should zener diodes be avoided in low noise circuitry?

[ftransform]

I am using a zener diode on a power rail circuit monitoring voltage but I am wondering if my power rails can be contaminated for sensitive electronics (this is a DC analog board). The zener has a ferrite bead and a decoupling capacitor. Basically a circuit that will issue an alert when the voltage is out of spec (more then +-5%).


I thought this was a great solution but then when I analyzed the situation on the oscilloscope (cheap rigol) I noticed that when given the right voltage the zener diode starts throwing turds like a irate monkey.

I cannot really measure the noise properties of my ebay zener diodes (I am trying to design this board to the best of my abilities) so I am asking if these guys are problematic when they are operated say 200mV above the point where they start spewing tons of highly visible on my cheap oscilloscope noise. I noticed that the noise point of my diodes is really sharp (hard to get the power supply to stay at that point) , but so long I can avoid this voltage are they considered to be problematic?

I do not care about the output noise of the power supply monitoring circuit (it is digital) but I do not want my rails contaminated.

[ftransform]

In order to be more specific, will zener diodes will be effecting power rails which are supposed to be ~ 60 microvolts of RMS noise.
Will zener diode, 5V, operating at low current (powering transistor gate) effect the noise level of my rail significantly?

When I put them in "noise mode" then the measured noise (putting a bead on a lead at the transistors that its connected to grounded pin with a near by 10uF tantalum + 100nF ceramic +10nF ceramic) is ~8mV rms wide band "white" noise which looks downright putrid. I cannot measure what they put out when they are at a normal voltage though, I do not have the money for a low noise preamplifier. I do not have anywhere near the equipment to test ripple noise in the <1 milivolt region. I suppose the ultimate test is to make the board and test it with a ADC, but that's a bit pricey.

help

A graph of noise spectral density vs current of typical diodes would be appreciated.
An uncited forum posts claims a noise spectral density of 50nV/sqrt(Hz), much less then my linear regulator.

[AndyC_772]

First thing to note is that many Zener diodes have a tolerance of +/-5% or so on their Zener voltage anyway, so as a way to detect whether or not a supply rail is within 5% of nominal, they may not represent an ideal method.

The noise itself can probably filtered out, though. Can you post a schematic so we can make suggestions?

[rf-loop]

Zener diodes are well known noise sources.

They have also been successfully used in the construction of cheap "noise generators".

http://pdfserv.maximintegrated.com/en/an/AN3469.pdf

http://bytecollector.com/archive/misc/BuildYourOwnNoiseSource.pdf

Btw,, noise generator is very useful test instrument and professional ones are also expensive.

[ftransform]

I understand this but I would like to know some figures for zener diode noise when it is not in its noise mode, i.e. normal operation.
I found that for my diode there is a 15mV range where the noise is very bad. Outside of this range the oscillations stop.

searching for zener diode noise spectral density turns up nothing.
I would like a graph similar to the one included in linear regulator datasheets.

The idea is to use a carefully selected zener at the gate of a transistor to indicate if there is a situation.

[gnif]

I would use a 5V buried zener voltage reference and a pair of precision operational amplifiers configured as a Schmidt triggers to set your upper and lower bounds on the voltage. I have been using MAX6350's for this kind of thing, and Maxim provide free samples, so wont cost you anything to have a play.

[kxenos]

I would use a VRef to generate 105% of the rail voltage. From this, with a resistor divider, I would get th 95% voltage. And then use these voltages to compare with the rail.

[ConKbot]

Something like a TL431 adjustable reference could be used as an almost drop-in replacement for the zener, if you have a set circuit in mind. Just add another divider for the input pin and it can regulate the output to 5v,

If you dont have a set circuit in mind it could be a bit more flexible too.

[AlfBaz]

Your opening post had me thinking and I've been tinkering with noise analysis in spice with various models at different zener currents.
From the bits I have found on the net the larger the current through the Zener the less the noise, and this seems to hold true in the spice tests.
It "appears" that at the "knee" in the reverse breakdown region is where the most noise occurs.

Going on your description and not having seen the cct, would it be true that when the rail voltage increases to the zener voltage, it conducts and this current is used to turn a transistor on?

[SeanB]

Zeners are unfortunately noisy devices, part of the design. At around 6V they are the quietest, but get noisy as the voltage goes up. You need to bypass it with a RC filter, as this will reduce the noise without oscillating if the circuit has negative resistance.