Harmonic Percolator (George Giblet's schem) build report

[alex frias]

Hi Alex - what other trannies have you checked?
(of Ge's, I currently only have 2N219 PNP ...and AC141, AC 187, AF239, and a 2SA235 which may be faulty)
Any of these you tested? 2N404 seems hard-to-find in EU. Maybe I havent' looked at the right places. (I suppose that circuit would work "flipped", with a npn ge as well)

Several AC188, AC128, 2SB75, 2SB54 and the 2N404A (also 5 unknown devices, but obviously they can't be used as reference). In this last attempt everything with medium to high hfe, many trannies sounded good, but, don't ask me why, THE SOUND I was looking for only was achieved with 2N404A. Some high hfe trannies exhibit a bit of the octave-down-at-attack effect.

I saw you tested a 2n2222 for the Si NPN. Could be that 2N2219 also works?

Don't know, but possibly...  i liked very much the combo: 2N404A hfe~250 + BSX20 hfe~100 (best octave-down effect), but the smoothest was the same Ge with 2N3565 hfe~120.

I soon will try to do a new "spectacular" video.

[alex frias]

The pics I saw that took me to the 10K Q1's collector resistor value was in another forum, they was posted as Chuck Collins HP-1 clone guts. So I think I may thank to Mr Collins again...

Another stuff I reread was the Barge (Mis?)Concepts texts and schem. I was so tempted to give their NPN/PNP idea a try that I built a 3rd version of the circuit.
Well, I don't know what the hell they did to the PNP/NPN circuit to get it not working, but their (mis?)conception is by far very confused.

I've built the circuit according the Barge schems and the same referred pics.

At one part of the text about their BP-1 device Barge people says: "In the original HP-1, the identity of both transistors is concealed by a combination of a metal cylinder and some potting compound attaching the cylinder to the body of the transistor. What's missing on these two potted and completely unidentified parts is any way of distinguishing between them. In any kind of electronics manufacturing, it's vitally important to be able to tell what part you're dealing with, and to conceal two different devices in an identical way makes it impossible to easily tell which is which. However, it makes sense to not mark them if you don't have to - if they're both the same and there's no need to distinguish between them. As a piece of confirming evidence, the importance of this one cannot be overemphasized, even though it completely escaped us before the electronic truth materialized." This assumption it's not the most clever I've heard. What I understood is that the "Interfax" was never a big industry, so the production line would not be hard pressed to get a very high standard efficiency. And if you are using only 2 different kind of trannies (PNP and NPN), just put them in separate baskets after measuring and dressing to make them unrecognisable, prior to sold them on the PCB. Or even better, separate them also in hfe groups.

What really crosses my mind is the possibility of running out of 2N404A's and finding out an easier-to-find (and perhaps cheaper Si) alternative to substitute them. I finished the soldering process late at nite, so no audio tests are already made. As DougH said, some stuff added, as the input diode to ground and the 1K + 2.2uF in Q1's collector to ground, could be a form of "band-aids" or another tricky effort to accomodate different trannies and hfe values in the same project. But those are just possibilities...

I would love to get comments about it...

[DougH]

And if you are using only 2 different kind of trannies (PNP and NPN), just put them in separate baskets after measuring and dressing to make them unrecognisable, prior to sold them on the PCB.

Or just put their "disguises" on after they are soldered to the board.

[alex frias]

Why not?

I hope to do tests with this new (NEW?) configuration, using NPN/NPN and PNP/NPN transistor-combinations to sse what we can get out of it.

Hard will be to find a Si PNP with different hfe values in my "stock", I think I only have 2N3406...

Anybody knows a complimentary PNP for the 2N3565?

[Gus]

Have not looked for a PNP but the few NPN 2n3568s I tested have a Hfe of about 120

[alex frias]

I think this could be a possible candidate:

[bool]

Now that's pr0n ... no, no ... it's a MILF

[Gus]

just checked 10, 2n3638s  lowest Hfe was 150 most were >250

[alex frias]

I admit my lack of attention to a very important thing: the collector-base resistor, in my third try, the version with the input-diode and the 1K+2.2uF in series from Q1 leg to ground, some more than exoctic glass resistors and unknown wax covered cap and trannies. The cap seems to be poted accidentaly, what could suggest the "disguises" was dressed after soldering... (CSI anybody?) Well, I don't know why I was thinking the collector-base resistor was at the same place the other versions showed, but as I replicate the pictured copper side of the pcb at the photos I perceived the differetial point.  It is actually a much more relevant clue than recognizability of transistors, I think. The resistor goes from base, I presume, to other leg. Now it even makes sense to me thinking about using two NPN's.

Barge pics:
http://bargeconcepts.com/bp/bppcb.jpg

Yesterday I had a very busy day, so a little time was left to the tests I was suposed to do. The 2N3565 remained as the Q2 during the tests. Those quick tests showed the Si PNP's didn't work at all for Q1. The Ge PNP's worked but not as I would mean "properly". The Si NPN's produced some good overdriven sounds, but nothing that special. The values I've used for collector-base for both transistors was 750K, guessing the axial glass resistors were the same value... Some pictures in the Barge site shows the guts of their machine. I will pay more attention to figure out what they used in place of those resistors. Maybe use two trimpots (doh) to aproximate a best-value.

[alex frias]

Well, I spend some time on the weekend trying to do my best with the supposed NPN/NPN version. I've changed the base-collectors resistors to what I understand from Barge  inside picture are 470K's. Just a so-so overdrive...

I then got back to the last PNP/NPN version I built.

My conclusion about this version is that it was the closest aproximation of the real thing sounds.

The component values used:

The caps are 100pF, .047uF, .1uF, .1uF e 47uF
Q2 collectror resistor 91K.
Q2 base-collector resistor 750K.
Q1 collector resistor 10K
Q1 base=collector resistor 220K
Diodes: 1N60
Diode resistor 4K7
Harmonics pot 100K A
Balance/Level pot 50K A
Q1 Ge PNP 2N404A hfe~380 (and a very leaking device)
Q2 Si NPN BSX20 hfe~75

To my taste, the more absurd the hfe and leakage for the Q1, better it is. But it lowers a lot of the overall gain. To keep the gain at a good level, the use of a Q2 with very low gain helps doing so. This balance between the transistors hfe are the key for THE sound, in my opinion it is getting cool above hfe 200 for the Q1.

I think I ended my tests as I'm very satisfied with the final sound. I tested the Doug's last shown schem with the same Q1 hi hfe / Q2 Lo hfe balance and achieved very close results.

Here a small video loosely playing with the prototype:
http://www.youtube.com/watch?v=VOAlUb7CCmg

[DougH]

Alex, that is the smoothest one you've demo'ed yet. Sounds beautiful! And the guitar cleanup is really great too. I hope that's the one that made it in your final build.

I found that octave sound too with higher hfe transistors for Q1. I didn't mention it because I wasn't sure that's what you were talking about. It's inconsistent and only happens with certain notes and fret positions. (I didn't particularly like it because it wasn't consistent so I tried to tune that out of mine.)

In any case, you've really nailed the sound AFAIC. Patience & persistence pays off with this circuit. Hopefully others can reference this thread for some tips on tuning, hfe, etc so they can diy some decent sounding builds.

[alex frias]

I think this is my final try about this one. I now have three PCB's populated with different versions. Both PNP/NPN's are very similar sounding when using close hfe transistor values. The twin NPN didn't work that great so, if I need it will be transformed in another great sounding PNP/NPN, or if somebody appears with some novelty about that version. I think Barge aproach is nothing like I would aprove in my own kind of research. At the technical side of view this circuit is side by side with the Fuzz Face: how far you can go with only 2 trannies and some resistors and caps. OK, the HP-1 uses TWO diodes and FF none, so it can be blamed by use of much more technology...

One thing that's not in the video but I find is a very cool point of the unit is the acoustic feedback capability of it. It's monstruous and at the same time musicaly beautifull.
The octave effect I didn't managed it completely (on Albini's vid it's more on the lower register, in mine it appears in lower-middle notes) , but I do like the inconsistence regarding fret position and guitar volume/harmonics control pot seting. Distortion, in my opinion is a resulting balance between consistency and inconsistency. It's like dirty ying/yan. I even like some cheeseness, but judiciously used in a musical context. I still hear music as some sort of magical noise structures, something tha communicates with us in a more subtle emotional state, but much more effectively than words can do. So the sky is the sonical limit.

Now I have a very different sounding and ready to be explored new pedal. I hope my added 2 cents may be of any help to others looking for the same kind of stuff.

[John Lyons]

It sounds good Alex!
Thak you for posting your tests and results.

john

[Caferacernoc]

A note on the octave down effect of this pedal. The second stage of the percolater/brick is very similar to this electra distortion build. Which is noted for it's octave down behaviour.

http://www.home-wrecker.com/electra.html

And a general note. A difference I see with both circuits(and the Trotsky drive) is that there is no "coupling resistor" in between the output capacitor and the clipping diodes compared to most Dist+ type circuits. I assume the diodes give a variable load to the transistor depending on how much the diodes are conducting signal to ground. Do you guys think that has something to do with how these circuits are so spongy and why they can bloom?

[alex frias]

Well, recently I read a topic here about the Jordan Bosstone and the supplied sound clip showed some octave down effect too (and several other noisy/sound artifacts...).

I think of building a J.Bosstone and try two or three things on it.

Regarding the Percolator: what I liked about it is that it is plenty of  "idiossyncrasies". The singing sustain and musical acoustic feedback are great bonuses. The sound artifacts, as its inconsistent octave down effect (only cranked up or close to it) and how it cleans up with the Harmonics or Guitar Volume control gives a seducing character to it.

My only complaint about the HP-1 is the noise ground, but... you know...

[DDD]

In spite of the present topic consists of six (!) pages now, there are no any attempts to explain why the subject creates so good sound 
So, let's try.
1. HP provides dynamic texture distortion due to the shift of the operating points of the PNP and NPN trannies caused by the "asymmetric" charge\discharge of the 47uF cap generally through the NPN transistor and clipping diodes. As we know Ge trannies are very sensitive to the operating point changes, so they produce much more "dynamic" tone change.
2. HP is not just "non-linear" amplifier, it's a kind of a resonant circuit with a positive feedback. Playing with all of its three caps one can convert it to low-frequency oscillator that can generate 80 to 200 Hz frequency pulses or to one-shot oscillator triggered by input signal. So it can divide the input frequency by two or even by three. That's why we can see the octave-dividing effect playing on several neck zones. This ability highly depends on coupling and electrolytic caps as well as on trannies' hfe.
3. Combination of the two above mentioned features creates very complicated effect on the circuit functioning, including DC and AC positive feedback loops and even kind of "automatic gain control" caused by the operating points shift, e.t.c. That's why the device is SO sensitive to it's parts characteristics.
As a matter of fact it's almost impossible to produce two HPs with the same sound. Also, they are sensitive to the ambient temperature changes.
So, I hope my "two cents" may trigger some more attempts to modify this wonderful device. For example, to change Ge trannies and diodes with Si ones (including Shottky),  use aluminium 47uF cap plus 1uF film instead of tantalum, e.t.c., e.t.c.     

[alex frias]

I think you have done it and helped a lot! Thanks.

I was going for the empyrical side of it and what I saw is very well complemented by your explanation. One thing is clear: it's virtually impossible to do two units sounding exactly the same! I've tried it out but any small difference matters, eventhough it may be subtle...

Could you make any comment about the NPN/NPN version? I really had no success with this version...

[DDD]

Hi alex frias,
i think npn\npn HP version is not as interesting as "classic" pnp\npn one is. Classic HP has much more dynamics in sound texture because operating point shift is applied to both trannies' emitters (pls note that emitter is a "control" terminal of a tranny). In the npn\npn arrangement operating point shift affects only one emitter, hence we can assuume that op. point shift produces less affect on the distortion texture.
By the by it seems to be interesting to change Ge PNP transistor with the modern low-noise Si one and additional resistor in series with its emitter terminal to reduce its "excessive" gain. This measure can lower the overall noise and improve high-frequency roll-off of the input stage\pickup inductance. So we can expect the better (brighter) tone with humbucker pickups (again - lower hum). In this case the input pot ("Harmonics") resistance is not to be 100 kOhms, but maybe 250 or even 500 kOhms. Moreover, it may be omitted at all, because the volume pot can do it's "duty" with no additional noise caused by "Harmonics" pot.
But please note: all the written above is only theoretical assumptions. I didn't check them in reality to my regrets. Moreover, I'm not sure the proposed mods will sound as good as the "Brick" or Ge-Si versions do. It's almost impossible to compete with Doug H, as we know :-( 

[alex frias]

OK, and Mr Hammond is among other things a real gentleman and it seems you are too!

I will try out some of your suggestions. I did some experiments with Si PNP, but at that time I didn't have any clue about the component values and their relation to the resulting sound. But at first I will keep the path I was folloing - PNP Hi hfe and NPN very low hfe.

A funny question: How can I simulate a leaking Ge device with Si trannies? I liked very much the leaking Ge PNP I used.

Regarding hi freq roll off, well, i't can be a good or a bad thing depending on the project...

[DDD]

Hi alex:
I'm sure it's actually impossible to simulate leaky Ge transistor with any components. Leakage is an internal property of the tranny. Moreover, it's combined with hfe and other mysterious Ge mojo :-)... Even if you'll find any theoretical solution, the sound won't be the same.
By the by, I have quite interesting Si project with BC550 (Hi-gain Si npn) and BC556 (Hi-gain Si pnp). As my spice simulator shows, the project have some distortion texture dynamics and frequency properties close to the classic HP.
At the same time I'm sure its sound will never be the same as that of the classic HP. Mojo is mojo :-(
If you have some time to try it, I'm ready to send you its chematic and some explanations. To my regrets I have no possibility to breadboard it in the nearest future. But it seems it worth some try as a differently sounding gadget.