What else does the static-protection zener do in MOSFET circuits?

[earthtonesaudio]

You know the anti-static technique of putting a 9.1V Zener diode from Source to Gate on MOSFET circuits (like the SHO and many others)?  I was thinking about this, and doesn't this effectively create a low-value resistance for positive-going signals that exceed 0.7 Volts?

If so, what does this do electronically?  Are those signals "bootstrapped," and therefore the impedance for them is raised?  Or, does it change the gate bias for positive signals only?

Clearly it doesn't have a huge effect on the sound, but I was just wondering what's actually going on there, other than static protection.

[R.G.]

What it does is create a low impedance for negative going signals larger than 0.7V. The zener is set up with cathode to gate, anode to source for n-channel MOSFETs. This is in expectation of a bias voltage of 1-3V. So in effect, negative-going signals have to be more negative than a diode drop plus the bias voltage to turn the diode on.

Some MOSFET protection schemes have two zeners in series-opposed so the gate can go +/- the zener voltage from the source.

[earthtonesaudio]

Cool, thanks.  So I guess if I want to keep the Zener from influencing the sound, I'd just make sure the signal can't turn them on.  That's pretty simple.

[DougH]

Another thing to keep in mind is that the main purpose of this is to protect the first stage. My understanding of the error mode that this protects against is static charge on a cable plug (from rubbing your feet on the carpet or living in Minnesota...) that is discharged through the MOSFET gate when the cable is plugged into the input. 2nd stages and beyond in a mult-stage MOSFET circuit should not be vulnerable to this, unless I have misunderstood the error.

[Gus]

It is not a static voltage as much as a low voltage.  Some mosfets "glass" "under" the gate break down at 20VDC gate to source
http://www.fairchildsemi.com/ds/BS/BS170.pdf

[R.G.]

I used to think it was first stage only, too.  But if you think about it, that only applies inside integrated circuits. What happens if you're building a pedal or repairing one and reach in and touch the board in the middle?

Maybe cheaper than a zener is a diode clamp from gate to V+ and gate to V-, both diodes normally reverse biased. That works as long as you can guarantee your power supply never goes over 20V. Both schemes are used in ICs.

[Bill Mountain]

WARNING!!! ZOMBIE THREAD!!!

I didn't want to make a new thread because R.G. touched on something I was wondering in his last post in this thread.

If my main concern with a the first stage is keeping the incoming signal below a certain voltage I can use anti-parallel diodes to ground like this:

http://www.runoffgroove.com/azabache.html

If I also want to use a MOSFET as the first gain stage would the anti-parallel diodes to ground still provide protection to the MOSFET?  It won't be going directly from gate to source but if the input signal never exceeds 1 or 2 volts (because of the diodes limiting the input voltage) would I be at risk of breaking the MOSFET?

If I put the diode clamp as R.G. mentioned would I still also limit the input voltage to the diode voltage?

[R.G.]

If I also want to use a MOSFET as the first gain stage would the anti-parallel diodes to ground still provide protection to the MOSFET?  It won't be going directly from gate to source but if the input signal never exceeds 1 or 2 volts (because of the diodes limiting the input voltage) would I be at risk of breaking the MOSFET?

If I put the diode clamp as R.G. mentioned would I still also limit the input voltage to the diode voltage?

Things get tricky here. MOSFET gates rupture suddenly and abruptly on any voltage significantly bigger than what the datasheet specifies. And that's the voltage on the gate pin, not anywhere else. Unless the protection is right at the gate pin, if there are resistors and such between the gate and the protection stuff, then one touch with a winter-staticized finger can zap the gate while the stuff elsewhere is charging up.

The problem with the input signal never exceeding one or two volts is the same static finger from handling, or touching something that's got 120Vac leakage on it, even at currents so low you can't feel a tingle.

If you put the BTB diodes right where the input wires come on board, you're probably OK - you've excluded a batch of possible situations which could kill the MOSFET, but possibly not others.

A zener on the gate/source is so quick and easy, it's cheaper than thinking about all the other things you might do. But yes, there are other ways.

[Bill Mountain]

Thanks for the wonderful explanation as always.

I'm looking at putting the diodes on the input to limit the max input voltage.  I can also easily add the G-S Zener.  I just didn't know if the Zener would become superfluous or not.

Thanks!

[amptramp]

In some cases, a zener will have a significant capacitance across the junction.  Some people put antiparallel diodes in series with the zener to isolate these capacitances.

[Bill Mountain]

In some cases, a zener will have a significant capacitance across the junction.  Some people put antiparallel diodes in series with the zener to isolate these capacitances.

Does this effect the sound?

[R.G.]

In some cases, a zener will have a significant capacitance across the junction.  Some people put antiparallel diodes in series with the zener to isolate these capacitances.

(1) for some values of "junction"; MOSFETs don't have junctions in the normal sense, but it does appear across the gate-source - which is itself a capacitance of some significant value. For most MOSFETs, it's trivial, some it's not, and for many this is bigger than the zener capacitance. At least for modern MOSFETs. The old 3N series of MOSFETs had very small junction capacitances, and this might well have made a difference there.
(2) the antiparallel diode scheme only works for signals and voltages small enough that the forward diode doesn't get turned on. When it does, the diode becomes a low-value resistor and turns on the connection to the zener capacitance. The problem then becomes the effect of the zener capacitance being switched in and out of the circuit by the diode as signal changes. This may or may not be worse than leaving a capacitor in there all the time.

A lot depends on the source impedance and how much, if any, unbypassed source resistance is in the circuit. If the source resistance is big, the transconductance winds up actively minimizing the gate-source impedance to some extent.

Does this effect the sound?

(1) Yes.
(2) No.
(3) For some values of signal and circuits. The Devil is in the details.

[amptramp]

In some cases, a zener will have a significant capacitance across the junction.  Some people put antiparallel diodes in series with the zener to isolate these capacitances.

(1) for some values of "junction"; MOSFETs don't have junctions in the normal sense, but it does appear across the gate-source - which is itself a capacitance of some significant value. For most MOSFETs, it's trivial, some it's not, and for many this is bigger than the zener capacitance. At least for modern MOSFETs. The old 3N series of MOSFETs had very small junction capacitances, and this might well have made a difference there.
(2) the antiparallel diode scheme only works for signals and voltages small enough that the forward diode doesn't get turned on. When it does, the diode becomes a low-value resistor and turns on the connection to the zener capacitance. The problem then becomes the effect of the zener capacitance being switched in and out of the circuit by the diode as signal changes. This may or may not be worse than leaving a capacitor in there all the time.

I think we are talking about small-signal MOSFET's here.  There are Zener diodes with a sufficiently high junction capacitance that they have been used, reverse biased but below the breakdown voltage, as tunable elements in audio filters.  Some zeners have as much as 15 nF of junction capacitance - that's 15,000 pF.  If you use zener protection, you can isolate this capacitance with a pair of series diodes so that the approximately 4 pF of a 1N4148 is in series with the zener.  The zener diode can be placed in series with a pair of antiparallel 1N4148's, meaning the zener capacitance is in series with the 8 pF and the total is slightly less than that.  Protection diodes are never intended to be turned on in a linear amplifier, so the amplifier behaviour is not really an issue.  In power supply circuits, the effect of the protection diodes may be significant if there are drive signals that can turn the protection diodes on.

[R.G.]

I think we are talking about small-signal MOSFET's here.

Yes, we are. Also small zeners.

There are Zener diodes with a sufficiently high junction capacitance that they have been used, reverse biased but below the breakdown voltage, as tunable elements in audio filters.  Some zeners have as much as 15 nF of junction capacitance - that's 15,000 pF.

I'm sure there are. But most of the smaller ordinary zeners are down in the 100pF range. Still not trivial, but the gate-source capacitance of the 2N7000 is typically 20pf, max 50pF. So the zener adds  a comparable, but not wildly larger amount. The junction capacitance of a zener depends on the junction area, and this in turn is proportional to the power of the zener, as you need lots of junction area to handle lots of current flow. So lower power zeners will have smaller capacitance than same-voltage zeners rated for higher power - the area of the capacitor "plates" is smaller.

If you use zener protection, you can isolate this capacitance with a pair of series diodes so that the approximately 4 pF of a 1N4148 is in series with the zener.  The zener diode can be placed in series with a pair of antiparallel 1N4148's, meaning the zener capacitance is in series with the 8 pF and the total is slightly less than that.  Protection diodes are never intended to be turned on in a linear amplifier, so the amplifier behaviour is not really an issue.  In power supply circuits, the effect of the protection diodes may be significant if there are drive signals that can turn the protection diodes on.

Yes. This is a useful trick if you find that a zener changes the sound of your MOSFET. In practice, I've never been able to notice the difference with/without a zener. It would be interesting to run some lab tests of frequency responses with a zener and zener plus diodes versus bare MOSFET. For guitar, let alone audio, it's probably fine, depending the source impedance driving it.

A quick of calculation says that 100pF and 100K (the Zl of a nominal guitar pickup at treble frequencies before self capacitance cuts it back down) is about 16kHz. So yeah, adding 100pF could just barely start affecting sound at the top end of audio for a 2N7000 with no unbypassed source resistance.

Once again, we discover that details matter!      The moral seems to be if you think your MOSFET sounds dull with a gate protector zener, either put in isolators or pick a lower capacitance zener. Or go without protection. 
All quite true - it is possible to find zeners that have wildly larger capacitance