Having problems with a MOSFET Booster build.
I originally built on vero and had this same problem, so for the second go around, I etched a board using MarkM's PNP image and layout from this thread: http://www.diystompboxes.com/smfforum/index.php?topic=54619.0
I measured all the resistors before the second build, and they checked out ok. I also tried replacing the BS170 in case it was faulty. No dice. I'm still getting these voltages:
(http://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=48803&g2_serialNumber=1)
Any ideas about what would cause that low gate voltage? Probably voltage drop across the 10M resistor, but why is the gate drawing current? What should I do next?
It looks to me like you have current from source to gate across the diode. .52v looks like a diode drop to me. Make sure you have the right component there.
Thanks for the idea. I removed the protection diode temporarily and also disconnected the 47p capacitor in case that was leaky, and the voltages are still the same. It's very strange.
Does it play OK ??
You can not measure the gate voltage with any ordinary meter. Gate is biased through 10 Meg. Your meter is 10 Meg. When you attach the meter, this voltage drops to about half.
If you have a second meter: monitor the source voltage while you poke the gate. I bet it drops.
Or: shunt the 10Meg gate resistor with 100K. Now the meter loading causes only a 1% drop, and you should see the nearly-right gate voltage. It will even play this way, though the 10Meg ensures "no" loading of the precious guitar signal.
Hi
I'm no expert, but 10M is a lot of resistance.
Quotethough the 10Meg ensures "no" loading of the precious guitar signal
I can see that it is a small load with respect to current, But it is a massive load in the sense of voltage - very 'stiff' - and prone to picking up static and noise. 1M is a much more sensible impedance. For most pickups tone doesn't suffer much until less than 220k.
cheers
Ohms law will help to understand what is happening.
Read PRRs post again and then look up your meters specifications (as PRR posted it should be around 10meg because of the voltage you measure). The input resistance in DC voltage measurements.
First thing, double check your pin-out on the MOSfet.
Also, I concur with Bret: Just use 1 or 2.2 Meg for R3 since it is well enough and avoid some problems.
Thanks gents. Great advice about the loading.
Quote from: PRR on March 05, 2013, 01:51:30 AM
Does it play OK ??
It sounds fuzzy instead of clean. O-scope shows entire bottom half of wave getting chopped off, even with 1Vp-p input sig. So that is like... saturation or something? It is amplifying, though. Output signal is much bigger than input, just clipped on the bottom.
Later today I'll try the 100k resistor for testing as you suggested and double check gate voltage with that in place.
Quote from: jymaze on March 05, 2013, 10:48:16 AM
First thing, double check your pin-out on the MOSfet.
That was my thought, too, so I even switched out the BS170 with a 2N7000 (flipped to correct pinout) in the vero build in case it was a bad batch or they had weird pinouts compared to normal BS170. Same problem. I will try that in this build, too, though... because who knows what will work!
Also, I am not sure your voltage readings are correct because of the high impedance node.
I have an hypothesis: Your booster is working just fine but there is just too much clipping by design!
As is, there is a lot of gain with this booster and it is normal that it gets fizzy because there is quite some clipping occurring.
If you want to keep it as clean as possible clean and use the full range of the 9v without clipping, add a trimmer in series with C5 and trim for max gain without clipping. Probably you can just put a 220 resistor in series with the C5 and call it a day too. There might be some clipping but possibly nothing you could identify as fizzy.
Even better: Build an SHO instead, it is a much better design to me, going from quasi-buffer, to clean booster, to slight overdriven sound. You just have to change a few components so it is not daunting, but the added value will be worth the effort I guarantee. Don't worry about having a C-law pot, B-law works just fine.
I positively LOVE mine, one of the most useful pedal ever and definitely the most simple!
Quote from: jymaze on March 05, 2013, 12:48:57 PM
Even better: Build an SHO instead, ... I positively LOVE mine, one of the most useful pedal ever and definitely the most simple!
Sounds like I need to try an SHO either way, whether I get this build working or not!
One simple test to do is disconnect C5. This sets the gain at about X1. Then you can do the series resistor and 100uf cap in parallel with the source resistor textbook gain setting circuit fragment jymaze posted this addition as well. I would try a 5K external control (about 1.75K resistance setting at min 5K in parallel with 2.7K). This would give a min gain of about X1.5
Did you look up Ohms law?
Quote from: Gus on March 05, 2013, 02:17:03 PM
One simple test to do is disconnect C5. This sets the gain at about X1. Then you can do the series resistor and 100uf cap in parallel with the source resistor textbook gain setting circuit fragment jymaze posted this addition as well. I would try a 5K external control (about 1.75K resistance setting at min 5K in parallel with 2.7K). This would give a min gain of about X1.5
Did you look up Ohms law?
I'll try removing C5 as well later today when I am experimenting. I understand the basics of Ohm's law and use it regularly, but I don't always know where/how to apply it---I sometimes don't know what properties of what components matter at a given test point. For example, since I completely forgot about the meter's loading effect, I didn't realize it was showing me incorrect gate voltage. I looked in my meter manual (Mastech MS8268), and the meter's input impedance (or resistance, since we're talking DC, right?) is 10M as PRR said. So it is definitely messing up my measurements there.
Still doesn't explain why when I scope the output, the whole bottom half of the wave is chopped off, though!
It is chopped because there is too much gain and the clipping is asymmetrical. I did a quick simulation in LTspice and it is confirmed!!! It is enough to add a fizzy quality to it.
The SHO is superior because you actually decide how much gain you want and also change the bias, so you get a palette of tones and amplitudes.
With the mosfet booster here you get a fixed sound that is divided by a volume pot... Not that great in my opinion... Variable gain is a must for a booster, would it be only to be able to use it as a buffer if needed.
Place the meter set to DCV across R3. The voltage drop should be close to 0.
What are you using to drive this circuit?
if a signal generator, what is the peak to peak signal input swing?
Is the diode in the circuit the correct way?
Quote from: Gus on March 05, 2013, 02:54:42 PM
Place the meter set to DCV across R3. The voltage drop should be close to 0.
What are you using to drive this circuit?
if a signal generator, what is the peak to peak signal input swing?
Is the diode in the circuit the correct way?
Ah - cool. I get it. If I look at the drop across R3 it puts the 10M meter in parallel with R3 to make 5M and doesn't load the gate or act as a voltage divider like it did when I measured voltage gate-to-ground? I'll try that when I get home.
Using a signal generator to drive circuit. I believe I had it set around 1V or 2V p-p, but I tried turning it up and down while scoping the output, and it was still chopping the bottom off the wave. That's why I thought it was a component failure or I was screwing up the biasing somehow.
I double-checked the diode polarity and even removed it temporarily (while being very careful about static), and it did not change the symptoms.
Quote from: jymaze on March 05, 2013, 02:53:22 PM
The SHO is superior because you actually decide how much gain you want and also change the bias, so you get a palette of tones and amplitudes.
I looked up the SHO just now, and it looks even simpler than this build. Definitely going to give that a try next. My vero build was the original MOSFET booster, which had the 5k gain control instead of the volume pot, and I was having the same problems with that (possibly because the meter was giving me loading problems as we discussed...). Looks like the difference in the SHO is that it is a "scratch ok" pot with DC on it which is actually changing the bias of the tranny like you said (instead of being in series w/ a bypass cap). I built a fuzz with that setup and loved it, so I'll probably give that a shot next.
+100 for Gus' suggestion of a 5k pot in series with the 100uF bypass capacitor. That 5k pot will give you much-needed control of the gain.
FWIW the stratoblaster is my favourite booster. Simple, low to mid gain. JFET sweetness.
cheers
Just had a couple minutes to experiment tonight. I put a decade box in place of the 10M resistor and turned it down to 100k as PRR suggested. It immediately fixed the voltage measurement problem - now I'm reading 5.69V on the gate.
And Gus - you were right, measuring voltage across the 10M resistor R3 showed just 100mV or so. Looks like my problem was a measurement problem. :icon_redface:
I'll try some of the other mods as soon as I get a chance, but I'm thinking jymaze is right and I might like the SHO or something better.
Replaced the 10M resistor with 2.2M. Output looks a LOT better on oscope - enough that I wonder if there was a bad solder joint on the 10M resistor, even though they looked ok. If I put in a 1Vp-p sine wave input, this is what comes out:
(http://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=48805&g2_serialNumber=1)
Looks a lot better than before, but still not what I'm going for. I'll try the bias pot mod and see what difference that makes, and if that doesn't do it, looks like I'm building an SHO or a stratoblaster.
Thank you all for the help!!
Put 100 ohms in series with C5! It will tame the excess gain a little.
Will do. That idea made me go read more about bypass caps, too---thanks for the motivation. I knew they increased the output of a stage by reducing/eliminating negative feedback at the cathode/source/emitter, but for some reason I never put it together that that less negative feedback at the source = more input gain within the same stage, and more NFB = lower gain in that stage... cool.
you could try replacing r1 and r2 with 33k each, with a 100k trimpot between them connected to r3/c3. then you could shift the bias point of the mosfet while watching the cro to get the maximum swing from your particular device. I'm surprised this hasn't already been suggested.