Mosfet protection diode 2

[sbirkenstock]

Hi Everybody,

I started the topic "Mosfet protection diode" a couple of days ago.
I don´t know how to inform everybody that there is some "important new input",
so I created a new task. Please let me know if there is a better method.

Here again is the schematic I am refering to:
http://runoffgroove.com/peppermill.html

I made a little test "bread".
I took an 18v power supply and put it through a 1Meg linear pot to ground.
So I can adjust the wiper from 0 to 18V.

First I applied the voltage to an LED going to ground. (just the led, no mosfet included here)

a)
If the LED is directed like in the schematic of the peppermill, and I apply the variable voltage to the cathode of the LED,
I can "turn up" the voltage to 18 volt. The LED is totally blocking it from ground.
Seems like the protection in the papermill is not working at all?

b)
When I turn the LED and put the cathode to ground and apply my variable voltage to the anode of the LED,
then at around 1.6V the LED starts lighting up, and if I increase the voltage further, the LED will burn brighter,
but the voltage still goes further up.
I did not try it further than 5 volt, but I guess somewhere the LED will burn up.
Of course my source is strong and durable, other than maybe a short peek from a static discharge, but still,
this does not seem to protect the gate.

Then I tried a Zener Diode (9.1V and 0.5 Watt)
c) The cathode of the Zener Diode is grounded,
the variable voltage applied to the anode.
Then at just a bit over 0.5 volt the diode will conduct.
That would protect the gate a lot, but the signal will be very limited.

d) now I grounded the anode of the Zehner Diode and apply the variable voltage to the cathode.
I turn up the voltage and it raises to 9.2 volt, then it stops raising. Cool!
That seems to be THE way to protect the gate.
BUT: at around 17 volts, the voltage on the cathode is rising again.
So possibly the voltage still could rise above 20V to destroy the mosfet gate, if the voltage is high enough.
So why is this happening? Am I acceding the 0.5 watt of the diode?
I did not have a higher wattage diode, but I just odered it to try it out.

Any comments?
Best regards,

Stephan

[amz-fx]

Stephan,

As I stated in your previous topic, the LED is not good protection.

After there were a number of posts made in that topic, I decided to put some LEDs on a breadboard to measure the reverse voltage characteristics. I had 3 different LEDs for which I have a datasheet, and they all say max 5v reverse voltage limit.  The spec is only the guaranteed reverse voltage but does not reflect the actual performance of the LEDs, which will be above 5v.

I only had a 22v wall wart by my test board, but I used it to measure not only the 3 LEDs for which I had datasheets, but also a handful of other LEDs that came from Ebay that I do not have datasheets for.

When I applied 22vdc in reverse polarity to the LEDs, not one of them conducted. (I limited current to max 10 ma. to protect the LEDs.)

Since the spec of the BS170 is shown as 20v max, every LED that I tested would allow potentially harmful spikes into the gate of the mosfet. You are just as well off using no protection than with an LED in place.

The voltage of a zener is not rigidly fixed at its specified rating. Once it begins to avalanche, as the current increases through the diode, the conduction voltage begins to slowly increase also. I suspect that as you are raising the voltage to 17v, the current through the diode is increasing too. The good news is that most any spike into the input will be very low current even though the voltage may be high. You can also add a series resistor on the input before the 0.022uF capacitor to limit current; I suggest 100 to 470 ohms.

Just use a 9v to 12v zener and it will be okay in your Peppermill.

Most mosfets get fried in pedal circuits when the input audio cable is being plugged or unplugged, possibly due to the power switching connections that are often used with a stereo input jack. If I remember correctly, the original SHO had the power switching on the output jack to prevent this interaction.

Best regards, Jack

[sbirkenstock]

Jack,

thank you very much!
Also for your great web site!

Best regards,

Stephan

[Rob Strand]

Great work!

I did the same experiment on three different colored LEDs. 
I used a 20V max rail + 100k resistor (for the hell of it).
Got the same result -  not one showed any obvious form of breakdown.

I'm sure I've seen them breakdown in the past so I hunted out an old LED and it broke down at 12V.    I found another that looked the same and it broke down at 9V.   I checked another 5 unknown origin LEDs of various colors, but still old, and they did not break down.  (Unfortunately I've chucked out a lot of the old LED because they are sooooo dim compared to the new ones.)   So maybe only the ancient LEDs show any breakdown.   It's quite possible some of these LEDs are pre-1980.

FWIW, at low currents zeners will start to break down at a lower voltage than the specified valued.  Often higher wattage zener will show this effect worse than lower rated ones.   The effect is more prominent  for lower voltage zeners:  < 5.1V sloppy VI curve  > 6.8V sharper VI curve.    For a 2V to 3V gate source voltage a 9V zener shouldn't have much effect on the biasing of a high impedance circuit.

[amz-fx]

Rob,

No doubt, that as manufacturing techniques improved on LEDs, the reverse voltages went up, but they apparently have just left the datasheet spec at 5v as in previous years.

Interesting results on the old LEDs. I have some quite old ones too, so I may hunt them down and see if I can get them to breakdown. If I get any results, I'll post here.

Best regards, Jack

[Rob Strand]

Thanks Frank Jack.  I checked some more. I can find only two LEDs that breakdown out of at least 7 which are "old and dim".  (I just update my previous post.).

The way the 5V spec is given is like < 10uA  at 5V.     Like you say,  in the bad old days they might have been closer to 10uA but in the new era they *way* lower.

[Rob Strand]

I couldn't resist finding some numbers on the web.

Found this 2005 patent:
https://www.google.com.cu/patents/CN1697203A?cl=en

Which makes the statement:
"Since the semiconductor LED breakdown voltage is typically about -20V"

One channel of my PSU died recently so I can only get to 21V or so - I know what's wrong I just need to fix it.

[PRR]

> BUT: at around 17 volts, the voltage on the cathode is rising again.

AND pulling-up the Source of the MOSFET.

The voltage *across* the MOSFET gate insulation is not exceeding the Zener voltage.

You are on stage in very tight pants. A crazed fan grabs your left foot and pulls. Your right foot is planted. When your feet are pulled more than 20 inches apart, your pants split. Bummer.



For the next show, you tie your feet together with a 9 inch rope. Now when your left foot is pulled, your right foot is pulled-along. No split, no major tailor bill for sequined satin repairs.

That's what the 9V Zener does. The distance between your feet (voltage across your Gate) does not approach a dangerous separation, the excess distance (voltage) appears somewhere else. Other-foot slide, source pin pulled-up.

BTW-- the Zener alone is not enough. There probably should be 10K-50K between any source of danger (generally the input jack) and the Zener/Gate junction. This causes almost no audio loss. It does limit the current when things get too wild. In this case, 33K at Gate and perhaps 1K at Source, the Source would not be pulled-up more than a wee bit.

Assuming the Zener can carry 10mA, a 33K resistor allows 330V(!) for considerable time. Zeners will carry far higher current for an instantaneous "zap", so it is also protected against shocks like you get from door-knobs in winter-time.