MOSFET input buffer weirdness [Madbean's Cosmo/screwdriver content]

[midwayfair]

This is a cross post from Madbean's forum; I figured I'm more likely to get a "why" answer here even if someone there knows how I can fix it.

Here's the build document with schematic for reference:
http://www.madbeanpedals.com/projects/Cosmopolitan/docs/Cosmopolitan.pdf

So here's what's going on:
The MOSFET does not work at all in proper orientation, and it works very poorly in an incorrect orientation.

Here's what I've done:
-I tried every mosfet I had, BS170s and 2N7000s, about 10 of them from both Tayda and Smallbear, and nothing works, so it seems unlikely that it's just an out-of-spec MOSFET.

-I also know that the circuit and layout is fine because every bipolar silicon I tried works properly (but it's loud and has a little more noise) as do FETs (when oriented correctly).

-I've verified that the MOSFETs I have on hand work in another pedal (but not another screwdriver, since I don't have an unsoldered socket), and I've used MOSFETs from this batch in other recent builds (two ROG tremolos and another Screwdriver).

Here's my voltages:

D 4.6
G 8.5
S 4.6

Which is just plain weird because there's definitely only 2.7K resistance to the drain, so I don't understand how it can be so low. I get similar voltages regardless of which BS170 I plug in there. I didn't bother taking voltages with a 2N7000.

-I've triple checked the datasheets, and the Tayda ones have the normal DGS pinout (and SGD for 2n7000). So if they were going to work, they'd work

--But here's where it gets weirder: If I swap the gate and the source, I get some sound, but it's gated. My voltages also look slightly more normal (9.4, 8.8, 0v), but they're still waaayyy wrong according to my other build (which reads ~9, 2.5, and .5v, which is exactly what I'd expect to see).

Any one have any ideas? I haven't changed the bias resistors yet. I've read RG Keen's article on biasing MOSFETs, but I couldn't divine from it a reason for the extremely weird voltages I'm getting. (Also, I well and truly suck at math.)

I guess if I can't get a MOSFET to bias in there, I could just change C1 and R2 to reduce the volume to tolerable levels, but I'd prefer the MOSFET because it's low noise.

One thing I haven't had a chance to do yet is breadboard just the input circuit for testing, just in case somehow every MOSFET I have is broken. I wanted to see if there was a more obvious fix before doing that.

[ORK]

D2 ?

[midwayfair]

D2 ?

You mean the Zener? My multimeter says it's fine, and if it were shorted, a bipolar or FET wouldn't work, right?

[R.G.]

Take the MOSFET out of the board. Apply power. Measure voltage on the holes where the MOSFET should go.

You should read:
D - 9V (or whatever your "9V" really is, minus about 0.4V for the Schottky protection diode)
G - half of what you read at the drain
S - 0V

[slacker]

Check R1 is properly soldered, if it's not then R4 will pull the gate voltage up to near the supply, turning the MOSFETt completely on, this will give you the voltages you're seeing. Also check the values of R1, R2 and R4.
Another check would be to measure the gate voltage point with the MOSFET removed, you should get about 4.5 volts.

EDIT: posted the same time as R.G.

[midwayfair]

Hi, guys, thanks for the help.

First, I was getting 8.8v on the gate. (This is what I get when measuring on the "far" side of a 1M resistor from the +9v, so I'm not sure why this was expected to be 4.5v?) To get it down to 4.5v, I would have had to go all the way up to 6.8M. I don't think the circuit was intended to need that kind of adjustment.

I breadboarded a small test circuit -- just a simple booster/buffer, basically Jack Orman's -- and used this to verify that indeed every MOSFET I have must have gone bad. Really weird, but it's the only reasonable explanation. Maybe I got some static into the baggie they were in or something (it was just a normal plastic bag), and it's been a while since I had last plugged them into something else. I'll probably end up using them as clippers, since my multimeter still registers normal voltage drops across the ones I tested.

[R.G.]

First, I was getting 8.8v on the gate. (This is what I get when measuring on the "far" side of a 1M resistor from the +9v, so I'm not sure why this was expected to be 4.5v?) To get it down to 4.5v, I would have had to go all the way up to 6.8M. I don't think the circuit was intended to need that kind of adjustment.

The gate is connected to a string of resistors between +9V and ground. From 9V, there is a 1M, a 100R, and another 1M before ground. The 100R is so small compared to the 1M's we'll ignore it. So there are two 1M's between +9V and ground, and the center point must be at about 4.5V, depending on the actual tolerance of the two 1M's.

If your gate point is at 8.8V, it leads to the problem I suspected and slacker actually posted, that being that the gate was pulled up, saturating the MOSFET. Your measurements support this guess.

I think that you do not have a bad MOSFET. I think you have a bad connection between the gate point and ground through the 100R and the second MOSFET.

I breadboarded a small test circuit -- just a simple booster/buffer, basically Jack Orman's -- and used this to verify that indeed every MOSFET I have must have gone bad. Really weird, but it's the only reasonable explanation. Maybe I got some static into the baggie they were in or something (it was just a normal plastic bag), and it's been a while since I had last plugged them into something else. I'll probably end up using them as clippers, since my multimeter still registers normal voltage drops across the ones I tested.

I think that you ought to not dispose of those just yet.

[midwayfair]

Holy crap, I can't believe I missed this after going over the schematic so many times ... I connected the 1M to ground as a pull-down instead of after the input AC coupling cap. (Thanks RG for the really specific breakdown of the voltage divider network ... I never would have seen the mistake otherwise). I was just wiring it as "normal" without thinking.

Thanks for the help guys!