Harmonic Perculator

[Sir H C]

It is pretty tricky.  It is an NPN and PNP gain stage, both are common emitter, with the common point for both emitters set as an AC ground with that big cap.  Given the variances of transistors and all that, they were definitely tweaked to get the biases right.  I have seen pictures of the insides of one on HC, and it was all sorts of whack components (glass encased resistors for instance).  Cool circuit, could be tweaked to be repeatable, but it should work if you play with the bias resistors to get the voltages ballpark.

[markm]

Isn't The BRICK the HP in disguise?

[Dan N]

John, thanks for the link to that photo. Besides a headache, I get somethng like this (reworked Barge schem):

http://users.rio.com/senorris/junk/notanotherschem.gif

Richard, mine does not seem to break up as much as the clip. The first one I built was much closer.

Chris, maybe he was tweaking values as he went. I suspect he was, eh, "different".

Today I tried different transistors, and pretty much kept getting a rather fat distortion. Q1 being PNP ge or NPN si did not make a whole lot of difference. One note, PNP si did not work at all.

[mac]

The dc circuit is very simple. As I said before, placing trimpots for the 4 resistors is the way to go. Then use whatever transistor you want, but 2 germaniums are better.

mac

[stobiepole]

I have a couple dozen 2n531 germanium transistors lying around. They're actually two transistors -an AC127 and an AC125 - paired in the same case, with five legs. I have always intended to use one to build a Harmonic Percolator. Problem is, I can't work out the pinout...the diode testing routine gives me nothing. If anyone can help me out with the pinout, I'd certainly be happy to part with a few.

Chris

[analogguru]

2N 531 is a single transistor in a TO-5 case -with only 3 legs.

Maybe there is a little mistake with the numbers ?

analogguru

[RLBJR65]

Isn't The BRICK the HP in disguise?

It's Doug's Harmonic Speculator which is his take on the HP.
Could be wrong but I seem to recall he was getting some grief over the name so he changed it to The Brick because of the way the finish on the enclosure turned out.

I found a few notes I made when I built Doug's version.
To my ears it sounded better with a low gain PNP GE and High(ish) gain NPN SI.
Don't know what PNP I used but the Hfe was only 36 and the NPN was a 2N5089 Hfe 591.

Really vague about the tone but I wrote that it was a nice OD to about 1/2; Distortion 1/2 to 3/4; 3/4 and up fuzz. 

[stobiepole]

2n531 spec sheet:

http://www.datasheetarchive.com/search.php?q=2N531

Five legs. Two transistors. Paul Perry told me they were probably used in portable radios...

Chris

[mattpocket]

Well, I told the dude I'd build him one, so I guess its time to order a million and one trannies and hit the breadboard!

See how things go.

These things are two knobbers aren't they? Do you think it would be worthwhile adding a tone control of some description?

Matt

[analogguru]

intersting:

My transistor-databook from 1981 says:

2N 531:  Ge-PNP   15V/25mA/100mW   ß: 25-30/1mA   fg: 3,5 MHz   Case:  TO-5 (standard pinout)
Manufacturers:  CSR-Industries,  Electronic Transistor Corp.

My transistor-databook from 1968 says:

2N 531:  Ge-junktion (planar)  15V/..mA/100mW  ß:25-30/1mA  fg: 1,5 MHz  Case: TO-9 (similar to TO-5)
Manufacturer:  General Transistor (Division of General Instruments)

hmmm....also the TO-9 case is known for maximum 4 legs...

Do you have a picture of this transistor ?

analogguru

[George Giblet]

Here's my take:

- Definitely at least two versions.

- For the second version Dan's "version 2" mark-up of Alphonso's circuit looks correct (after tracing against the PCB).  The clear things look like resistors close-up

The 101 cap is 100pF and the 503 cap is 50nF.  The only thing I would question is the 220k resistor which could just as well be another 737k.  The top transistor is an NPN silicon (say Alphonso's 2n3545) and the bottom is a PNP germanium (say Alphonso's 2N404A).

The circuit to me looks fairly straight forward.  The large 47uF cap forms a low impedance DC source.  The bottom transistor, Q1,  forms a CE amplifier with a bypassed emitter.  The top transistor, Q2,  forms a CE amplifier with a bypassed emitter.  The output from the Q1 feeds the input of the Q2.  The result is a two stage cascade amplifier without feedback.

The two transistors are biased at the same collector current due to the series connection.

The "manual" quoted supply current is ~50uA, which puts the collector of Q2 at about 9V - 50uA*91k = 4.45V, and the collector of Q1 at about 20k * 50uA = 1V.  The biasing of Q2 is fairly stable being a feedback bias.  Assuming the gain of Q2 is about 100 at the low currents we are dealing with the base current of Q2 will be 50uA/100 = 500nA and the voltage drop across the 737k will be 737k * 500nA = 0.37V.  That means the transistor is bias with a very low CE voltage.  The base will sit at 4.45V - 0.37V =4.08V and the emitter will sit at 4.08V -0.6V = 3.48V.

Now the biasing of Q1 implies that the GE transistor is leakage is the primary source of base current for Q1.  Remember the collector current is small here 50uA so the base current will be small and the leakage could easily bias the transistor.  The connection of the unknown 737k/220k resistor between Q1's B and E implies that some leakage current must be shunted away - this makes the biasing dependent on the transistor, which is bad for reliable circuit builds.  The resistor connection in Alphonso's circuit between C and B makes more sense when Q1 isn't leaky, and here Q1's CE voltage will be quite low. There's no point guestimating the leakage but I suspect the circuit may pull a little more than the assumed 50uA and that the collector of Q2 is a little lower than 4.5V in both versions of the circuit.

It appears both Q1 and Q2 are biased with very low CE voltages (which makes the circuit less sensitive to biasing issues).


(Forgot to mention, the 2u2 + 1k network lowers off the gain of the first stage quite a bit above 70Hz or so, pretty much all guitar frequencies.  The other circuit with the 2x100nF's would appear to have more gain.)

[markusw]

I got curious too now and gave it a try on breadboard, fed in a sine wave and ckecked the output with a software spectrum analyzer.
I used the Barge schem which should be indentical to Dan's version 2 mark-up of Alphonso's schem (besides Q1).
Si PNP as Q1 didn't work regardless of the orientation. A low leakage Ge PNP (AC128, hfe about 80) as Q1 like in Dan's/Alphonso's schem worked.
The only way of getting approximately the harmonics as indicated in the data sheet of the Percolator was with a NPN as Q1 like in the Barge schem. Also output levels were higher than with the AC128. Also I prefer the sound with the two NPN's.
I tried different NPN's (2N2222, 2N2369 and 2N3904) but there wasn't a real difference.
Don't know how it is supposed to work but with two NPN's the harmonics seem to be OK and the sound with guitar too.....

FWIW......

Markus

[George Giblet]

> Si PNP as Q1 didn't work regardless of the orientation. A low leakage Ge PNP (AC128, hfe about 80) as Q1 like in Dan's/Alphonso's schem worked.

That makes sense.  The mark-up circuit can *only* work if leakage is present so it can never work with Si PNP's; also if a PNP doen't work it is unlikely to work in the reverse direction.

>The only way of getting approximately the harmonics as indicated in the data sheet of the Percolator was with a NPN as Q1 like in the Barge schem.

The Barge version (2xNPN) has got to work, at least to some extent, whether it is the original is another matter!  In fact the equal transistors seems to point even more towards the clear C-B resistor being  the same value.  I had another look at the PCB's, in the image CIMG3335.jpg you can just make out the orientation of the transistors and it does imply the Barge circuit is correct for that PCB.  The two transistors look the same but it's hard to be sure.

The thing that is crazy about the barge version is the 1k + 2u2 is loading an emitter follower (ie. a buffer) it doesn't really make sense to do that.   I can only guess that someone fiddled around until they got a version that worked!

The pics corresponding to the other circuit show different looking transistors for Q1 and Q2, so I guess that version does use PNP.

As far as the PCB pics go there's two working circuits: barge NPN+NPN and the PNP+NPN with the 2x100nF (circuit not entire clear yet).  The Alphonso/Dan NPN+Leaky PNP in ckt works but there's no PCB pic for that version.

[markusw]

That makes sense.  The mark-up circuit can *only* work if leakage is present so it can never work with Si PNP's; also if a PNP doen't work it is unlikely to work in the reverse direction.

I have to admit it was just trial and error. 
So I put in trannies for Q1 and checked if there is any orientation in which they worked.

In fact the equal transistors seems to point even more towards the clear C-B resistor being  the same value.

Will give it a try with two 1M trim pots. Maybe the harmonics change?

I had another look at the PCB's, in the image CIMG3335.jpg you can just make out the orientation of the transistors and it does imply the Barge circuit is correct for that PCB.

Thanks for that hint!  I missed that pic...

I can only guess that someone fiddled around until they got a version that worked!

Sounds like my approach 

[mac]

That makes sense.  The mark-up circuit can *only* work if leakage is present so it can never work with Si PNP's; also if a PNP doen't work it is unlikely to work in the reverse direction.

When I bradboarded the HP I found that  a pnp 2sa49 leaking 20uA did not worked as well as a pnp 2sa101 leaking 100-1500uA, both with hfe around 50.

And a npn 2n388 leaking 30uA worked better than a npn 2sd352 leaking 150uA, both hfe near 50.

Moderate leakage for pnp, and none to very low leakage for the npn if Ge.

My all Ge did not have the final gain  of the (pnp Ge low gain, npn Si high gain) pair, but sounded more defined, fat and fuzzy.

mac

[dano12]

Interesting commercial development on the Percolater:

http://cgi.ebay.com/Harmonic-Percolator-INTERFAX-exact-duplicate-HP-1_W0QQitemZ110112284343QQihZ001QQcategoryZ41416QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

[johngreene]

Interesting commercial development on the Percolater:

http://cgi.ebay.com/Harmonic-Percolator-INTERFAX-exact-duplicate-HP-1_W0QQitemZ110112284343QQihZ001QQcategoryZ41416QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

That's this guy:

http://pages.prodigy.net/chuckcollins/percolator.html

nothing new.

[johngreene]

> Si PNP as Q1 didn't work regardless of the orientation. A low leakage Ge PNP (AC128, hfe about 80) as Q1 like in Dan's/Alphonso's schem worked.

That makes sense.  The mark-up circuit can *only* work if leakage is present so it can never work with Si PNP's; also if a PNP doen't work it is unlikely to work in the reverse direction.

>The only way of getting approximately the harmonics as indicated in the data sheet of the Percolator was with a NPN as Q1 like in the Barge schem.

The Barge version (2xNPN) has got to work, at least to some extent, whether it is the original is another matter!  In fact the equal transistors seems to point even more towards the clear C-B resistor being  the same value.  I had another look at the PCB's, in the image CIMG3335.jpg you can just make out the orientation of the transistors and it does imply the Barge circuit is correct for that PCB.  The two transistors look the same but it's hard to be sure.

The thing that is crazy about the barge version is the 1k + 2u2 is loading an emitter follower (ie. a buffer) it doesn't really make sense to do that.   I can only guess that someone fiddled around until they got a version that worked!

Not really that 'crazy'. It provides some high frequency filtering. I honestly don't think an emitter follower would find a 1K + 2.2uF as much of a load.

The pics corresponding to the other circuit show different looking transistors for Q1 and Q2, so I guess that version does use PNP.

As far as the PCB pics go there's two working circuits: barge NPN+NPN and the PNP+NPN with the 2x100nF (circuit not entire clear yet).  The Alphonso/Dan NPN+Leaky PNP in ckt works but there's no PCB pic for that version.

From what I have seen, the barge NPN+NPN and the Chuck Collins PNP+NPN are the only two circuits that have been verified. All other schematics I have seen were drawn in error. IMHO.

--john

[George Giblet]

> Not really that 'crazy'. It provides some high frequency filtering. I honestly don't think an emitter follower would find a 1K + 2.2uF as much of a load.

There's no doubt it will drive it but from a design point of view it just doesn't look right.  At 50uA, re for Q1 will be about 500ohms and that means the output network will form a divider of x 0.66 at high frequencies.  With the 2u2 cap the shelf EQ will start boosting frequencies at about 70Hz and shelf off at about 50Hz ie a bass boost.   They are wierd frequencies and a buffer stage with a gain loss in a distortion looks a little counter productive too - of course it's only a best guess.

The circuit 'notanotherschem.gif' corresponding to the 'oldpercolator.jpg', with the NPN + PNP is starting to look more like it.  The only thing wrong there is the missing parts (and is the 10k resistor in series with the diode correct?).  It seems that's the one Chuck Collins is pushing.  On that one I'm sure the PNP and NPN emitters go together; you can see the emitter tag on the PNP.   Sensible R1 and R4 could be worked out but that's not going to make it 'original' though.

Now what about those traces by Alphonso? where do they fit in?(!!!)

[johngreene]

> Not really that 'crazy'. It provides some high frequency filtering. I honestly don't think an emitter follower would find a 1K + 2.2uF as much of a load.

There's no doubt it will drive it but from a design point of view it just doesn't look right.  At 50uA, re for Q1 will be about 500ohms and that means the output network will form a divider of x 0.66 at high frequencies.  With the 2u2 cap the shelf EQ will start boosting frequencies at about 70Hz and shelf off at about 50Hz ie a bass boost.   They are wierd frequencies and a buffer stage with a gain loss in a distortion looks a little counter productive too - of course it's only a best guess.

So maybe a bass boost is what he was going after. Impossible to determine what his motives were, but it just doesn't seem -that- unusual to me.

The circuit 'notanotherschem.gif' corresponding to the 'oldpercolator.jpg', with the NPN + PNP is starting to look more like it.  The only thing wrong there is the missing parts (and is the 10k resistor in series with the diode correct?).  It seems that's the one Chuck Collins is pushing.  On that one I'm sure the PNP and NPN emitters go together; you can see the emitter tag on the PNP.   Sensible R1 and R4 could be worked out but that's not going to make it 'original' though.

Now what about those traces by Alphonso? where do they fit in?(!!!)

Version 2 of Aphonso's schematics is almost identical to the Chuck Collin's picture. So I believe that one. The 10K resistor in series with the diode is in the CIMG3335.jpg picture. So it is probably correct but since that resistor is different in the oldpercolator.jpg pic I have a feeling that it varied quite a bit over the life of the pedal. I doubt that the 220K went to ground as shown in Aphonso's version 1 schematic. If you connect that 220K to the collector of Q1 and make Q1 an NPN, you have the Barge schematic. The 2.7K is probably not correct but it really wouldn't matter much except for current consumption.

I didn't see 'notanotherschem.gif' anywhere. So I can't comment on it.