.5 volts.  I tried a new diode, and it instantly slipped to 3.5 volts.  I thought maybe i was loading it down, so i buffered it with an opamp. The voltage held around 4.7 volts for a few minutes, and then fell to 3.5v again.  I double checked the resistor, thinking maybe it was not resisting enough, and overheating the diode.  Nope.  The resistor read 9.9K on the ohm meter.

Zeners drift, but they're not that bad.

I suspect that you have something else going on, like perhaps your +v source is dropping under load as something else - or the +V source - heats up. The zener is then starved for voltage and drops. It can also happen that you are overloading the zener some way. The buffering says no, but that's a good one to check for.

If it were mine, I'd try that again and when the voltage on the zener drops to 3.5V, I'd measure the voltage on both ends of the 10K feeding it.

That is exactly what i did.  V+ was never below 8.5V.  This is kind of what i thought about the buffering though.  It should aleviate all problems with loading.  I just breadboarded the voltage divider again, and tried it with 4k7, 10k, 22k, and 100k resistors.  all read around 3.5 volts instantly.  I'm starting to think this batch of zeners may simply be mislabeled.

[Sir H C]

It used to be concidered that zeners 5.7 volts and under were just "bad", very weak knee compared to the higher voltage ones. 

[George Giblet]

> nd tried it with 4k7, 10k, 22k, and 100k resistors.  all read around 3.5 volts instantly.  I'm starting to think this batch of zeners may simply be mislabeled.

The quoted zener voltage is specified at quite high currents, often 20mA - your test currents are very low.  Try using a 220ohm resistor.  If the voltage is still low then that means they are probably mislabled.  If the voltage is 4.7V that just means the zener characteristics aren't guaranteed at low currents - a new batch or different brand may help.

The phase 90 uses a 10k resistor with a 5.1V resistor and it's *usually* about 4.8V.

[R.G.]

The quoted zener voltage is specified at quite high currents, often 20mA - your test currents are very low.

Yep. Good point. The datasheets are a bit murky on this one, but a couple of them say to test at 20ma.

Fairchild's datahsheet says that the zener impedance is 19 ohms at 25ma and 1900 at 0.25ma, so you can count on a lot of falloff at low currents.

But still, if a test with 4.7K as a series resistor ( almost 1ma ) doesn't get better than 3.5V, I'd say they may be mislableled too.

[Brian Marshall]

thanks for all the advice guys.  I was starting to think i was crazy for a bit.  I tried higher currents, went as low as a 470 ohm resistor, and the voltage didn't get anywhere near 4.7 volts.  I'm thinking that my original reading on the first one may have just been wishful thinking.