My Overdrive: Samples and a "Buzz/Fart" Question (Again)

[John Lyons]

I was thinking that the transistor could be clipping as well. If you lift the diodes from ground you will hear what's going on there to be sure.

Side note: One thing that the series cap will do is clip the lower frequencies less, which will make the circuit a little fuller because it passes the low notes through more since they aren't bound by the diode clippers. Not a bad thing!!

John

[DougH]

Component values on the schematic would be helpful.

[railhead]

Well, I added a socket to mess with the series cap between diode and ground, but I'm still getting farting. Not as bad, but it's there. Looking at my schemtic again...



R1 is 3.3K and R4 is 680 ohm. Could increasing R1 (to 4.7 or something) drop the input to the transistor, thus decreasing its potential for clipping?

[jakenold]

Try to give us the component values. Then it's possible for us to actually help you with the specific circuitry. 

[railhead]

Here's the full schematic with my "best" values:

[PerroGrande]

Yeah -- your transistor is clipping for certain when you have the 50K pot set to its largest value.  With a 100mV input, I would expect it to start clipping when the total collector resistance (fixed + pot) is around 44K or so (this will vary a bit depending on the Hfe of the transistor you have).  Adding more total collector resistance will make the transistor clip sooner, which is not what you want.

With a 200mV input, which is certainly possible with humbuckers and the pot dimed (max resistance), the clipping is very hard and probably contributes to what you are hearing.  Plus, with all those diodes in the clipping chain, they're not going to get into the clipping game until the transistor gain stage is fairly "hot".

Several things to test:

Roll off either the gain control and/or the volume on your guitar and see if the farting stops.  If it stops and you still have the tone you want, then you have your solution -- you'll need to tame the max gain of the transistor (decrease the value of the gain pot to 20K or so). 

[DougH]

Okay,

First off, that's a really lousy way to control the gain. You will be radically changing the bias of the transistor (in not good ways) when you adjust the gain, as PerroGrande implied.

Secondly, the R3 c->b feedback resistor is not a good way to bias the base. You would be better off to establish a voltage divider from Vcc to ground to set the base voltage.

[PerroGrande]

Hi DougH,

I agree.  Using collector feedback to bias the base isn't the best solution.  Although it is noticeably better than a fixed value resistor to Vcc, it is still more subject to Hfe variations than the "4 resistor" method that we all know and love. 

Your gain-control comments are well-taken, too.  I wouldn't be surprised if turning the gain control knob didn't produce a little bit of "noise" (akin to a noisy pot) when you rotate it.  You *are* changing the DC bias characteristics of the circuit when you change the knob.  That said, I've seen plenty of circuits, some of them commercial (!), that do this.  If it were me, I'd put a fixed resistor on the collector and a pot on the emitter (1 lug to emitter, wiper to bypass capacitor, 3rd lug to ground) and control the gain that way.

Nevertheless, the circuit in its current incarnation *can* be made to work without clipping.  The gain will have to be tamed, however, to do so.

[railhead]

Like I said, this started off as a Trotsky drive, so the gain control is a hold-over from that circuit.

I'll start messing with your ideas...

[DougH]

it is still more subject to Hfe variations than the "4 resistor" method that we all know and love. 

I can't remember if there's a temperature stability issue with this method as well. My copy of "Art of Electronics" is at the office and I'm home until after the first of the year.

I know there's a lot of commercial stuff out there that is really lousy design, especially in the "booteek" stompbox world. Don't get me going on that...

[Gus]

Connect a voltmeter to the collector node, change the gain setting, note the collector voltages

The source feeding the circuit works with the C to B feedback R. 

Edit I just noted John posted the same idea
Maybe something to try a 1k pot wired like the FF type gain control for the emitter R and a collector R value you like.

[railhead]

If it were me, I'd put a fixed resistor on the collector and a pot on the emitter (1 lug to emitter, wiper to bypass capacitor, 3rd lug to ground) and control the gain that way.

So you mean replace R1 and the Gain pot with a single fixed? If Doug's 20k idea worked, plus the existing 3.3k, maybe something like 20k, fixed or so would be good enough to start with? What about R3?

As far as moving the pot, I'm not sure how to implement the wiper bypassing the cap (still new to this building stuff!). What cap are you saying to (potentially) bypass?

[PerroGrande]

Well, Vbe decreases about 2.1V per degree C.  This can creep up to bite us if the emitter voltage is well below Vbe.  This condition can be created in this circuit when the control is dimed.  Rising temperature would cause rising collector current and correspondingly change all the other operating points of the circuit.  Again, setting the maximum collector resistor to a lower value will help this somewhat, but at the cost of gain.  Bypassed emitter resistance is a better solution even with this biasing scheme.

However, the biggest issue with this remains the variability of the operating points of the circuit with transistor Hfe. 

For example -- running a sim with a 12K fixed collector resistance (gain ~17):

(Vc, Vb, Ve)

2N2222: 4.6V, 847mV, 247mV
2N3904: 5.8V, 811mV, 182mV
2N4401: 6.23V, 748mV, 157mV
2N5088: 2.9V, 997mV,346mV
2N5089: 2.53V, 1.02mV, 367mV

(Models using MultiSim)

[PerroGrande]

Something like:



will improve the gain control and DC biasing a bit.  You can diddle with R4 to set the maximum gain point. 

Disclaimer:  I have not rigorously tested or sim'ed this circuit.     It still suffers from many of the biasing problems that using collector feedback inherently has.  However, this circuit (or ones like it) allow the gain to be changed without destabilizing DC bias. 

Output is taken from the Q1 collector as before.

[railhead]

I gothca -- but I'll still keep my C2 on the signal going to the diodes, though, right?

[g3rmanium]

First off, that's a really lousy way to control the gain. You will be radically changing the bias of the transistor (in not good ways) when you adjust the gain, as PerroGrande implied.

Still, the first sample sounded really good. So if you can reduce the hard clipping then why not keep this style of gain control?

[PerroGrande]

Yep -- keep C2 and the clipping diodes just as before.

[railhead]

Something like:

How's this look:

[PerroGrande]

Looks pretty good -- give it a try!

[railhead]

Will do -- I just need to breadboard this instead of burning up copper clad!