Harmonic Percolator - make one!

[iainpunk]

i build a metal Simplex once, its a nice little fuzz/distortion with heaps of gain, but i put in 47k resistors for R2 and R4.
its indeed quite similar to the HP in topology, but can be made to sound really different.

funny how that website describes the schematic as ''although similar topology as old Fuzz Face.'' which made me chuckle a bit.
they have very little in common, only the fact that its 2 transistors cascaded...

that's also not how my metal simplex sounds, i guess we can hear the electric guitar being acoustic about as loud as the amplifier is in his clip.

on another note, the HP/HIC i have on my breadboard now is really harsh and gated, how do i lower the gating effect without lowering the harshness? the first stage is not gated, and the 2nd stage is quite gated, and there is a small cap in between them to make it harsher and more nasal, if i increase the gain of the first stage, i lose the harsh nasal vibes, how do i get more sustain, without losing the harshness?

cheers, Iain

[Rob Strand]

funny how that website describes the schematic as ''although similar topology as old Fuzz Face.'' which made me chuckle a bit.
they have very little in common, only the fact that its 2 transistors cascaded...

Yeah, you see that a lot.

It's a while since I've played around with the HP.   Using the designators on this schematic,


I don't have a perfect way to tune it but I think the key is the VCE  (voltage between C and E) of each transistor.
For Q1:  want VCE1 0.2 to 0.4V ; higher values of 0.6 to 1V are workable but I think the low range is what people want.
For Q2:  want VCE2 0.9 to 1.5V ; wider range 0.6 to 2.0V is workable

The steps to set-up the circuit:
- Choose part values for R3, R4 either like in the George Giblet circuit or other. 
- wire-up the George Giblet circuit.
- Adjust R2 to set the VCE1 voltage
- Adjust R1 to set the VCE2 voltage
- repeat the last two steps one or two times if the required R1 and R2 values change a lot from the initial values.

The problem with adjusting R3, R4 is you have to make large adjustments to affect VCE1 and VCE2 and even then you need to have R1 and R2 reasonably close to start with.

So one tweak which I haven't played with is adjusting R3 and R4 *together*.   If you keep the sum of R3 and R4 constant you can trade the gains of the Q1 stage and Q2 stages.  For example if we start with R3 = 91k and R4=20k the sum is R3+R4 = 111k.  We can decrease the gain of Q1 by changing R4 to 10k.  In order to keep R3+R4 constant R4 needs to be 111k - 10k = 101k.   The reason for keeping R3+R4 constant is it doesn't affect the VCE1 and VCE2 that have be selected.

In order to set the VCE voltages we might end-up with R1 and R2 values which are much different than the original values.  A low value will decrease the gain of the stage.  A high value will increase the gain of the stage.    Way off values might mean the transistor isn't suitable.   Adjusting R3 and R4 together is a way we can fix the gain to some degree to compensate for way off R1 and R2.

[rankot]

So one tweak which I haven't played with is adjusting R3 and R4 *together*.   If you keep the sum of R3 and R4 constant you can trade the gains of the Q1 stage and Q2 stages.  For example if we start with R3 = 91k and R4=20k the sum is R3+R4 = 111k.  We can decrease the gain of Q1 by changing R4 to 10k.  In order to keep R3+R4 constant R4 needs to be 111k - 10k = 101k.   The reason for keeping R3+R4 constant is it doesn't affect the VCE1 and VCE2 that have be selected.

In order to set the VCE voltages we might end-up with R1 and R2 values which are much different than the original values.  A low value will decrease the gain of the stage.  A high value will increase the gain of the stage.    Way off values might mean the transistor isn't suitable.   Adjusting R3 and R4 together is a way we can fix the gain to some degree to compensate for way off R1 and R2.

In my last build, I did exactly that - mounted 91k collector resistors and 500k pots in parallel with them, then adjusted pots by ear - and to me the best position was close to the cutoff for both transistors, just tweaked a little to avoid some nasty decay oscillations on low notes. Then desolderet pots, measured their resistance, calculated parallel resistance of that with 91k and that was it.

[Rob Strand]

In my last build, I did exactly that - mounted 91k collector resistors and 500k pots in parallel with them, then adjusted pots by ear - and to me the best position was close to the cutoff for both transistors, just tweaked a little to avoid some nasty decay oscillations on low notes. Then desolderet pots, measured their resistance, calculated parallel resistance of that with 91k and that was it.

That's method is fine too.  Decreasing the collector resistors (R3, R4) makes the VCEs go up.   If you VCEs are too high you need to decrease R1 and/or R2.

Digit only adjusted R4 which is also fine but you can end-up with R4 being too low.   I think Digits has R4 around 12k, which is in he ball park of 20k.    However it's possible to have many set-ups which give the same VCE provided the sum R3 + R4 is constant.  For Digits's case R3 + R4 = 12k + 91k    = 103k.   So if we kept R4 at the standard 20k would could set R3 = 103k - 20k = 83k and it would produce the same VCE bias points for Q1 and Q2.    I haven't done the experiment to see how the sound changes when you  keep R3+R4 constant but allow the R3, R4 values to change.

[rankot]

I believe that hfe also has impact on this, and those are the final resistor values for my two build, with different transistors:

Q1 MP20, hfe 52, Vc=4.67V, R4=56k
Q2 2N6517, hfe 140, Vc=5.84V, R3=39k

Q1 AC152, hfe=41, Vc=0.66V, R4=5k1
Q2 2N2369, hfe=55, Vc=5.27V, R3=30k

Both found using "pots and ear" approach.

[Rob Strand]

Q1 AC152, hfe=41, Vc=0.66V, R4=5k1
Q2 2N2369, hfe=55, Vc=5.27V, R3=30k

That one is quite interesting.   It would be good if you had the emitter voltage so I could check the VCE voltages of Q1 and Q2.

If magic sound comes from a low VCE on Q1, then in this case Q1 would be expected to be quite leaky and the low R3+R4 sum value is a way of counteracting that.     If the VCE of Q1 is say 0.3V, then the emitters will be at VE = 0.66 + 0.3 = 0.96V.    So for Q2 the VCE would be 5.27 - 0.96 = 4.31V, which is much higher than normal.   If we assume VCE of Q1 is 1V, higher than normal, that puts VE=1.66V but VCE2 is still quite high at 3.61V.    Whatever way we look at it VCE2 is quite high so maybe it's not that important.   

[rankot]

Unfortunately, I didn't record the leakage and now I'm too lazy to desolder trannies and measure again. However, emitter voltage sits at 0.94V on this one with AC152 and 2N2369 (I can't measure another unit, since I gifted it yesterday to my late friend's son).

Probably hfe has some role in this, because 2N2369 has hfe of only 55, while gifted unit use 2N6517 with hfe of 140.

[Rob Strand]

Unfortunately, I didn't record the leakage and now I'm too lazy to desolder trannies and measure again. However, emitter voltage sits at 0.94V on this one with AC152 and 2N2369 (I can't measure another unit, since I gifted it yesterday to my late friend's son).

Probably hfe has some role in this, because 2N2369 has hfe of only 55, while gifted unit use 2N6517 with hfe of 140.

Thanks.  We can see VCE for Q1 is 0.94V-0.66V = 0.28V which fits the theory of that VCE1 needs to be quite low.  VCE2 on the other hand is 5.27- 0.94 = 4.3V,  way over the normal.   So maybe VCE2 is of secondary importance.

The hfe definitely comes into it.  Higher gain transistors will give lower VCE voltages.    Which is clearly seen on your two NPN transistors.   The pnp is tricky because high leakage will cause VCE to drop and so you might not see VCE matching up with the hFE.

[rankot]

And I think bigger hfe is better for Q2, because this one I have with lower hfe still has some low frequency oscillations (reduced when I put 2u2 capacitor instead of 100n at C3), while the one I gifted has none. Maybe I should try different transistors on my unit.

[Rob Strand]

And I think bigger hfe is better for Q2, because this one I have with lower hfe still has some low frequency oscillations (reduced when I put 2u2 capacitor instead of 100n at C3), while the one I gifted has none. Maybe I should try different transistors on my unit.

I don't have any good answers for the DC problems.   Some other pedals put resistors between the output of one stage and the input of the next and others use small emitter resistors.   The other way is playing with the cap values like you did.  Hard to know in advance if they change the sound for better or worse.

[iainpunk]

a simple 330 ohm to 1k emitter resistor on the Q2 side made all my oscillations disappear, under the 330 oh, they were still there and above 1k the amount of gain started dropping off to much, but mine isn't a ''true'' harmonic percolator, but my Instant Coffee version, i posted earlier.

i came across some issues with gating, i have Q2 in reverse Beta, with the right diodes*, this gives a beautifully broken tone, but it gates out the circuit, and if there is no current through Q2, the current through Q1 starts to sag until the middle cap is empty, which in turn made the gating worse.
my solution is a 10k from 9v to the middle cap to keep current through Q1 going, this made it way less gated, and retained the beautifully broken sounds. the cap voltage is 7,62v.

TLDR: if you have problems with gating, a resistor parallel to one of the stages might relieve the gated-ness

cheers, Iain

*i find that diodes have way more impact in the instant coffee than they have on the Bazz Fuss. i have a BYX10 on Q1 and an amber LED on Q2.

[iainpunk]

Instant Coffee Update:

there are 3 things i considder virtues of a fuzz, and i want them all, which is harder than you'd think, most of the time its a choose 2 situation:
- low end
- sustain
- raspyness

after a day of experimenting, i have come to a conclusion; having Q1 as high a gain as possible is better for everything, both bass, sustain and raspyness, the gain of Q2 determines its place on a kind of spectrum, from thicc and bass heavy to raspy with less bass, there is no magic combo with much bass and raspy harmonics. leakage is a bad thing here, it loses bass and gets less raspy, it gets just thin and soft.
after an hour or two, i selected a ASY29 as Q2, in reverse Beta of course, it gave me the best balance.
here is the new full schematic:

Q1 : BC519
Q2 : ASY29
D1 : BYX10
D2 : Amber LED

cheers, Iain

[Phend]

Hum, so to sum it all up,
NO two Percolators are the same.
GE is temp sensitive,  so even in one it might be different day to day.
Plus the permutations of values exceeding 10^4, sounds like a great challenge.
Next build will be the elusive HP !

[Mark Hammer]

Will you get something EXACTLY like what anyone here got?  Not likely.  Will you get something pleasing and musical?  Very likely.

[Phend]

Thanks Mark, can't get better than that. !

[Rob Strand]

Hum, so to sum it all up,
NO two Percolators are the same.
GE is temp sensitive,  so even in one it might be different day to day.
Plus the permutations of values exceeding 10^4, sounds like a great challenge.
Next build will be the elusive HP !

I suspect there is a formula but no one has worked it out yet.    A few points were made in this thread.

Some years back I posted a circuit from one of the magazines from back around 1980 and discussed some of the issues with getting second harmonics from a transistor stage.  IIRC it was posted in the Lounge section.

There's also this,
https://usermanual.wiki/Document/FullWaveSignalRectifierwithOneTransistor.9622082.pdf


Maybe this one, it's more Bazz-Fuzz but it makes some good comments,
PE Oct 1978 - Simple Fuzz, p1082

[Phend]

i think it's soft clipping the signal with the transistors, and using some kinda SCARY PHASEING MAGIC i totally DON'T UNDERSTAND to make it do all this!

seriously, i think something about the circuit seems to  cancel dissonant harmonics.

i can get octaves out of it now...easily. up. down. both. it's insane.  no idea how/why it works. also discovered adding power supply filtering seems to affect it adversely!!

the latest is the best sounding one... if you make that 91k resistor a 47k from the collector of the pnp to the top lug of a 50k pot, and tie the wiper of that to a 47r resistor, you will get ALL the shit in albini's video. seriously.

it's bizzarre. you can make feedback come out in harmonic chords some times.

it's crazy man, like really crazy how ccrazy this crazy thing really is..

once my girl gets up, i will try and shoot some video. that 91k is the main key to percolation. the other is the 1m feedback resistor on the npn.

Question, what does the 46R resistor do ?
The way I understand this is,
Replace 91k with 47K, attach that to lug 1 of 50K pot then a 47R to lug 2 (wiper) then to gnd. Lug 3 is no connection.
But 47R ? What is it for ?

[antonis]

And exactly where is that 46R (or 47R) notorious resistor..??

[Lino22]

Is this pedal explained somewhere? It looks pretty unique. Does it just create a massive gain that is clipped by Ge diodes? Is the output really quiet?

[Phend]

And exactly where is that 46R (or 47R) notorious resistor..??

Quote from pinkjimiphoton
"the latest is the best sounding one... if you make that 91k resistor a 47k from the collector of the pnp to the top lug of a 50k pot, and tie the wiper of that to a 47r resistor, you will get ALL the shit in albini's video. seriously."
It isn't shown on the reference drawing from digi2t