Harmonic Percolator (George Giblet's schem) build report

[JOHNO]

The thread called "Harmonic Percolator vs barge concepts bp1" theres a post the by analog man? i think. Its about the 3rd of 4th post in that thread.

[slacker]

Cheers

[Gus]

Good feedback thanks for building the two NPN idea.

Lets think about this and the octave part both stages are inverting amps and can have different gains so they will try to wiggle the 47uf cap node and the waveforms,  When I have time I will look at this with a scope.  Maybe try smaller/bigger cap values allow more or less wiggle.

[JOHNO]

Im not sure if you were adressing my post gus but mine is the pnp npn version. the transistor is 2n217s. I wrote incorrectly im my other post. Its from a radio aparenltly built in 1964.

[DougH]

albini is defenantly getting an octave down thing happening to my ear when he is playing around the 5th fret.

I got more of that deep grunty sound on my wound strings by re-voicing it according to George's schematic.

Here's the vid with Albini demo-ing the Percolator: http://www.gearwire.com/bp1-albini-comparison.html

[alex frias]

i really appreciate the Gus aproach, as understanding the project principles could let us even investing on an evolution side of the thing.

Well, I don't know about someone else, but I found fascinating the sound achieved by mr Albini on that video. At first, what amazed me was the fact he has a so poor technique and the resulting sound remained interesting. With my new "components setting" percolator circuit I realised the indisciplined Albini's playing is part of the magic. My new combination of elements, in my own opinion, caught some of the magic I was looking for since I've heard it. Even when I play consistently, THE SOUND is there, but if I let the notes ringing, as a deliberately resonant playing, you can really find the magic.

The component values was close to Geroge's schematic with few changes: the output cap is 1uF, not 0.1uF. 100K for both colectors resistors, not 91k.  This configuration really gave me the closer sound to the Albini's version I've got . I don't even know if the Albini's Percolator is an example of how an original one is intended to sound.  But tha sound seduced me. It is far from the splat sound i got from the other versions with high hfe trannies, when the Harmonics, or guitar Volume pots are at Max. The sound is now piercing, compressed, but not that squeezed to death sound i've got.. I tested some 2N404A and 2N3565 trannies (and also several other types). but the best I found was an 2N404A hfe=30 and an 2N3565 hfe=140 combination. I will do the measures tonight, so I hope to post them tomorrow.

I've even tried the input-to-ground diode and the 1k resistor + 2u2 F cap in series from Q1 to ground. The diode is not so interesting in my last configuration. More about it later...
DougH said the RC combo refined the sound. I can't agree more. But it just happen to me when I used other "wrong" high hfe trannies. The diode also helped to reduce the splatyness in the sound. but when i got the magic sound with the "right" trannies, both elements was not needed, they even took the sound to other planet, in my opinion.

[DougH]

I'd love to see how your voltages came out, Alex.

I tried an NPN with ~90hfe but it did not sound good. But I had another wiring problem (see my "xover distortion" tweak earlier) so I may try it again, or at least see if I have an NPN that is lower than 200.

I hear what you are saying about "deliberately resonant playing". If you pick softly and let the notes "breathe", you start to see what the fun is about.

[alex frias]

Well, I did the measure on the working beast. But I really don't know if you would like the results...

I've remeasured the trannies too. I've found two Si NPN's I liked most:
The infamous 2N3565 (not easy to find here in Brasil) was hfe=230 and a 2N2222 with a hfe=205.
The Ge PNP remained the 2N404A showing a hfe=40.

I changed both collector resistors from 100K to 91K. Feedback resistor for Si NPN = 680K and the Ge PNP = 220K. Input cap = 0.047uF polyester, the input-to-ground cap = 100 pF ceramic disc, bases cap = 0.1 uF tropical fish and output cap = 1uF polyester NP. The emissor to ground remained Tântalum 47uF. I, at this moment, gave up the 1K + 2u2F parallel to the Ge PNP collector resistor. And, as my config was very close to some of the known version pics, decided to keep the input diode-to-ground out of the circuit. Harmonic's Pot = 100K and Output Level Pot = 50K.

With both NPN's the measures was the same (power supply: 8.98v.):

Si NPN Collector = 4.70 v

Ge PNP Collector = 3.65 v.

Emissors = 4.03 v.

I did the output diode "clipping" stage with two 1N60 and a resistor of 10K. It works OK, but I will really put the switchable option to use a Si/Ge diode combo for more level and less compressed output in the finalized pedal.

Post comments:

I would like to get more "octave effect" and a bit less splat, retaining the wonderfull sustain when used with both guitar's Volume and pedal Harmonic's pot all opened up. And I like the HP most with my Parker P-30 (almost a Strato) with Golden lace Sensors. Hotter pickups are less interesting with HP in my opinion.

I hope to post some sound samples soon if anybody is interested.

[DougH]

Wow, that's really fantastic!! I could not get those voltages to come out like that with any of my transistors, even with similar hfe and resistor values. (Did you get the emitter and Ge PNP collector voltages reversed in your post?)

For more octave I would suggest lowering the Ge PNP collector resistor value. That seems to work with mine.

I played with mine for a few hours (too long) last night. Personally I'm kind of tired of the "octave up" sound. It sounds like an axis fuzz which is fine but I've already got one of those.

I went back to the 2n3565 NPN (hfe 200) because it sounds smoother than the other NPN's I tried. I am using a Ge PNP 2n404a 67 hfe but I have another Ge I'm thinking of using instead, one Joe G gave me a long time ago. I think it's a 2n614 or something like that? (Can't remember the p/n.) Anyway it has 160hfe and sounds smoother with less "octave up" than the 2n404a. I'm thinking of pulling the 2n404a out of my Brick, IIRC it was hfe 100. For some reason, higher hfe Ge's give me more of the sound I want. (Less splat and misbiased octave sound.) Incidentally I tried a couple AC128's and an AC188 and could not get them to work well. I couldn't get them to develop enough gain.

I am testing with a Squire Bullet Strat into my homebrew "vox" amp. One thing that is cool is G at the 5th fret D string produces this really wild harmonic/octave bloom that just takes off, even at real low volume. The circuit is very sensitive there. I agree that it sounds better with single coils.

FWIW I'm leaving out the 1k/2.2u filter now. I must have accidentally wired it up wrong the other day. I took it off and added it back and now it is just chopping out too much gain, which is what I expected to begin with.

Glad to hear you are making progress. Would love to hear a soundclip!

[bool]

Has anybody used BD 135/136 in the circuit? Or some other Si/Si combo that worked?

[DougH]

I'm going to try 2n3904/2n3906 tonight.

[alex frias]

The measurement was made using an asian "not-that-cool" digital multimeter direct on the built circuit working as it is.

Each value was measured from the referred point in relation to the ground. not divided in point-to-point parts (like Q1 collector-to-emissor voltage). But just a bit of basic arithmetic and you get all those values. Did I make it clear?

I've tested other Ge PNP's : AC128, AC188, 2SB54 and 2SB75. Nothing even touched the 2N404A, at least in my sampling universe.

I've tested other Si NPN's too: 2N3904, BC238, BC548, 2N5088 and I even tried a MOSFET 2N7000 (remembering Tim E.), which worked but with low gain and a mediocre overdriven sound. With Ge NPN's the thing didn't work at all, just crackling sounds at max.

I tested a pot in place of the fixed PNP collector's resistor to find a better value, but late at night it seemed going to the undesirable side, so I will try it out once again.

Soon I will provide a very unpretentious soundclip, maybe tomorrow.

I really like that "hoarse" (I don't know if it's the right word, as it's so similar to porn stuff...) and the octaved down effect. And I really would like to preserve those qualities and, if possible, even to refine and get more of them.

It's cool what we are geting here. I'm very happy with it.

Recently, in another internet stompbox forum, with a very good help from Mr Graemey, an old mysterious fuzz, the Baldwin Burns Buzzaround was reborn to original specs. Now I have two clones here, one PNP and one NPN.

I couldn't imagine this 2 months ago. And what a sound!

[Gus]

did some simple math

As a guess for Q1 using George G schematic

20K collector
Ic .05ma
Emitter at 3.5VDC

for a 2.5VDc drop C to E and 1VDC across the 20K
2.5VDC - .6VDC (Vbe) = 1.9VDC C to B

1.9VDC / 220K = about .009ma

.05ma / .009ma = about 5.5 Hfe?

So one would need a very low Hfe device
OR
greatly increase the C to B bias feedback resistor (1.9VDC/1meg =.0019ma,  .05ma/ .0019ma = Hfe about 26)
Or pick a higher total circuit IC and lower collector resistor value (maybe .1ma and 10K or ....)

This circuit is interactive and sensitive to the parts used

[alex frias]

Hummmmmm...

P.S. -> Please, where you read "Emissor" in my posts, substitute it for "Emitter".

[DougH]

Alex, please ignore my comments about your voltages. I incorrectly read what you posted. I understand now.

Gus, thanks for the tips. I should have done the math myself. I got into a vicious parts swapping cycle last night, thinking I didn't have much time. Next thing I knew it was late and I got about 5 hours sleep...

So, with a ~40 hfe 2n404a, which I have, it looks like I'll need roughly a 1.5M feedback resistor to get the target 50uA current. I will give this a try- maybe if the voltages and currents make sense it will provide a reference for how it is "supposed to" sound. Then I can tweak from that point to make it sound like I want.

Alex, one thing I forgot to mention is I tried a Mullard Ge NPN I had and it sounded fine. I didn't notice much difference between that and Si NPN but if you want the novelty of "all Ge" it will work with the right transistor.

EDIT: Going back through George's math in the other thread I think one of the key points is that with Vce so low, there's a wide range of bias settings that will work. This reinforces my observation that I'm not getting drastic changes in tone with bias and/or feedback resistor changes.

[kvb]

I'm just posting to mention that I did try Gus's two npn idea. (2n2222a) It works just fine. I also spent a bit too much time jammin and messed up my bed-time.  For most of that time it was working (with low volume) even though I had forgotten to connect Q2's collector.
When I started taking voltage measurements I noticed the problem.

I used 1M pots at the feedback adjustment nodes. Lowering Q2's just seems to drop the volume. Changing the value at Q1 can add some voulme, but the distortion does not seem to change much.

The tone is similar to the Differential Distortion in the sense that as the signal/note fades, the distortion seems to remain fairly constant.
Also, as the note fades there is not any nasty crackling that I've noticed in a few fuzzes that I've breadboarded.
There seems to be a good amount of cross modulation. I identify this by bending one note while holding another and listening for the "divebomb" sound.

I did not notice any octave sounds. Maybe I should change the 100u cap and listen for differences.
today I'll try different components and get some voltages.

If I should post future comments in a new thread, just let me know

[jaytee]

Interesting thread. It makes me wonder why they went and changed Q1 from a silicon to a germanium transistor. Maybe for lower gain or they thought it sounded better? I think the biggest problem with that is the leakage of the germanium. With the circuit only drawing 50uA it's all too easy to get at least that much leakage which will throw everything out. Maybe they had to select for very low leakage? It might be worth thinking about a low gain pnp silicon for Q1 to get around that the only disadvantage I can see is if it doesn't sound as good.

[DougH]

Well, I think I finally came to the conclusion of this for me. Last night I experimented with Q2 100hfe NPN and Q1 2n404a 40hfe. I used a 2.2M feedback resistor on Q1 and got the target voltages to come out pretty close to Gus and George's predictions. Unfortunately it sounded pretty horrible- thin, misbiased, fizzy sounding. Maybe that's the way it was intended to sound but I kind of doubt it. So after changing  back to stock feedback resistor and playing with the 2n404a's at 40-67hfe and a 2n614 I had with 160hfe, I came to the conclusion that with low hfe this had a wheezy "helium octave" sound (on the upper part of the neck) that I didn't like and just couldn't tune out. The 2n614 sounded much better in that respect but didn't have enough gain.

So I changed course and decided to just re-voice my Brick instead with the coupling cap values in George's schematic. I also swapped out the 1n46's for 1n34a's and used a 10k series resistor with one of the diode legs. There it was- nice full fuzztone with a fairly unique low freq growl that you hear in the Albini vid. No fizz, no wheeze, up above the 12th fret it maintains the thick, fat fuzz. Nice gain range and at lower settings it gave the strat a nice overdrive sound. It was late and I had to keep the volume down so I wasn't able to check the sustain/feedback/bloom stuff, but it did real well at low volume and sounds pretty sensitive.

Then I remembered that when I built the Brick I tried three different 2n404a's with hfe of 40, 67, and 100. I had chosen the 100 because it sounded the best. My conclusion is that first off, you need a 2n404a with an hfe of 100 or so. With hfe much lower than that, it does not sound good IMO. With the "right" transistor in there, you don't need the input diode or 2.2u/1k filtering or etc. At this point I suspect some of those circuit variations may have been band-aids to account for variations in transistors that were available. Secondly, I don't think collector voltage bias setting is that significant. With Vce of 1v, you have a lot of range within 9v to work with and move around without affecting the overall sound very much. It's like the first stage of a BMP with the grounded emitter in that respect. I don't know which version of the HP the "spec sheet" was intended for, but I don't believe the 50uA spec unless it had a 91k Rc on Q1 as well as Q2. With my circuit I was able to get the good sound with a 22k Rc on Q1 and Ic much higher than 50uA. So IMO the 50uA spec is essentially a red herring. Another thing I found out is it has plenty of output with the 1n34a's. The key is to use the 10k series resistor. That also affects the tuning and overall tone of it (you will see what I mean if you experiment with that resistor). One of Alf's original schems showed a 1k there IIRC, which will really limit the output signal swing.

This is not any kind of definitive "mystery solved" thing for the HP in general. But it does conclude this exercise for me and I'm really happy with what I ended up with.

Here's the schematic for my "updated" Brick: http://www.aronnelson.com/gallery/main.php/v/DougH/Brick/brick.gif.html?g2_imageViewsIndex=1

I'll try to come up with a sound clip later in the week.

I think I'd like to try to get hold of some more 2n404a's and sort them for higher hfe and build another one up on the breadboard. It would be interesting to see if I could get any sort of consistency and it might reinforce/disprove my premise that Q1 needs to have higher hfe.

[DougH]

With my circuit I was able to get the good sound with a 22k Rc on Q1 and Ic much higher than 50uA. So IMO the 50uA spec is essentially a red herring.

Actually, I take that back... I just did a quick calculation with the Q2 Vc I measured last night (2.1v) and it looks like Ic was around 69uA, not that far off... Hmmm...

Still, as far as good sound is concerned, I don't think bias voltage setting is that significant for this.

[alex frias]

Doug, I always found suspicious that current drain issue. At least with much accuracy...

WEll, I didn't have a lot of time to record the clips last night, so they was played in a hurry, but I think they will give a bit of the circuits sound.
Please, folks, forgive the lack of inpiration and cold-hands-playing!

There are both soundclips:

http://www.4shared.com/file/52700761/b7922c15/Perc2N2222.html
http://www.4shared.com/file/52700843/60f66886/Perco2N3565.html

and this is a hand drawn schematic:



Those clips was recorded using a Parker P-30, with Golden Lace Sensor pickups, bridge pickup only.
The controls on the project was all crancked up, only the guitar volume control was tweaked while playing.
A Marshall Plexi emulation, as ZOOM insists that we believe in it, and just a bit of digital room reverb was used and all the recording was done through headphones, as no crancked amp or loud acoustic feedback was allowed late at night. Next time I will try it earlier...

Still not there, but the results are very interesting in my opinion. With more spare time I will try different values on Q1 collector to ground resistor, supposedlylower ones, as Doug said.