Input cap toggle on Jack's MOSFET Boost... How can I do this an not have a pop?

[R O Tiree]

Right... I've spent all day messing around with this and here goes:

1. Adding all those extra components without paying attention to the input impedance, the relative impedance of the input cap and its pull-down resistor at the frequencies of interest, etc, changed the character and behaviour of the MOSFET Booster beyond recognition. What I ended up doing was to turn it into an insane Treble Booster. Oops. I went back to stock and had a re-read of Jack's blog entry and the AMZ page and found that it was designed to have a flat-line amplitude response across the entire guitar frequency band. So, having put the sim schem back to stock, that's exactly what I had.

2. I then copied fuzzo's schem of the Catalinbread into the sim. Biased it at 4.5V and observed the same behaviour (flat-line amplitude response) right up to the point where I cranked it right up to full gain. Hmmm... a different shape to the distorted waveform. I'd given them both a 100k Level pot, so they were both driving the same load.

Here's the AMZ/Orman MOSFET Booster @ 82Hz, full gain:



You can see the clipping is happening on the bottom of the waveform and the output cap/level pot alter the shape just a little.

OK, here's the CatalinBread:



Clipping is occuring at both top and bottom. This will, obviously, sound different.

OK, curiosity got me, and I decided to try to make the Orman's waveform look more like the Cat...



OK, that worked, by changing the bias to 4.5V and altering the output cap to 220nF. Still wasn't quite right, so I removed the little 47pF cap and voila!

Curiosity bit me again, and I thought it might be cool to turn the Cat into an Orman...



Biased the Cat to 5.3V or so, gave it a 100nF output cap and added the 47pF. Voila^2 !!

So, the sound would appear to be down to the bias voltage chosen (somewhere between 4.5V and 5.5V), the size of the output cap and a piddly little 47pF between gate and Gnd.

[Mark Hammer]

Joe Gagan addressed the issue of attempting to allow more bass into a circuit via the input cap by simply using a parallel cap in series with a variable resistance.  While not exactly the same as if there was no parallel cap, using a standard-value pot, like 500k or 1M, one can fade in the extra bass via the larger cap.

So while I've got a couple of guys on the line who understand the finer points, in what ways is a variably impeded, as opposed to switched, cap better or worse than the assorted switching schemes being tried out here?

[R O Tiree]

Given the input impedance of 10M (Reference: AMZ MOSFET Booster page, third paragraph), changing the input cap from 1nF to 100µF makes a whole 2% difference in the signal passed on to the gate! (Calculated using |Ztotal| = SQRT ( Zr^2 + Zc^2 ) and observed in the simulator.)

As Jack says on that page, "NOTE: The frequency response of the AMZ Mosfet Booster is flat and extends down low enough for both bass and guitar use. There is no need to mod the design to add more bass; it will not make any audible change."

There's no point, in this particular circuit. As to variably impeded as opposed to switched in other circuits, I'll have a bash with a BJT circuit tomorrow (Si Fuzz Face?) and see. That would probably have a lower input impedance?

[Mark Hammer]

Not that I want to heap any more on your plate than you've already got, but one circuit where changing the input cap makes a very interesting difference in the qualitative properties of the resulting tone is the Jordan Bosstone.  The stock unit has a 100ok attenuator pot on the input followed by a .022uf cap.  Change that cap to a higher value, like .33uf and you start getting weird sub-octaves and unusual ghost tones.

[fuzzo]

i don't know if the Catalinbread uses a trimer to set the bias. I did in mine 'cause I didn't have 62K R I though it will be easier to have the bais with a trimmer.

Anyway , Why not changing the 100uF cap to lower value to have a so-called "treble booster" like Mark said in this thread instead of using a toogle input cap switch ?

I think I'll try that, like this way I could use the Switch on my small stone box. I'll put a 1uF or 2u2 with the 100uF actived via the switch.

[R O Tiree]

Not that I want to heap any more on your plate than you've already got, but one circuit where changing the input cap makes a very interesting difference in the qualitative properties of the resulting tone is the Jordan Bosstone.  The stock unit has a 100ok attenuator pot on the input followed by a .022uf cap.  Change that cap to a higher value, like .33uf and you start getting weird sub-octaves and unusual ghost tones.

Well, that was a fun day...

1. First point, it makes no difference whether you use a large (470k) pot to "bleed" another cap in or switch it in the Jordan Bosstone - the pot needs to be 470k otherwise you get increased signal getting through. Phase angle difference either side of the input cap network is not affected by this pot, bizarrely. What does affect it is the input impedance of the transistor. I did some scratchings on the back of a cigarette packet and, with about 38nA or so going through Q1's emitter, that means the Shockley resistance times hfe (I guessed at 500) in parallel with the 150k from base to ground of Q1 gives an impedance of around 100k. This gives a phase-lag of around 60 degrees or so on the low E string, improving at higher frequencies. (If it was going into an opamp, then the phase-lag is about 2 degrees max at low freqs and almost zero the higher you go).

So, to answer your question, whether you switch or "bleed" with a pot, the results would appear to be the same, with no phase-angle changes due to the "bleed" pot. I found that (in the sim) switching/bleeding to a 330nF cap at the input just passed all frequencies of interest and made a fairly boring square-ish wave with about an 80% duty cycle with single frequencies. Much more interesting seemed to be dropping it to 10nF instead. However, when I hooked up a couple more signal sources and "played" it an open E5, I got all sorts of interesting outputs at all sorts of values from 10nF up to 330nF. As you said, octave-down and other seemingly unrelated tones as well, not to mention the harmonics generated when the diodes clip... Not for the fanit-hearted, then.

Now, OT, perhaps, but this pedal is full of weirdness. Whether it's "musical" weirdness or not is in the ears of the listener... I thought I'd share.

2. When I first sparked it up, the output waveforms were quite bewildering... huge oscillations and much weirdness in the output waveform compared with the voltage at Q2's emitter:



Every time the trace at Q2's emitter (blue trace) "kinks", dV/dt in C4 is changed, affecting the current flowing. Because this cap can charge/discharge very rapidly, even the slightest kink causes radical changes in the output waveform (red trace). Once the diodes start clipping, then things get very spiky indeed. Changing C4 to a 47nF smoothed some of the rough edges off, though.

3. I think the kinks in the waveform at Q2's emitter are due to having it connected to Q1's base, and I think C2 has enough charge to re-bias everything when both transistors switch hard off but I'm going to think about it for a day or so. It seems to be a sort of elastic-band push-me-relax-you thing going on.

[dap9]

Remember that the total capacitance of caps in series is calculated (mathematically) in exactly the same way as resistors in parallel and vice-versa. So the total capacitance of C1 and C6 when SW1 is open is:

C = 1(1/C1 +1/C6) = 0.909090909...nF ~ 0.91nF

When SW1 is closed:

C = C6 = 10nF

There are various configurations in the schems I've posted, from parallel to series to swapping - so, there are several ways of doing it, some better than others. This one makes sure that the junction of C1/C6 is always biased at 0V, so popping just cannot occur. Whatever happens at the right-hand end of C6 is irrelevant. DC-blocking by C6 and the 1M resistor to Gnd at the junction of C1/C6 ensures this.

In previous schems that I've posted, we've been swapping caps or putting them in parallel, with variously successful results... and a series one which was disastrous! Seems to me that this one is the most effective, based on what I've found out from the simulation program.

It's been a voyage of discovery, hasn't it?

I should take some ritalin...  I totally skipped over the first part about how caps in parallel are calculated...  I just assumed they're like resistors.  I've since educated myself and get it.  But what I can't figure out (if it's even doable) - how do I go about sticking to the stock setting of .001uf and then having the other cap be a .047uf?  My initial question got answered earlier on in the thread, but I'd like to try this other idea of having two caps parallel just to improve my knowledge.  The sad thing is, IIRC, this is some basic math where you have two of the values you need, and just need to figure out the third value.  Sadly, I've forgotten how to do it.  Thanks to anyone who takes a shot.

[R O Tiree]

The point is, with this particular circuit, that Jack Orman, the designer, figured out the correct values (on a scruffy piece of paper - check out his blog) that would give an absolutely flat-line response as you go up in frequency. This means that, from 41Hz (bass guitar's low E) up to as high as you can get on an electric guitar, the signal gain is constant within a couple of percent. It also happens that it doesn't matter what value you put in front of the circuit, you won't get any more bass gain at all. As I said in an earlier post, you could stick a 100µF cap in there and there'd be absolutely no difference at all.

OK, with that out of the way, go back to my very first reply and you'll see a simple way to go from 1 -> 47nF. If you really want to go from 1nF to 48nF (1 and 47 in parallel) then you'll need to just switch in and out the 47nF, but make sure you put a LARGE pull-down resistor to ground in front of the 47nF, otherwise it's pop-city again.

[dap9]

The point is, with this particular circuit, that Jack Orman, the designer, figured out the correct values (on a scruffy piece of paper - check out his blog) that would give an absolutely flat-line response as you go up in frequency. This means that, from 41Hz (bass guitar's low E) up to as high as you can get on an electric guitar, the signal gain is constant within a couple of percent. It also happens that it doesn't matter what value you put in front of the circuit, you won't get any more bass gain at all. As I said in an earlier post, you could stick a 100µF cap in there and there'd be absolutely no difference at all.

OK, with that out of the way, go back to my very first reply and you'll see a simple way to go from 1 -> 47nF. If you really want to go from 1nF to 48nF (1 and 47 in parallel) then you'll need to just switch in and out the 47nF, but make sure you put a LARGE pull-down resistor to ground in front of the 47nF, otherwise it's pop-city again.

Ok, thanks, man, I'll go back and re-read everything.  This has been (and continues to be) a great learning experience.  One thing I'm still trying to wrap my head around is the below diagram.  I haven't had a chance to breadboard it, but the way I see it, the 1nf cap is never out of the circuit.  In the position of the picture, doesn't the signal still go through the cap?  Or does it "go around" it?  Thanks again for all the info you've provided!

[R O Tiree]

Yes - when the switch is "closed", the signal just takes the easiest path and bypasses the 1nF cap completely.

Bear in mind, when you bread-board it, what I said earlier in the post at the top of this page, about relative impedances and how I'd messed the circuit up and turned it into an insane treble booster...

The pull-down resistor (R8) doesn't need to be huge (100k?) as long as it's ahead of the cap. Do bear in mind, though, that with another circuit where changing the input cap does make a difference, you must pay attention to the relative impedances of the pull-down resistor and the cap at the frequencies of interest and this diagram has a drastic effect on the amount of signal passed. A better way of doing this would be to gave a pull-down cap at the front and then use a 470k pot to bleed the other cap in parallel.

The source bypass cap in the diagram above is too small. As I said above, it takes ages for the sim to "fill up" all the caps and reach a steady state, so I made that cap smaller. Mind you, that doesn't seem to make much difference either - you need 100µF to get an absolutely flat-line response, but 47µF will get you about 95% at very low freqs and 10µF will get you about 85%. Any more than 100µF just doesn't do anything extra for you.

The reason it turned into a treble booster was that the 1M resistor I had behind the 1nF cap in this switching arrangement just wasn't anywhere large enough and that messed up the impedance that low frequencies "saw" and they were allowed to bleed to ground.