Harmonic Percolator (George Giblet's schem) build report

[alex frias]

Mr Hammond broke my heart... Well, then I will retake the research. I agree with his comment about the Brick's being a bit thin sound, as the Mr Albibi's Percolator has an almost octave down kind of effect attached to it's sound. As I just bought some 2N404A (they looks like new ones) I will try them with the George's circuit. At least in the brick version, the Si I've tested sounded better than Ge as the NPN tranny. I must retry it with the George's circuit.

To be continued...

[Gus]

It is good to read people have tuned this circuit to other things but it might be cool to build it as stock as possible to understand what the design is about if you don't have access to a real one that is in good shape and working correctly

I think to understand what the original "is about" one has to measure a real one.
OR build one
Using the math done by George G. The 50ua(.05ma) current draw is important, knowing this and knowing the stock collector resistor values gives two DC points,  Ohms law R X I =E(voltage)  George took the more knowable transistor specs the Si NPN and did the math.

http://www.diystompboxes.com/smfforum/index.php?topic=55987.20

Break the two gain stages out
Use a adjustable power supply to try and bias the stages/ select a transistor separate because of the interaction when placed in series

Think of the first PNP stage as one that runs at 3.48VDC and you want the collector at 1VDC above ground with the 20K(20000 ohms X .05ma = 1VDC) and using a bias resistor of about 220K.  This stages gain will be a mix of the openloop gain of the transistor circuit  and the feedback (C to B) resistor divided by the source R before it

Think of the 2nd NPN stage as one that runs at 5.52VDC and you want the collector at .97VDC above ground (4.55VDC drop across the 91K) and using a bias resistor of about 750K.  This stages gain will be a mix of the openloop gain of the transistor circuit  and the feedback (C to B) resistor divided by the source R before it (750K/20K(collector R of 1st stage))

Then connect them together in the circuit.  I don't see the stages being emitter circuits for the other side I see a 47uf cap holding the emitters at a  more fixed DC value (AC ground) and dividing up the power supply voltage to the stages  That point might wiggle a bit and maybe that is part of the type of feedback in the video not knowing what the 47uf electro cap specs are after all the years in the circuit.  SO maybe try different caps values.

[DougH]

I played with this a lot more this morning.

Here's what I found:

1. Si NPN with hfe ~200 sounds good. My 2n3565 was 220 and I found a radio shack grab bag xsistor hfe 182 that I like the sound of better. I doubt that is due to hfe alone, since they are so close. Just pick something that sounds good with hfe in that general range. I tried hfe 97 xsistor and it didn't sound good to me.

2. Ge PNP with hfe ~40. This sounded better IMO than the same 2n404a type with hfe ~60. Again, I'm guessing any Ge PNP with low hfe should work. I tried one with hfe ~250 and it sounded good but lost a lot of the "harmonic magic".

3. Pay Attention to Gus's post about voltages. The only way I could get these voltages was with Q1 Rc = 100k. At 20k the Q2 Vc was 2.7v. The key IMO opinion is what Gus alluded to- keep Vce on Q2 to around 1v or so. That's where it gets really sensitive and you get the harmonic blooms and etc. Set Q1 Rc to set Q2 Vc to taste.

4. The diode shunt on the input (in HC photos and Alf's version 2) compresses the signal and smooths things out a little. Yes it actually is audible and sounds a little better IMO with it in.

5. The 2.2u/1k network on Q1-C helps smooth out high freq splat as well. This and the input diode will give it a more "refined" sound.

6. 10k in series with clipping diode gives a little more volume and better octave harmonics.

7. Like I mentioned before George's circuit voicing with the .047u/.1u/.1u coupling caps was spot on and sounds very good.

Here's a summary of what I've got on the breadboard: http://www.aronnelson.com/gallery/main.php/v/DougH/Perc/myPerc.GIF.html?g2_imageViewsIndex=1

I'm going to perf this up and replace my Brick circuit board with it.

This is a fun circuit. Try picking lightly and letting the notes bloom. Or just finger with your left hand without picking. It pretty much plays itself.

Hopefully other people may try some of the debated circuit fragments on the breadboard and report back how they affect the sound.

[Gus]

Circuit looks interesting Doug.  I wonder, a med power Si for Q1?

For some reason I find this circuit  interesting

Q2 is set up opposite of what a RM is.  The RM is often biased more to cutoff and Q2 in the HP is closer to saturation This is the higher gain part of the bias range.

I have not heard or seen a real HP except in the video and pictures on the web.

[foxfire]

looking at the HC pictures of the HP, what is the black cylinder in the middle of the board? it's got a bit of that red goop on it. and if it's already been mentions and i missed it, sorry. rylan

[foxfire]

never the mind... rylan

[DougH]

One minor typo in my post above:

I found a radio shack grab bag xsistor hfe 182

Q2 hfe was actually 200. Not that it makes a significant difference...

[George Giblet]

Thanks for all your efforts.

Regard the biasing.  The calculation I did in the old thread were based on the 50uA supply current and working from there.   I've never been able to get a situation where I did actually get 50uA with the values in the circuit  - it's always more.

Using the values in my schematic (ie 750k's, 91k's etc)  the bias voltages I got (from some old simulations) were, going "down the chain"

                           Q1 & R4 Value
            Si+20k  Ge+20k Si+91k Ge+91k
Q2.C     3.1        2.8      5.3       5.1
Q2.B     2.6        2.2      4.9       4.7
Q2.E     2.0        1.6      4.4       4.1
Q1.B     1.37      1.41     3.74     3.92
Q1.C     1.29      1.36     3.69     3.89
** Q1.E  = Q2.E in all cases

So, (designators are for my schematic)
- When R4 = 91k the current is closer to 50uA.  Does that make the value correct? perhaps not!
  It could be the 50uA spec was for R4=91k; there were variations of the circuit over time.
  The 20k could be correct and the current was never updated in the spec!
-  The Q2.Vce voltage tends to stay around 1V by nature
   This is a consequence of the transistor gain and the value of R1.
   Smaller R1 values will drop Q2.Vce
- The Q1.Vce voltage is controlled by R2.
   Note Q1 operates with a very low VCe because R2 is quite low.


To: danielwarner
> In the photos posted on Harmony Central, there are three diodes on the PCB. Is it just a different version of the Percolator?

Yes there are other versions.  The Barge concepts schematic matched this circuit.  There was uncertainty to whether Q1 was a PNP or an NPN in the barge circuit.  Some of that was discussed in the old thread.

[Gus]

I built a circuit not tested yet

Q1 a 2n2907 PNP 190   moto metal can
Q2 a 2n2222a NPN 210 moto metal can

Q1
Using a supply set to 3.53 VDC (control is kind of coarse making it hard to set the voltage to a the .01)
Using a 10K collector resistor I wanted .1ma Ic
A 1.2 meg C to B  gave 1.057VDC at the collector of the 2n2907 for Q1

Q2
Using a supply set to 5.5VDC
A 47K collector resistor
A 2n2222a
A 470K C to B gave .957VDC C to E

Connecting the two fragments together
9.02VDC
4.37VDC Q2 collector
3.42VDC emitters
.986VDC Q1 collector
0VDC      ground


   Look close at the first stage.  It is basically the same as a FF input.  The FF bias and feedback comes from the C to B that is offset by the Vbe drop Q2 emitter tracks the Q1 collector offset by Vbe

Look at the HP there is a C to B feedback bias resistor.

Think about how the FF interacts with the pickup(s) tone control volume switching of a guitar
The HP is a lot the same: however it has a PASSIVE volume pot that adds series and shunt resistance to the input summing node of Q1 circuit.

Also what happens if you place FF after a device with a low output resistance?

[DougH]

George, my voltage readings came out similar to yours (for the Brick too). I think this is fairly "typical" if we are only tweaking the Q1 collector. To reach the 50ua goal as in the HP spec sheet I think you have to isolate ea stage as Gus suggested and bias them individually. It will probably take large tweaks to the feedback resistors unless you have the same transistors with the same characteristics that Giese used. I played with the feedback R's yesterday for a few minutes, but could not get the voltages and target Ic. I'm going to try it some more and may try Gus's approach (although I don't have an adj bench supply - so I'd have to jury rig something up). I'd like to get to the target 50uA if at all possible, just to see if there's any difference in sound.

[slacker]

I gave this a try today using Gus's method, I just hooked up an LM317 to get the voltages for the 2 circuit fragments and used a BC179 (hfe 250) for Q1 and a 2N3904 (hfe 190) for Q2.
I used a  680k for Q2s feedback resistor giving 0.94 volts on the collector but I had to use a 4M7 for Q1s feedback resistor get the correct voltage on the collector.

Joining the 2 halves together I get about the same voltages as Gus.

Q2 collector 4.54
Emitters 3.61
Q1 collector 0.95

I built the rest of the circuit following George's schematic at the top of this thread.

I've only had a quick play with it but compared to the Gearwise review of the BP-1 it seems to be in the right ballpark on the wilder sounds at least.
Compared to the cleaner sounds mine sounds too "misbiased" and splatty, I think I'll have to play with the collector resistors a bit more and try some of Doug's suggestions to see if I can improve things.

[Gus]

slacker

  Thanks for the feedback on you build.  I posted that bias method because I am wondering what the real design was about.  I think we will need to find a Q1 transistor(s) that will bias with resistor around the stock 220K.  4.7meg for Q1 will have the transistor operating close to its open loop gain in that fragment so "splatty" make sense.  The Q1 part looks like it be more of a problem to get "right"

   If you change the collector resistor of Q1 you will affect the gain of Q2 the 680K/20K part of the gain.  It seems to build it stock you will need to select transistors and feedback/bias resistors to get to adjust the operation points and gains of each stage

Maybe people here will find other voltage divisions and biasing/gain setups that they like better.  I think the simple break it in two parts to bias it will help with what ever people find they like if they want to reproduce the circuit

I posted an idea for a two npn circuit in another thread.
http://www.diystompboxes.com/smfforum/index.php?topic=55115.0

I did some thinking for a two NPN, Q1 collector resistor to the emitter of Q2 and Q1 emitter grounded

Q1 maybe try a Si med power device like a TIP29 or devices around hfe100
power supply set to 3.3VDC adjust the feedback resistor for the  collector at 2.5VDC ref to ground(1VDC drop across the 20K)

Q2 power supply at 5.5VDC and the collector at .95VDC(91K x .05ma = 4.55VDC drop across the 91K) adjust the feedback/bias resistor.

For the output I would try the warp type control(AMZ) the diodes to the outer legs of a pot and the wiper to ground a 10K should allow a little more output and you can adjust to taste

[slacker]

I think I'll have to go and read all the threads about this, because the answer's probably in there, but I presume the hfe of Q1 is important for setting the collector voltage using your method? Unfortunately I haven't got any other PNP transistors so I can't experiment with it any more at the minute.

I've got some low gain NPNs though so I can try your 2 NPN idea out.

[Gus]

two npn idea

http://www.aronnelson.com/gallery/main.php/v/gus/npnhpfragment.GIF.html

[JOHNO]

Howdy folks. Ive been following this thread for a couple of days now and whipped this thing up on the bread board today. I used a 2n2175s for Q1,its out of and old transistor radio that was made in australia. It has "AWV" printed on it,I guess thats the maker.HFE is about 40.Q2 I just used a 2n3904 HFE 210.The resistors I use after some experimenting are 100k and 680k as in dougs schematic.And on Q1 20k and 220k as in geoges schematic.Also I dont have pots connected to the circuit I just grounded a 100k on the input and also went to 100k on the output as it seemed to increase the octave effect. Now while im talking about the octave effect I notice through some fiddling that by lowering or raising the cap between the bases of the transistors i could change the range of the octave. But Im only getting the octave effect over about one octave on the guitar neck no matter what cap I use there.. I hope you understand what im saying as i dont explain myself well in writing. Anyway by fiddling with the cap i could change which octave was making the octave effect.LOL im confused now.Also twiddling that cap changes the voltages on the transistors. Are you guys getting the octave thing happening? Theres not been much mention of it. Anyway try caps from say .047uf up to .27uf or more or less if you like.

Heres my voltages with a 0.47 cap

Q1    E 1.51  B 1.42 C 1.36
Q2    E 1.51  B 2.09 C2.33

Im not sure if my little rant will help you guys as im out of my league but its too late now LOL
Have a nice day JOHNO

PS I left out the 1k and 2.2uf to ground as it took away too much grind for my liking also used 10k on the diode to ground as doug suggested,but i do have the diode on the input as in alfonsos schematic

[JOHNO]

>the cap between the bases.< Opps thats a typo. between Q1s collector and Q2s base is what i meant.

[slacker]

I had another crack at your idea Gus taking a different approach, I'm not sure how valid it is from the point of view of trying to understand the original but it seems to work.
For the PNP fragment I used the stock 220k feedback resistor and then "piggybacked" 2 BC179s, bases connected together and a resistor joining the emitters. I adjusted the value of the resistor to bias the collector to about 1 volt then hooked it up to my original NPN fragment and the voltages looked good.

Sounds pretty good now, at maximum gain it has a nice compression on the attack and it cleans up quite nicely with the volume control. Compared to just building the circuit stock using the same transistors which biases up with Q2's collector at somewhere around 3 volts it's a lot tamer, and to be honest I prefer the wilder version. Interesting though. 

[DougH]

Re. JOHNO's post:

That's good info and pretty consistent with the kinds of voltages I have been seeing.

Without having a real perc with xsistors Mr. Giese picked (and not somebody's clone), it's difficult to know where on the respective characteristic curves the operating points for these two transistors are supposed to sit (or how significant that really is). It's also difficult to know what the real "sound" is although you can get a good idea from the Albini vid. You can sort and pick through transistors until you find a pair that work with the known component values and with the spec'd Ic of 50uA.

Barring that, what I do know from experimenting with this a lot is you can get a decent fuzz sound with a lot of harmonics and an "almost octave" wildness (similar to the Axis fuzz) when it is cranked by using a variety of transistor types and making the necessary adjustments in Q1's collector resistor. Aside from the normal considerations of biasing and voicing with coupling cap size, you can get pretty darn close to the sound in the vid.

To be honest, I'm not really hearing anything real "special" in the vid or the circuits I've experimented with, other than it being a nice and fat, harmonically-rich fuzz with gain to burn. IMO there are a lot of ways of getting to that sound. From the standpoint of solving a mystery however- I completely understand the interest in it. I wanted to move on to other stuff and don't want to spend the time sorting through my transistors for the "right" one for this. But if you guys come across something definitive I would seriously love to hear about it.  I'm real happy with the circuit I have on the breadboard (I posted the schem earlier) and prefer it over the Brick. So that's what I'm going to build.

One thing I'd like to add is that all these Perc threads have been an absolute gold mine of info on this circuit. It would be nice to see this linked up in the Wiki somewhere.  All the variations, filtering options, biasing info is a great resource for anyone who wants to experiment with it.

And finally- here's a fun tweak I discovered by accident: Try connecting the output pot to the junction of the 1n34a and 10k (4.7k in George's schem) instead of the junction of the diodes. Increase or remove the resistor for more volume. This adds crossover distortion which sounds a little gritty but it will sustain almost infinitely at very low volumes.

[JOHNO]

>"To be honest im not hearing anything special in the vids or the circuits".  True it may not be any thing special but albini is defenantly getting an octave down thing happening to my ear when he is playing around the 5th fret. Im getting that too. I guess its just a happy accident as I havnt auditioned any other transistors in this circuit. Im only playing it through my champ and it has to be one of the best fuzz pedals i have built. I havent built a decent fuzz face yet and ive built a lot of them. Anyway thanks to you guys yet again for all your work and putting it out there for all to see cheers, JOHNO

[slacker]

True it may not be any thing special but albini is defenantly getting an octave down thing happening to my ear when he is playing around the 5th fret. Im getting that too.

Which video is this? I've only found the "Pussy Fuzz" one and that doesn't sound that crazy to me.