Unistage Screamer (idea for a distortion circuit)

[earthtonesaudio]

***Not Verified!***

I had this idea last night for single transistor distortion inspired by the classic Tube Screamer gain stage. 

Input goes through a 100k resistor (voltage to current conversion) and into the emitter.  Yes, the emitter is used as an input.  On purpose.
Signal output at the collector is fed back into, and biases, the base (negative feedback loop), and is also clipped with diodes.  The cap in series with the diodes keeps the bias from being messed up.
Large value cap from base to ground increases AC gain.  I would have included a series resistor there to be more true to the TS architecture, but honestly I forgot to draw it in. 
10k from collector to +V and 1k from emitter to ground creates a DC gain of 10.  AC gain is substantially higher.

...At least that's how I think this works.  I might be way wrong.  Try it out for yourself!


Parts values not set in stone.  Diodes should be Si or Ge or whatever you want, but LEDs probably won't work as well.
Transistor should be... I have no idea.  High gain, low noise Si probably.
1M resistor could be made into a gain control, tone control could be added to output, etc.

[ItZaLLgOOd]

I used a similar idea a while back for one of the monthly contests (Darling Muff was the name if I remember correctly).  It was the darlington transistor one.  I used 2 2n3904's for the darlington.  I didn't know what to expect as I was just tinkering around with AMZ's Muff Boost on the breadboard but it sounded very tubescreamer like.  I didn't do the resister and cap to ground though. I'll have to try it out.

EDIT:  Seems that all of the stuff that I had hosted is gone.  I'm pretty sure that I still have it if any one is interested.

[Purple People Eater]

Interesting. It looks like THE EASY DRIVE'S cousin. I wish I had time to breadboard it and see how the emitter input works.

[Ben N]

So the emitter and base are supposed to analagous to + and - inputs of an opamp? OK, but wouldn't you need to buffer the input?

[earthtonesaudio]

I thought a bit more about this snippet and I'd change some values:

-100k input resistor should be lower, 10k or less if possible.  This might make the guitar signal disappear though...
-10k collector resistor should be higher, 100k might be good.  But, this will decrease headroom, increase distortion.
-Output volume control should be 1M or eliminated altogether.  Plug straight into a tube amp, perhaps.

So the emitter and base are supposed to analagous to + and - inputs of an opamp? OK, but wouldn't you need to buffer the input?

Yes, the emitter is supposed to be the + input and the base is supposed to be the - input.  But it's far from ideal.  Changing values as I stated above might improve things a little, but the compromise-free way to improve the characteristics would be to add input and output buffers, and possibly an active load in place of the collector resistor.  But by that point it's basically a discrete op-amp and it seems somewhat ridiculous.

Interesting. It looks like THE EASY DRIVE'S cousin. I wish I had time to breadboard it and see how the emitter input works.

I never have time to breadboard things these days!  The best I can do is draw up a circuit and post it here.

[Ben N]

I still think it is a cool idea, but as a snippet, not as a standalone circuit. I don't think it needs an output buffer, but the input impedance is very low, so you have to assume it is preceeded by a very low output impedance. Anyway, we wont know more until somebody breadboards (or at least sims) it.

[DougH]

As you suspected, the 100k is too big. 100k/1k divider on input knocks the signal amplitude down by 100 to start with. With 10k/1k you will get almost unity gain, but no clipping through the diodes. (I simmed it and played with it.) You have a 3-way tradeoff between reasonable input impedance, minimizing input voltage attenuation, and minimizing degenerative feedback (you can't bypass the emitter with a cap since it is the input). Not to mention having enough current to bias it properly. It's very tough to solve with just one transistor and a high impedance guitar pickup as the source.

Part of the problem is you can't treat the emitter as the "+" input. A diff amp tries to keep the + and - node voltages the same. So in a TS circuit, the diff amp puts out an output voltage that tries to make the + and - voltages equal- that's how the feedback loop with the big resistor produces the gain. A CE transistor circuit does not work this way, it is just a simple gain stage

EDIT:

Okay, here's an update. Reduce that 100k down to something really small, a 1k or even 100. With it at 1k the diodes will clip and you will get almost 20db of gain out of it. Not bad for a simple circuit- might sound interesting. Of course, you will have to drive it with a buffer given that Zin is so low. But it might be worth experimenting with. (Check out Gus's overdrive for a simple one-transistor "screamer" sounding circuit too. Should compare these...)

[earthtonesaudio]

Cool, thanks for the info Doug!

I had the different ways of using the common base circuit floating around my head when I drew this, so there are elements there that would make it useful as a current follower, and other elements that suggest it would be used as a voltage amplifier... but apparently it can't do both at the same time!  Ah well...

[rnfr]

if you are going to have two transistors anyways, could you just use a differential amp?

[earthtonesaudio]

Yes, but using more than one transistor is problematic... I'd have to change the name.   

[Barcode80]

i like the idea doug id proposing. maybe the klon opamp buffer from beavis's site would be ideal. but then, it does make it a little closer to a plain ole distortion, but it isn't.

[Ben N]

Yes, but using more than one transistor is problematic... I'd have to change the name.   

Nah, a buffer isn't really a stage.

[DougH]

Did some more simulation of it this morning. If you lower the 100k to 100 you will get around 36db gain out of it. If you eliminate it (I simmed it with 1 ohm) it gets around 45db of gain (!). As you lower the value of that resistor you want to increase the input coupling cap. 1u looks pretty good for 100-1k. Might want to go higher if you eliminate that resistor completely. You could possibly make that a variable resistor (1k pot) and it would serve as a gain control and would act sort of like a treble booster as you turn it up, kind of the way the Gainster gain control works. Of course, make sure you buffer the input. And I'm guessing an LPF cap a la the .22u on the TS will be needed too. The output signal swing is limited to about 3v p-p, btw, but that's okay. Some fuzz boxes are much more compressed than that.

This looks interesting. I'm going to try it on the breadboard this morning.

[earthtonesaudio]

Yes, but using more than one transistor is problematic... I'd have to change the name.   

Nah, a buffer isn't really a stage.

Well, in that case, carry on!

Did some more simulation of it this morning. If you lower the 100k to 100 you will get around 36db gain out of it. If you eliminate it (I simmed it with 1 ohm) it gets around 45db of gain (!). As you lower the value of that resistor you want to increase the input coupling cap. 1u looks pretty good for 100-1k. Might want to go higher if you eliminate that resistor completely. You could possibly make that a variable resistor (1k pot) and it would serve as a gain control and would act sort of like a treble booster as you turn it up, kind of the way the Gainster gain control works. Of course, make sure you buffer the input. And I'm guessing an LPF cap a la the .22u on the TS will be needed too. The output signal swing is limited to about 3v p-p, btw, but that's okay. Some fuzz boxes are much more compressed than that.

This looks interesting. I'm going to try it on the breadboard this morning.

Cool!  Thanks for the continued interest, Doug.  One of these days I'll get to spend some time at the breadboard myself... sigh.

[DougH]

I tried it out this morning- .68 input cap->JFET buffer->1u coupling cap-> "Uniscreamer" stage.

I omitted the 100k resistor for testing. I also subbed a .22u for the 1u base to gnd cap. That tightened up the bass. I also added a .01u from collector to ground for a little high freq rolloff. Used an NPN with ~400hfe, so I used a 3.3M feedback resistor to bias the base and I got around 4.4v on the collector.

My gain predictions from the sim were wrong- as it doesn't seem to have as much gain as predicted. But that is dependent somewhat on the size of the input coupling cap(s) too, so YMMV. It sounds nice, very warm and sweet. Not as much gain on tap as my TS9 that I compared it too but good nonetheless. It makes a great booster for the amp too, turning up the volume can really push it. I don't think it needs a gain control- maybe just build it as a one knobber for boosting or etc. It doesn't produce a lot of clipping on its own, but the diodes do make a difference as I accidentally had them disconnected at one point. When I hooked them back up they added warmth and sweetness to it.  I think a linear taper volume control is the way to go here, as with audio taper you have it 2/3 of the way up to get unity gain.

The only issue is the hfe-dependent biasing using the feedback resistor. You will have to pick transistors for hfe or adjust that feedback resistor depending on what transistor you use. It might be worth trying a divider on the base to make the bias independent of hfe and temp. All in all it reminds me of Gus's overdrive circuit. I suspect it might have a little more gain although I haven't compared the two. It would make a good gain stage or mild overdrive on its own.

[earthtonesaudio]

Added the mods Doug described, and filled in details as I went.

[DougH]

Thanks for writing that up. I used a 10k in place of the 3.3k source to ground and there is a 1M/1M divider on the gate in the source follower. You might also want to note that the "unistage" setup  is for a transistor with hfe=400. The 3.3M will need to be adjusted for other hfe values (see pg. 71 Art of Electronics "Bad Biasing" paragraph). It's probably worth trying a voltage divider on the base to stabilize the bias and make it work for a variety of hfe values.

BTW, one thing I noticed with my 10W push-pull amp is this circuit really brings out the "sag". I installed a series "sag" resistor in the power supply of this amp but it was so subtle I didn't really notice any compression before. (It rounded off the highs a little which is why I used it.) But this little circuit really emphasizes that and gives it a nice subtle compression that really feels good. I don't know if it's this circuit compressing, or if the extra gain is just bringing it out with the amp more. I'm going to try it with my S/E amp again and see if it behaves similarly.

[Ben N]

I think the basic circuit sans buffer might go very nicely on the tail end of a Mr. EQ.

[DougH]

Yes, it would work good as an output stage, without needing the buffer.

Incidentally, here's the schematic of what I did, with the corrections:

[earthtonesaudio]

Yeah, it wouldn't need the input buffer if driven by a low output impedance.

I corrected the 1.1 schematic according to Doug's reply:



[edit] I was just a moment too late!  Thanks Doug.