Harmonic percolator (non-mojo builds)

[bmsiddall]

Greetings from Australia.

I'm one of many here bitten by the HP bug. I'm currently putting together build #5 (i think) after admitting an addiction  .  I'm using the Giblet schematic with mods:
1uf instead of 100nf coupling caps
Fixed 100k resistor instead of harmonics pot (prefer using the guitar vol control).


Diodes tried: D310 (very good), 1N695 from small bear (excellent), D9K (excellent ++). I have some D9J diodes coming which seem to have an even softer knee than the D9K- will check these out too. https://www.diystompboxes.com/smfforum/index.php?topic=118007.0

I've tried many transistor types and hfe.  For the PNP,:
2SA102 (hfe 35, didn't measure leakage) very good.
2N1305/7/9 (hfe from 60-250) were mostly excellent.  Low-high leakage worked, although bizarrely, 180 + 300uA was noticably less fuzzy than lower hfe/lower leakage AND 250 hfe/1.3mA leakage.  Yes, that was 1.3mA.
2N404 were very good-excellent.  generally I like hfe for these around 80-100 rather than lower.
GT404 (hfe 60-120, 100-200 uA leakage).  Nope.  Just makes it into a tame overdrive really.
GT109 (hfe 50-200+) - these are the really tiny Russian gnats!  Uniformly excellent.  i had nearly decided on a 2N1307 (80 hfe, 50uA) until i started loading these.  Just more of everything!  Haven't decided which one but i currently have one socketed with hfe 80, 20uA.

For NPN, have tried 2N3565 (hfe 205), BC208 (hfe 130), BC407B (hfe 500), 2N697 (hfe 50) plus a few other randoms- they all sound fine.  This is a pretty forgiving circuit!

One persistent irritant (and probably why i keep building them) is the octave down is hinted at rather than obvious.  I have owned the Latent Lemon version and its octave down sticks out like dog testicles.  Almost like a separate string was being hit at the same time.  Wish i could find a way of getting there- have tried 100k trimpots for R3/R4 without luck.  Suggestions welcome!

[Mark Hammer]

Ironically, I'm just finishing boxing one up for a guy.  He put a matchbox of old Ge transistors into my hand and asked if I could do something with them for him.  I measured their respective hfe and after musing for a bit thought that an HP might be just the thing for him (he plays with an alt-country band).

I used a 2N3565 and one of the germaniums that is semi-rectangular (no visible part number).  I drew from the schematic that is near identical to the one you show, but also included a germanium diode to ground in parallel with C1.  When I brought him an unboxed-but-functional board to try out, he liked it very much but found it a bit shrill, so I stuck a 1K8 resistor in series with the volume pot, and wired up a 3-position toggle to provide two different treble rolloffs, or no cut.  Sounds great.

I should note that there were 5 or 6 2N1309s in the box, so I may try a build with one of those.  I have not noticed any octave down, hinted-at or otherwise.

[bmsiddall]

Thanks Mark.  Interesting- I've not thought any perc build I've done has ever been shrill at max harmonics- quite the opposite.  I'm guessing you mean with harmonics or guitar volume rolled back?

I'd be interested in hearing your thoughts on the 2N1309.  FYI, i think the higher hfe (generally >180 i think) sound better with lower hfe npn (90-130 or so), but there has been so much swapping I'd need to take another swing at this aspect.

I did forget another mod.  1nF capacitor across 20k Q1 collector resistor- I saw this somewhere or another.  i'm not sure what it does or if it adds anything to the tone, so might cut it from the final build.

Regarding the input diode, what would this do tonewise? 

[duck_arse]

a few weeks ago, I broke my coffee pot, a moka pot type. so I've been building a percolator to put innit.

I upped R1 to 1M5, and put a 6k8 and 10k trimmer at R4. I also tried all my available transistors in, kept coming back to the 2SB156, either 76uA/88 hFE or 102uA/42 hFE, and a (pulled) 2SC1000 BL. or a 2SC734 Y. I erased my extra transistor notes this very evening for room on that page, so the only best russian I can remember is the "P416b" which was very very good. very good. very good indeed, if'n you've got one.

but the Si I'm building with is a PN3694 which measured 100 hFE's on the dmm.

I started out using digi2t's notes, went looking for the octave down. I'm not the best person to be tuning these, technically or earally, but I'm not so sure it is octave down and not the circuit motorboating/oscilating at "a" freq it likes. with 680nF out cap and 220nF Q1C to Q2B, it was there - but the key seems to be the tantalum cap. if you use 47uF Al, it can get all blurty and loose, if you add a 47uF tant it tightens up somewhat, and if you use just the tant, it can blurt, but tightly, or more tightly than w/ the al electro.

again, but - shifting the caps to 470nF out and 330nF up/down, the blurts have mostly left. so I'm intent on switching a tant in parallel w/ a fixed Al electro for tighter and woolier.

if you reduce the value of the E-E cap, and strike a string, you will set a very ugly oscillator ringing, which is why I think not a real octave down.

[edit :] oh, and as pointed out by Gus, all the stuff following/loading the output cap is important in tuning the blurtingness. he didn't say it like that, though.

[Mark Hammer]

Regarding the input diode, what would this do tonewise?

I imagine what it does would depend on the input signal level.  As a germanium diode, it has a low clipping threshold.  But since it is situated before any sort of gain stage, whatever clipping it imposes would be relegated to the initial peak, and even there only on one half-cycle.  So I imagine it would be a nuanced effect, but when it occurs, would make the tone seem to change over the life of the sustained note.  I do know that on this one, there is a pleasing "gargle" to power chords at higher sensitivity or when sensitivity is dimed but the guitar is rolled back a little.

[bmsiddall]

I tried the input diode this morning. Subtle difference, maybe a bit more cultured?  Not better or worse to my ear, so might leave it as is.

I also went through my bag of GT109 and tried 200 hfe/300uA, 333/50, 820/80 (the bag ranges from hfe 9-820, amazing!) and all sounded good.  Further blind swapping lead me to hfe 52/20uA as it seemed to retain the nasty, but gave a little more sting to unwound strings.

Coincidentally i have some p416b incoming soon.  Will report in unless I succumb to the temptation to box up with the gt109- it's sounding great right now 

[SpencerPedals]

I was just looking at a vero for the percolator yesterday and found some folks talking about a 1n cap to ground in parallel with the diodes here:  http://tagboardeffects.blogspot.com/2010/07/harmonic-percolator.html

My breadboard is tied up at the moment so I can't percolate as I would like, but I would see if that does anything for you since it's a two-second mod with a very common value cap.

[Mark Hammer]

Personally, I like to run my my amp bright.  So I tend to want to add in mods that will tame the high end on fuzzes and overdrives, so that I can hit the stompswitch and not feel like needles are piercing my eyes.  The added cap sounds like it is in keeping with that strategy. 

[DDD]

The HP mojo depends on the type of C4 capacitor.
Aluminium C4 supports mojo, Tantalum C4 removes it.
The reason is that Alu electros have high ESR, and HP transforms to start-stop multivibrator and frequency divider (under some conditions).
Tantalum caps have negligible ESR, hence positive feedback due to their low "resistance" is weak, and HP works as usual amplifier.
Try low-Ohm resistor in series with the C4, and you'll get some interesting and weird results.

[bmsiddall]

Couldn't help myself- curiosity got the better of me. Had a good win at work and celebrated by ordering another Latent Lemon HP with fixed harmonics (has balance and sputter/bias pots only). Previously had one and sold it after I got serious about DIY.
Yesterday I tested it side by side with my GT109 build. What a humbling experience- it smashed mine out of the ground. More gain, fantastic HP grind, and octave down so noticeable it's like playing a downtuned 12 string. Back to the drawing board I think. Shit 😕

[bmsiddall]

Got a chance to revisit the circuit and built another one using the Giblet consensus schematic (20 K,220 K, 750 K, 91K resistors).  Nailed it this time! 

I went through a whole mess of PNP for Q1.  2N404, 2SB171, GT402 (sounded pretty good this time), ASY25, ASY26, GC117, MP25, MP16, GT403, P416. 

Of all the transistors tried for Q1, ASY 26 and GT403 were both terrible- burbly, farty and misbiased.  HFe were in the 40 ballpark but ASY 26 leakage was very low, GT403 leakage very high (300-1000 uA).  Strangely the 2N404 were not much better, which makes me think I should stop trying to buy these things on eBay .  Of the others, ASY25 and 2SB171 generally excellent (provided gain/leakage figures are appropriate), while most of the Russians tried were smoother, with less growl- pleasing but I wanted more of the nasty.  I spent a lot of time tuning it using headphones and an amplifier modeller, and don't think I ever noticed before that the higher the Q1 HFe (>80 or so) the less fuzz and lower volume you get.  There are some interesting interactions going on between the 1st and 2nd stages of this thing.  If leakage is >100, I started to notice the subtle octave down ghost notes in several places on the fretboard.

The furious transistor swapping got me a Percolator with the right growl (ASY25/3 with <60Hfe, 150uA leakage) and the ghost notes were appearing on some frets on the higher/unwound strings.  Swapping q2 gave subtle differences but did influence if and where the ghost note appeared.  After trialling Hfe between 80-280 (Hfe of 29 was pretty anaemic here!) )- I settled for a 2sc536 (Hfe 179), which gave the most distinct ghost notes (around fret 5 B string and frets 12-14 G and B). The tweak that got me there was pulling one end of the q2 750k resistor and soldering a 5M trimpot in series. Increasing the resistance kept lowering which frets/strings the octave down appeared. Now have >3M here and there's a great grunty growl on the wound E down to the 3rd fret.  It's fantastic!

The only tweaking to the circuit was the 750k resistor, but I guess I'd have more success with other transistors if I had tweaked the Q1 resistors.  Might try that next time.

[glops]

Got a chance to revisit the circuit and built another one using the Giblet consensus schematic (20 K,220 K, 750 K, 91K resistors).  Nailed it this time! 

I went through a whole mess of PNP for Q1.  2N404, 2SB171, GT402 (sounded pretty good this time), ASY25, ASY26, GC117, MP25, MP16, GT403, P416. 

Of all the transistors tried for Q1, ASY 26 and GT403 were both terrible- burbly, farty and misbiased.  HFe were in the 40 ballpark but ASY 26 leakage was very low, GT403 leakage very high (300-1000 uA).  Strangely the 2N404 were not much better, which makes me think I should stop trying to buy these things on eBay .  Of the others, ASY25 and 2SB171 generally excellent (provided gain/leakage figures are appropriate), while most of the Russians tried were smoother, with less growl- pleasing but I wanted more of the nasty.  I spent a lot of time tuning it using headphones and an amplifier modeller, and don't think I ever noticed before that the higher the Q1 HFe (>80 or so) the less fuzz and lower volume you get.  There are some interesting interactions going on between the 1st and 2nd stages of this thing.  If leakage is >100, I started to notice the subtle octave down ghost notes in several places on the fretboard.

The furious transistor swapping got me a Percolator with the right growl (ASY25/3 with <60Hfe, 150uA leakage) and the ghost notes were appearing on some frets on the higher/unwound strings.  Swapping q2 gave subtle differences but did influence if and where the ghost note appeared.  After trialling Hfe between 80-280 (Hfe of 29 was pretty anaemic here!) )- I settled for a 2sc536 (Hfe 179), which gave the most distinct ghost notes (around fret 5 B string and frets 12-14 G and B). The tweak that got me there was pulling one end of the q2 750k resistor and soldering a 5M trimpot in series. Increasing the resistance kept lowering which frets/strings the octave down appeared. Now have >3M here and there's a great grunty growl on the wound E down to the 3rd fret.  It's fantastic!

The only tweaking to the circuit was the 750k resistor, but I guess I'd have more success with other transistors if I had tweaked the Q1 resistors.  Might try that next time.

I've also tried a 2M trim for the 750K and have built several of these with a 25K trim for the 10K/R4. I always wanted the octave down too but then it was too frustrating having it exist constantly. I remember being able to tune in the octave down with the trim in the 750K position but then think that didn't work when I tried again at a different time. For me, sometimes Percs sound good and sometimes they don't and for me I think I notice this the most when there is a change of weather as I always socket the GE and they seem to just change in sound with the weather in this circuit.

[bmsiddall]

I hear ya.  I've got a 2nd perc that I've just tuned with an ASY25 without the 5M trimpot.  Sounded incredible yesterday, but this morning not so much.

For me, more than any other fuzz I've built, how i react to the HP is mood dependent

[duck_arse]

you blokes should look to that tantalum cap on the emitters. try with a standard electro for "octave down" and switch in an additional tantalum to kill the oct. it might work.

[bmsiddall]

I might just try that.

Btw, ordered one of these and some slider pots.  Anyone have suggestions for cuttings slots? I'm thinking of going to a lasercutting firm, but if the job is too small I might have to break out the Dremel (and the cold sweats!).