Box of Rock Bias Issues

[Cruton]

Hi everyone,

So, I've built a couple Box of Rock pedals [edit: *modified* Boxes of Rockses, see schematic below] previously, with great results. The last one I built was in the horizontal orientation a la Zvex, with board mounted pots. My friend who received that one as a gift recently asked me to build him another, but without the extra SHO booster. I said sure, and that it'd be easy enough to fit that in a 1590B in the typical vertical orientation. I made a new perfboard layout to fit it in the 1590B vertically:





Based on this schematic:




**note that the 5k resistor R25 is the variable resistor that serves as the gain pot. R8/R9 form the tone pot, and R12/R13 the volume pot. This is the same schematic I had used for my previous builds.

I got it all boxed up and it sounded horrible. With the gain all the way up, it sounded like a raging fuzz, and with gain all the way down it would go into terrible blatty sounds. More concerning, there was this constant low frequency "jack-hammer"-like noise that got more perceptible as the gain turned down. The voltages on Q2 and Q3 checked out fine (~4.5V on drain, ~2.2V on gate, and ~0.9V at source), but Q3 seemed a bit weird (~4.5V - 7.6V on drain and ~0V - 1.7V on source depending on gain control, which seemed right based on my sims, but the gate would only move up to ~2.8V when the gain control was all the way down, whereas my sims suggest it should've hit ~3.8V).

So I went through and checked the trace against the layout and schematic, checked continuity/grounds everywhere, and then went at with the audio probe. This started to get really confusing for me. The audio was blatty and jackhammer-y at the drain of q1, sounded great at the drain of q2, and then SUPER LOUD jackhammer at q3 drain. I could omit any one gain stage by removing the MOSFET and jumpering the transistor socket from gate to drain, and it sounded fine. Put the third MOSFET back in and boom, blatty jackhammer.

This next part is (extra) foolish... I ran an exacto between the traces with it still hooked up, hoping I would clear a bridge and it would spring back to life, and I would have found where the problem was. In doing that, I momentarily connected the top of Q1's drain 5k1 drain resistor (R27) directly to the +9v input, and viola, jackhammer gone! I was sure I'd found my problem, though I had no idea why bypassing the series 82 ohm resistor and 47u filter cap to ground would have cleared up that issue. I dropped a bit of solder to tie R27 straight to the +9v, and the jackhammer was gone.

Then things got really weird. Although the sound was now "clear", the minimum gain setting on the pedal was a super high-gain distortion, and the max gain was a ripping fuzz. AB'ing it with my old box of rock, I confirmed that the minimum gain on the new build is way higher than the max gain on the functional circuit. Hmm.

Back to testing voltages, I found that now all 3 MOSFETs have gate voltages of ~0.8V. Drain and source voltages are still spot on. For q1, changing the gain control still changes the drain voltage from ~4.5 - 7.6V and the source voltage from ~0V - 1.7V, but the gate voltage stays at 0.8V no matter what.

I have cleaned the traces again, traced the board against the layout against the schematic again, and pulled the board out of the enclosure to measure the voltages at every connection. It's just such a dead simple circuit I can't see what's happening-- there's 4.5V at the drain, then on the other side of the 10MEG resistor (the node connected to gate) it's somehow 0.8V for all 3 gain stages.

I've tested the voltages with the MOSFETs taken out of the circuit, and there's ~9.5V at drain, 0V at source, and, of course, 0.8V at gate. Somehow.

I've reached my wit's end (it's a short trip) with this thing. I'd like to just rebuild it but without knowing the issue I'm afraid I'll just end up in the same boat again. Perhaps one of you will be able to look at my layout and say "well there's your problem, dummy". I can only hope.

Cheers

[Mark Hammer]

The Madbean clone of the BoR shows 1M resistors where you show 10M, and also shows larger-value source resistors for the 2nd (180R) and 3rd (330R) gain stages.

[Cruton]

Hi Mark,

Thanks for the reply. Yes, the schematic I'm using differs a bit from the original BOR schematic and the madbean clone. I wouldn't doubt that my choice of component values alters things a bit. However, I think the main issue I'm struggling with is that the voltages I'm measuring don't match with what the simulation shows given my choice of values.

The sim shows:

Q1
        Gain minimum                 Gain Maximum
D:          7.6                                      4.3
G:          3.8                                      2.1
S:          1.7                                      ~0

Q2

D: 4.5
G: 2.2
S: 0.1

Q3

D: 4.5
G: 2.2
S: 0.1

[Rob Strand]

I got it all boxed up and it sounded horrible. With the gain all the way up, it sounded like a raging fuzz, and with gain all the way down it would go into terrible blatty sounds. More concerning, there was this constant low frequency "jack-hammer"-like noise that got more perceptible as the gain turned down.

People have had issues with the first stage of the SHO and the BOR oscillating.   The problem seems to be related to the inductance of the wires to the gain pot going to the source of the MOSFET.  The solution is to add 1K in series with the gate then a 100pF cap from the gate to ground.

There were a few posts about this around end of 2018 and early 2019.

[Cruton]

Interesting, I'd built a couple without board-mounted pots in the past without encountering that oscillation issue. If that had been the source of the original problem, is there any reason that bypassing the power filtering resistor-cap combo would've resolved it?

I'm still totally lost as to what's putting all the gates at 0.8V now that the jackhammer sound is resolved. I've been messing around with the circuit sim, trying to figure out if there's any combination of misplaced connections, drifted component values, etc. that will give me correct drain and source voltages but a super low gate voltage. So far, I've not been able to figure out what sort of mistake would give me that result. Anything I can do that drops the gate voltage messes with the drain voltages too... Hmmm.

[duck_arse]

your meter will load the 10M resistors and shift the bias voltage. if you have the correct source voltage, you don't really care about the gate voltage. can we see pics of what you have built, please?

[Cruton]

Ah, well that's good to know. Then maybe the voltages are fine, and there's some other reason it's behaving much differently than previous builds... Maybe I should go measure the voltages on the previous build with the same meter and see what I get.

Here are pics of what I've put together, top side and bottom side.






I ran out of the large 10M resistors, but had a couple of the smaller ones around that I used to finish up the build. I think the spacing around those will be the only difference between the build and the layout in the original post. Well, should be the only difference...

[Rob Strand]

Interesting, I'd built a couple without board-mounted pots in the past without encountering that oscillation issue.

So in this case there's no wiring to the source of the MOSFET.   When you add wires to the pots the inductance of the wiring causes the MOSFET to turn into a high frequency oscillator.   Oscillating MOSFETs is a known problem in many areas of electronics.

If that had been the source of the original problem, is there any reason that bypassing the power filtering resistor-cap combo would've resolved it?

The cause is related to what is happening around the gate, the source and ground so bypassing is ineffective to a large degree.   Adding the RC circuit to the gate seems to be the best way to squash the oscillation problem.   In the earlier thread I mentioned a few things were tried but the solution which worked was the RC network.  Another build not long after that had the same problem and the RC network fixed that one as well.   IIRC, you might be able use 100 ohm instead of 1k.

[Cruton]

I meant that I'd built some without board-mounted pots (i.e., with the wiring to the source) without running into any oscillation problems. That said, maybe you're right and there's something about this one that made that wiring scheme lead to oscillation when it hadn't before. Then again, if you're right that bypassing that power filter section would be ineffective at getting rid of the oscillation, maybe that's not what caused the noise after all.

In either case, I'm now just left with the weird issue of having waaaaaaayyyy high gain, but an otherwise fully functional (and quiet) effect.

[Rob Strand]

I meant that I'd built some without board-mounted pots (i.e., with the wiring to the source) without running into any oscillation problems. That said, maybe you're right and there's something about this one that made that wiring scheme lead to oscillation when it hadn't before. Then again, if you're right that bypassing that power filter section would be ineffective at getting rid of the oscillation, maybe that's not what caused the noise after all.

When a circuit is *right on the edge* of oscillation the tiny effect of bypass caps could push it into the good zone.     It's like balancing a pin on a wire,  it only takes the air of butterfly wings to make it fall.

It's hard to generalized but it is clear this circuit has some issues.  So it would be wise to bullet-proof it with the resistor and cap.

One other thing even if the pots were PCB mounted the tracks still act like inductors, so a PCB doesn't guarantee freedom from oscillations.

[Cruton]

I see, thanks for explaining that. If I end rebuilding the thing I'll think about adding that resistor/cap combo to gate of q1 to stabilize things a bit.

I'm still stumped on the source voltages. Taking duck_arse's suggestion that the meter is just loading the resistors, I measured the voltages on the old build I had on the pedalboard (also has off-board pot wiring, no oscillation problems). I got the following:

Q1
        Gain minimum                 Gain Maximum
D:          6.4                                      2.28
G:          2.4                                      1.10
S:          2.8                                      0.00

Q2

D: 4.6
G: 2.2
S: 0.1

Q3

D: 2.7
G: 1.3
S: 0.1


I dunno what's up with q3 on that one, but hey, it sounds great. Anyway, not trying to get my new build to match those same voltages, just would like it to have reasonable ones.

Following up on duck_arse's comment about gate voltage not mattering, I found this old thread https://www.diystompboxes.com/smfforum/index.php?topic=110930.0 about debugging a BOR where Paul. R explains the way the MOSFET biases...

> What causes the voltage to drop in half at the d side?

It doesn't. The MOSFET's Gate voltage is naturally about 2.2V. But this won't happen until the Drain puts about 4.4V across the 1Meg+1Meg resistor string to the Gate.

So it isn't half of 9V. It is twice the MOSFET Gate voltage, forced by the Gate resistors. (These were selected to put Drain near half of battery voltage.)

So, I feel like that's starting to get at some solution to my problem, or at least a way of interpreting the ~4.5V drain but a gate voltage of 0.8V, but I'm still a bit lost as to what sort of goof-up on my part would produce this voltage scenario.

[Rob Strand]

D: 2.7
G: 1.3
S: 0.1

I dunno what's up with q3 on that one, but hey, it sounds great. Anyway, not trying to get my new build to match those same voltages, just would like it to have reasonable ones.

The problem seems to be the MOSFET itself.  The gate voltage is much lower than the others and that causes the drain voltage to be low (due to the way the biasing works).     Pretty much all MOSFET circuits I've built have gate (to source) voltages around the 2V to 2.2V mark.   If it were my unit I'd probably swap Q3.  If you like the sound of the 1.3V MOSFET then keep it in there but maybe you should compareit with a "good" one first.

[Cruton]

That makes sense. Is there anyway that a bad MOSFET could also result in the weird gate voltages in the new build too? Though I guess based on what I've learned here, I would think that if that were the case, a 0.8V gate would result in a ~1.6V drain... hmmm.

After I built that old one with the possibly bad Q3, I had a couple of spare MOSFETs but not the 3 I needed for this new build, so I ordered a handful more from Mammoth and just tossed them in the parts bin with those other two. So, not sure which ones ended up in this build, but I've tried swapping them around with no real change. I wonder if I ended up with a handful of bad ones by chance.

[Rob Strand]

That makes sense. Is there anyway that a bad MOSFET could also result in the weird gate voltages in the new build too?   

Sure.  The gate to source voltage of a MOSFET is a characteristic of the each *individual* MOSFET.  MOSFETs of the same type will vary a small amount around an average value.  If you have a "bad" MOSFET it will always be bad.

Though I guess based on what I've learned here, I would think that if that were the case, a 0.8V gate would result in a ~1.6V drain... hmmm.

That's correct.  The circuit produces a drain voltage which is double the gate voltage.  So for your Q3 measurements 2x 1.3V  = 2.6V, which is quite consistent with the measured drain voltage of 2.7V.   So the MOSFET is working but it's not working like normal MOSFETs of that type.

One small point is if your multimeter is has a 1M input impedance.  The measure gate voltage will be somewhat lower than the real gate voltage, say 0.67 times, because the 1M meter impedance loads down the 2x1M resistors on the gate.  A 10M input impedance multimeter will only drop the voltage by a factor of 0.95.   The drain voltage is hardly affected by the meter.

So working backwards from 2.7V.  We expect a gate voltage of 2.7/2 = 1.35V.  The allowing for 10M meter loading we get an expected measurement of 0.95 * 1.35 = 1.28V which agree with your 1.3V.  So perhaps your meter is 10M input impedance.

After I built that old one with the possibly bad Q3, I had a couple of spare MOSFETs but not the 3 I needed for this new build, so I ordered a handful more from Mammoth and just tossed them in the parts bin with those other two. So, not sure which ones ended up in this build, but I've tried swapping them around with no real change. I wonder if I ended up with a handful of bad ones by chance.

The way to tell would be to measure the circuit voltages.    You know the badness is a low gate voltage and drain voltage.

[Cruton]

Thanks for explaining the meter loading in better detail. That leads me to believe the 0.8V on the gate of the 3 different MOSFETs in the new build couldn't be better the meter loading alone.

That also leads me to have no idea how I could get 0.8V at the gate and 4.5V at the drain, if the drain is double the gate voltage.

Sorry you're having to spell this one out for me, and thank you very much for your help.

[Rob Strand]

That also leads me to have no idea how I could get 0.8V at the gate and 4.5V at the drain, if the drain is double the gate voltage.

Sorry you're having to spell this one out for me, and thank you very much for your help.

It's hard to understand because it doesn't make sense.  If I saw those measurements I would start thinking the MOSFET is damaged somehow.

If the drain is 4.5V then you would expect to measure a gate voltage between 1.5V (1M impedance meter) and 2.25V (>10M impedance meter).   To get 0.8V you would need a loading of 276k to pull the gate voltage down.  A meter won't do that.  On a good MOSFET no DC current should flow into the gate but if it's damaged then perhaps the gate is leaking current and pulling the gate voltage down.

[Cruton]

Maybe I'll order up some more BS170, keep track of the new ones this time, and pop 'em in there to see if that fixes it.

With the MOSFETs removed from the circuit I get a sensible 9.5V at drain and 0V at source, but still 0.8V at all three gates. That makes me think I messed up the layout somehow, but I haven't been able to find my error. There's just not that much going on in there, I dunno how I'm so blind to it.

[duck_arse]

Cruton - pardon, I hope I haven't confused you, but I always get my source and drain confused. so I shoulda said if you have the correct drain voltage, you don't really care about the gate voltage.

[Cruton]

Thanks for the clarification duck_arse (Stephen? Just saw that). Can you help me understand why that's the case? I'd been trying to make my way through this article https://www.electronics-tutorials.ws/amplifier/mosfet-amplifier.html on MOSFET biasing, and it seems to suggest that if Vgs isn't greater than a threshold value then the MOSFET shouldn't turn on.. I haven't wrapped my head around this enough to understand how meeting that criterion or not would affect the drain voltage you'd get.

In looking up the datasheet for the Fairchild BS170s I have to see what the threshold is, and while the typical is 2.0V, the minimum is, interestingly enough, 0.8V. Maybe I'm getting closer to the culprit.

If this batch has a threshold voltage of 0.8V, could that cause the huge increase in gain relative to my other builds with 2.0V at the gates?

[Cruton]

Okay, so this is... SUPER embarrassing.

Just out of curiosity, I dug out my old, nicer DMM to see if I would get the same readings. I had been taking readings using a cheaper one that I bought on sale for no other reason that it had transistor terminals for measuring Hfe.

The gate voltages are spot on when read from my old DMM (~2V). They are 0.8V when read from the new cheap-o meter.

Lesson learned-- don't buy cheap s***. I also am now highly suspicious of any use that new DMM could be giving Hfe readings.

As far as why it sounds so much higher gain... no idea. Some more debugging to do there I suppose, but at least it's not because it's all biased all screwy or something. If I can't sort it out maybe I'll just swap out the source resistors on q2 and q3 for the stock BOR values or something higher to try to match the sound of my previous builds. Still, I've confirmed I used the 100R's in the previous ones, no idea why it's different now.

Thanks for your help, and please accept my sincerest apologies for not having done a second meter verification the first time around.