"Piggybacking" trannies for lower gain

[aron]

Wonderful tones! No noise, easy to bias, great bass response, bell like tones!
I've just started to use Fuzzface (the easy face is my first ever) and I'm addicted to it...

Thanks for making people all around the world more happy about playing a guitar.

OH MY!!! Stop it!!!! It's only 11AM here!!!! I can't wait to try it!!!!

The Hornet will get a face lift!!!! :twisted:

[WGTP]

I seem to recall in R.G.'s FF article reference being made to controling the asymetrical vs. symetrical distortion on the Fulltone 69 by adjusting the hFE on Q1 and Q2 which resulted in the "ideal" hFE's of 90 and 110.  Although it appears the hFE's are a lot lower, the "balace" between Q1 and Q2 may have an impact on that ratio.   :shock:

Maybe a dual gang pot, one for each Q is in order.

[brett]

Hi guys.  Thanks for the thanks.  It's really a matter of being lucky, not smart.  I just made a lucky find when following up a suggestion by the real brains behind this (RG).  It does seem to be a really worthwhile find, too, judging by the response of people who have tried it.

First, concerning versions with emitter-bridging resistors. In the Miss Piggy, above, I nominated 1k resistors for Q1/Q2 and Q3/Q4.  That seems too small now,  I think that Kleber AG's idea is more like what will work well.  Perhaps 2.2k on the first gain stage and 3.3k or 4.7k on the second.
I tried lots of combinations of transistors last night, and a 1 k resistor only added, on average, about 3 to the hFE.  Previously, I have found that 3.3k gave an increase in hFE of about 10, and 4.7k up to 100 (on PN100s with natural hFEs of about 400).

I know that a few of you are probably already ahead of me on this, but I've done some work on piggybacking different types of transistors and across hFEs:  Basically, some types appear to have characterictically higher hFEs when piggied than others.  e.g. 5 PN100s stayed in the range hFE = 13 to 15, no matter what was on what.  With 5 BD139s (a mid power device) the hFEs were 2 and 3 (!).  When a BD139 is piggied on a PN100, the hFE is only 2 or 3, but the PN100 on the BD139 was 14, which leads me to think that some types of transistors are "lightweight piggybackers", and don't drag the hFE as low as others. Of the four types I tested, the BD139 was "heavy" (hFE~ 2 to 3), BC548s and BC549s were "medium" (hFE~5 to 7), and PN100s were "light" (hFE~10 to 15).

Anyway, keep the piggyback testing going lads and lasses!  
Anybody tried this yet in a Tonebender (if it works in a FF, it'll probably work in a Tonebender, right?) or a Rangemaster (I seem to recall that Germanium is prefered in the Rangemaster. Yes??  If so, maybe the piggyback is a goer in there, too) or what else?  Of course, it's gunna have its limitations.  I guess even if it applies mainly to fuzzfaces it's still opening up new possibilities for lots of future distortion pedals.

A bit OT: There's also another story to the Hornet (with piggybacked transistors) that started this thread.  It was my first attempt at a commercial sale.  So it's sitting in the local guitar shop waiting to be sold (on comission).  It's funny, but I feel like getting it back out of the shop because I am getting a sense that it's become part of stompbox history  (OK, in a MINOR way, but history all the same).  It's weird...  :wink:

[gez]

A bit OT: There's also another story to the Hornet (with piggybacked transistors) that started this thread.  It was my first attempt at a commercial sale.  So it's sitting in the local guitar shop waiting to be sold (on comission).  It's funny, but I feel like getting it back out of the shop because I am getting a sense that it's become part of stompbox history  (OK, in a MINOR way, but history all the same).  It's weird...  :wink:

Get it back and auction it on EBay - it's got enough publicity! (future collectors item and all that).

Nice work by the way!

[javacody]

What's funny is that I built my Sili Face with two sockets connected together for each transistor after reading RG's comments in the original thread, but was too lazy to experiment with piggybacking right away. It wasn't until I read about Brett's success that I popped the two extra transistors in there. What a discovery. We all really owe thanks to RG for suggesting the idea and to Brett for having the gumption to be the first one to try it. Good job guys! I've been playing my piggy Fuzz Face non-stop for the past 3 days. My main od/distortion sound used to be a tubescreamer. Not anymore!

Here is the layout I used on perf. I used the RunoffGroove layout and modified it a little bit. I'm not claiming this as my own, and I'm just placing it here as a help to anyone who may want to perf it. If B has any objections, I will redraw it.



It may make sense to not solder together the two collector pins of the sockets, simply leave the second socket's (for the pair) collector unsoldered. I left them soldered and bent up the collector lead on one of the trannies.

EDIT: I also need to thank B Tremblay for the nice layout and great article on RunoffGroove.com. Read the Sili Face article at http://www.runoffgroove.com/sili-face.html

[B Tremblay]

What's funny is that I built my Sili Face with two sockets connected together for each transistor...

Can you describe the difference between the stock Sili-Face sound and the piggy-backed version?

Also, you connected the collectors, whereas Brett's Miss Piggy left them unconnected.  Have you tried it both ways and this is better?

There's no need to completely re-draw the layout you posted, but it may prevent some confusion if you add a note that this is your layout for a modified piggy-backed Sili-Face.

[javacody]

The sound differences are very hard to put into words. With your original input resistor, I don't think there would be enough fuzz, so I removed that from the piggybacked Sili. The stock Sili sounds great. I would call it classic si transistor Fuzz. The piggybacked version is ge all the way.  The fuzz is warmer, rounder, and seems to "breathe" more. I have a very hard time translating what I hear into words, but it sounds a lot like Hendrix on the Are you experienced album (of course, I'm playing a strat with Fender Custom Shop 69's, which also really affects that), but I can also hear some Clapton in Cream type sounds with it.

Also, concerning the collectors, I only connected them on the sockets, I bend one collector's lead up out of the socket and no I haven't tried just parallelling the transistors, but I've heard (and had experience with this with op-amps) that two transistors in parallel rounds out the sound a bit, giving the fuzz a more tube amp like sound, whatever that means.  

[Bill_F]

I just tried the schematic that Brett posted using NOS Philipps BC109's. I put 10k trimpots on the emitters of each pair of transistors so I could play with the gain. I tried it with just the two transistors biased correctly and then put in the other two and biased them again.

The results were a marked improvement with the trannys in parallel. There was no loss of distortion and it just sounded better, more germanium like. The brittleness of the SI's were gone and it sounded warmer and fuller.

Guys I think you've definitely found something here. Thank you for sharing it with us apprentices.

Bill

[petemoore]

I'll never forget the way she looked the first time I saw her.
 Brett, this schematic couldn't be easier to follow and use to create a board.

[Ed Rembold]

I breadboarded Miss Piggy last nite,
but was hoping someone else would post first-

I mean no dis-respect,  to R.G. or Brett,
but there is No danger of this taking the place of Ge transistors.
Yes, it distorts!
but has none of the character of a well-tuned Ge FF.
I really don't think the "piggy-back" transistors are functioning as anything else but diodes to AC ground.
I removed the piggy-back transistors and installed a 1N4148+1N34A.
(in series) in each spot. Same sound.

history is hard to make, I think.
Ed R.

[idlefaction]

mine does.  i used 12k and 15k emitter resistors to get gains of 90 and 110 and i like it more than my germ ff

personal taste... maybe your 'tuned' Ge ff sounds much better than mine!  

[brett]

Hi again. Regarding

I mean no dis-respect, to R.G. or Brett, but there is No danger of this taking the place of Ge transistors.

I think that constructive criticism is the way to go, so don't apologise for not liking something.  Like you, I feel that there is a significant difference between piggybacked Si and Ge.  My personal feeling is that *some* combinations of piggybacked transistors sound good and have *some* of the good characteristics of Ge (low gain being part of it).   I was lucky that I used 2 x PN100s in my first circuit.  They offer higher gain (about 20) and seem to work better than many other combinations.  So perhaps people should nominate what trannies and synthetic hFEs they're working with.  If Ed's ff had hFEs of 3 and 3 I wouldn't expect it to sound very good, either.  

Concerning the opposite (high praise):

mine does. i used 12k and 15k emitter resistors to get gains of 90 and 110 and i like it more than my germ ff

This is interesting!  That's what I originally hoped to achieve with piggybacking.  Of course, some Si transistors already offer those gains without piggybacking. e.g. the BD139 (mid-power) has gains from 80 to 200.  I've already used them and liked them (especially for Q1), and refered to their sound in previous threads as "hybrid Si-Ge".  

It also appears that there's some fertile ground between hFE 20 and 50 that's waiting to be explored.

[Boofhead]

I really don't think the "piggy-back" transistors are functioning as anything else but diodes to AC ground.

I've only just read this thread and I agree, it's not a new idea.   GFR's had his idea on his page for an while but the idea has been around a lot longer than that.

There's no mystery why the gain drops either.  The voltage drop across a  diode junction increases as the current through it increases.  Without the series resistor, the two parallel BE junctions are forced to have the same BE voltage drop.  The base emitter voltage of the active transistor Q2a, the one with the collector connected is largely determined by it's collector current.  The base emitter voltage of the transistor with the unconnected collector, Q2b, is determined by the base current.  So to rough approximation, it follows that the base of Q2b is about the same as the collector current of Q2a.  In other words it's just a current mirror(!!!!) a very common circuit topology, which has a gain of 1.  If you go through the fine grain detail of transistors the current mirror isn't quite a perfect mirror and you end-up with a gain of 2 or 3, which are the figures which were push around previously.

[R.G.]

I mean no dis-respect, to R.G. or Brett,
but there is No danger of this taking the place of Ge transistors.

Respect isn't an issue - it's a techie idea. It works for some things, not for others. No need to pre-apologize.

Yes, it distorts!
but has none of the character of a well-tuned Ge FF.
I really don't think the "piggy-back" transistors are functioning as anything else but diodes to AC ground.
I removed the piggy-back transistors and installed a 1N4148+1N34A.
(in series) in each spot. Same sound.

history is hard to make, I think.

Every ear and taste is different. I guess more importantly, every guitarist looks for a slightly different set of things for their perfect sound.  As all of us know by now, there is no one thing to satisfy everyone's perfect sound, and in fact different people will perceive the *same* sound different ways.

Is the piggybacked transistor a perfect emulation of germanium? It can't possibly be. But it does seem to make a distortion sound that a number of people like. It's another technique in the tool bag. It side steps some of the things that people have not liked about silicon at some times in some circuits.

And yes, I agree, history is hard to make. In fact my personal opinion is that really interesting history can be best appreciated from a safe distance.  

Philosophy discussion over, technical discussion follows:

I really don't think the "piggy-back" transistors are functioning as anything else but diodes to AC ground.

The original idea was that a junction paralleled with the base-emitter of a silicon transistor would steal an amount of current from the functioning base, and by sheer larceny lower the current gain of the silicon to some lower level that's unobtainable with modern "semi-perfect" devices. I reasoned that equal voltages on the BE junction would drop the current gain by half.

That turned out to be wrong - the effective current gain dropped more than that, by actual test in a base-current/collector-current gain tester. There's something going on in there that lets the paralleled base-emitter steal more than its fair share of base current. I think that is because the paralleled BE junction has no transistor action going on, so there is no Shockley resistance in the emitter, and for equal voltages, the current density in the paralleled BE is higher; so it steals more than the naive half.

In the fuzz face circuit, yes, I'd agree that the action is as it might be with a diode to ground - that is what a paralleled BE junction on Q1 does. However, I think that in circuits with emitter resistors, you would still get a difference in effective transistor gain from the paralleled BE junction. It's still stealing base current, so the composite transistor still can't amplify as much as it used to.

I removed the piggy-back transistors and installed a 1N4148+1N34A. (in series) in each spot. Same sound.

Hmmm... what led to putting in two diodes, the Si and a Ge?

There is obviously no magic about the paralleled junction being a piggybacked transistor; anything that steals current in the same voltage range as the BE naturally functions will lower gain. I think that the most interesting circuits would have a nonlinear element like a semiconductor junction that was stealing current at the static voltage of the biased original transistor. If the paralleled element needs too high a voltage to conduct ( like, say, an LED) it won't ever get enough voltage from the silicon BE voltage to conduct, so it won't steal base current, and will have little or no effect. If the paralleled element has too low a voltage (like a Ge junction), it will steal almost all the base current and the transistor won't have much normal function at all.

I think it's like a see-saw - the more interesting things happen when the original and paralleled junctions are nearly matched. The simplest way to do that is to stick in another one just like the one that's in there. The current density argument and the various discussions on adding a resistor to the paralleled junction reflect the difference between the active transistor junction and the dummy/paralleled junction.

Anyway, that's what led to my question about why two diodes. Just offhand I would have expected that to boost the conduction voltage high enough to not have much audible effect.

I've only just read this thread and I agree, it's not a new idea. GFR's had his idea on his page for an while but the idea has been around a lot longer than that.

I'm certain it has. There are not many new ways to connect up NPN junctions after the field has been plowed for fifty years. Maybe it never got the attention it deserved. It took some active experimentation to turn an incomplete speculation into something interesting.

The base emitter voltage of the transistor with the unconnected collector, Q2b, is determined by the base current. So to rough approximation, it follows that the base of Q2b is about the same as the collector current of Q2a. In other words it's just a current mirror(!!!!)

Interesting. That is a valid way to look at it. I think that by not connecting the collector of the "dummy", it's forcing the current mirror to be quite non-perfect. I wonder if the mirror gets more perfect - and the effective gain drops to nearer one - if you connect the dummy transistor's collector to its base. That gives you the classic simplistic current mirror circuit.

Hmmm... you could connect up a number of "active" devices in parallel, with one "dummy" device connected collector to base, then base and emitter parallel with the active devices, and have a more-than-unity current mirror, a common thing in IC circuitry.

I'll have to think some more about that. Thanks!

[Boofhead]

That gives you the classic simplistic current mirror circuit.

Yes. No problem there provided the transistors are matched.  If you use a diode there is some mismatch.  The mismatch isn't a "problem" just it makes thing harder to pin down precisely.

As far as playing around with this goes, IMHO, putting a resistor in series with base of the unconnected transistor seem to be the best thing to try.  Adding the resistor gives am increasing HFe vs. current curve which generates even order harmonics in a *reliable* manner.  In the past I've spent some time searching for transistors with this characteristic only to find different batches an brands of transistors don't maintain the characteristic.

I suggest a switch with resistances in the 1k to 22k region for the series resistor and a second switch contact which switches different collector resistors in so the Q2 collector voltage is at 4.8V for each and every base resistor option.  This should give you a controllable smoothness FF.  2N3904's should be fine for the job.  The transistor with the unconnected collector can be a transistor or diode. For consistency of results I'd start with a 2N3904 from the same batch, but sticking in other transistors and diodes will still work- I think with the switch the different tones will be close to these options anyway.

[javacody]

Hmm. I'm not sure I agree with the naysayers. I have yet to try this with perfectly matched transistors, and the only transistors I have tried are bc109's in the range of 490 to 530, but the gains I'm getting are much higher than 1 to 3. I would consider gains of 8 to 12 (which I am seeing) quite a bit higher than what was mentioned. Also, this setup almost nails the smooth germanium sound to my ears, but I'm no self proclaimed expert, so take it as you will. There may be subtle differences, but this is the best fuzz I've found (not saying much, again, not an expert) yet. It smacks the dunlop reissue (with ge transistors no less) silly.

I guess I'm one of the pathetic few who would rather play guitar and drop in a couple of cheap and common transistors that retain their sound no matter the temperature than sort through a couple of dozen germanium transistors to find the "magic pair" which only keep the magic sound within a very finite temperature range. But hey, thats just me. For one, I'm not going to try to dimish this discovery, nor try to steal the credit from RG and Brett, who at least deserve a good clap on the back for waking a few of us up to some much better options than we had before. Good Job RG and Brett! Even if some others would dimish your contributions, I appreciate this wonderful discovery which is new to me and has significantly improved the fun factor of my playing!

[Boofhead]

Even if some others would dimish your contributions, I appreciate this wonderful discovery which is new to me and has significantly improved the fun factor of my playing!

I don't think anyone's out to diminish contributions by anyone.  GFR's site has been up some time and I know there are threads in the old archives discussing lowering the gain of Q2 for a FF-  in particular putting diodes, resistor and voltage dividers on the base of Q2 - the threads occured around 2001 and 2002.  I believe shortly after that Joe Davisson brought out his Obsidian transistor version which is a FF with a divider at the input of Q2.

[petemoore]

New, or old, this technique is new to me [I used to try it wayy wayy back with a Tweek-O or a FF but never really discovered anything interesting...at that DIY Stage I was mostly interested it 'keeping' it working].
  The fact is, that it does in fact allow absolutely wonderful Fuzz Face results [with whatever Q's you Choose], and can be used as yet another option to get desired results.
  I got from point A to Point B...with this "new" [to me] FF Q Gain selection technique...And I 'feel' like I know more about what a "FF" [what does a FF sound like?] sounds like.
  Most important of all of this is the way the darned ckt sounds with these low gain 'PiggyQ's'. By far the most fantastic and flexible [well of course you know I'm mixing ckt's and use an amp...lol] sound pallette to date...it all became new again too...the sound just draws you in and time really flies when You're Having Fun...there's just something about the way a FF with low'er' Gian Q's ... really quite amazing and hard to describe the touchy feels, the attackk envelopeS, harmonic structure, and general 'workableness' of it.
 SOme us may choose to go for the Low Gain FF sound and Mix that with Ckt Fragments LIke a high gain transistor gain stage, a Booster, A tone control [with OA drive] etc. ...TONS OF TONES to be had messin' there...

[javacody]

GFR's site has been up some time and I know there are threads in the old archives discussing lowering the gain of Q2 for a FF- in particular putting diodes, resistor and voltage dividers on the base of Q2 - the threads occured around 2001 and 2002.

That my be true, but did anyone ever come up with the very simple idea of simply using another tranny? It doesn't sound like it to me. I don't need to mess around with all that other stuff (allthough, if you can provide me a link, it would make interesting reading for sure) I can just chunk another tranny in there. That is the ultimate simplicity and a beautiful thing for sure. Something, that even a relative newcomer like myself can easily do and have great results. It is also now indexed in such a way that newcomers can easily find it. Two very important points that should be remembered. Of course, new ideas in hindsight seem simple, but no one else put it together in just this way, which to me, is significant, and the credit of the discovery goes with the people who did so. Just my 2 pfennigs.  

[idlefaction]

new ideas in hindsight seem simple....  heh, nobody's really figured out why it works yet so it's not that simple  

btw for whoever it was thinking about piggyback transistors used in power amps, that's used a lot with matched FETs.  you can even get pre-packaged stuff like this - a 'HexFET'  is six piggybacked FETs in the same three-leg package.  with FETs it basically raises input impedance, lowers output impedance and power handling, and i saw a mathematical proof of why it works one time but forget the principles now.

it's totally different from how bipolars work though.  bugger aye!