Si FF rant

[Meanderthal]

Our bass player was pretty much carted to gigs in a cage and let out to play.

LOL! sounds just like most guitar players I know(the good ones).

I detect a little ummm envy from you guys... We can't help it that 'ours' are bigger than 'yours'.

Ok, I'll shut up now before someone thinks I'm serious.

[Jay Doyle]

I think the trimpot in the drain/collector/etc circuit has been done to death and misapplied quite a bit. It's really a stupid way of biasing a circuit, if for nothing else, the way it will change gain and make the gain of a stage inconsistent from device to device. I suspect it got started with the JFET "amplike" circuits as a quick & dirty way of dealing with inconsistent JFETs. There are app notes on JFETs that describe many alternate methods for biasing. I suppose some might be more suitable for dealing with the inconsistent JFET but I haven't researched it. In the end it will further complicate a circuit that is already more complicated than is necessary IMO.

Here is the link to the best biasing app note for JFETs: http://www.vishay.com/docs/70595/70595.pdf

And actually, in my opinion, it doesn't really complicate the circuit any to take out the trim on the drain and move it to the gate/base as a variable voltage divider feeding the bias resistor. Take the 'standard' trim-in-the-drain scheme: gate resistor to ground (two in the case of a BJT/MOSFET, the other goes to V+), resistor and cap on source to ground, and the trim on the drain.

Now, if you replace the trim on the drain with a fixed resistor, keep the source the same and move the trim to be a variable voltage divider that feeds a gate resistor,  you've only added one resistor (on the drain) in a JFET circuit and kept the parts count the same for a MOSFET and a BJT. Plus you have complete control over the whole biasing range, aren't messing with the gain, current (which moves where you are in the transfer curve), Miller cap roll-off, etc., you don't have the problem of DC bias across a trim on the output, and as you can see from the above referenced App Note PDF, it flattens out the load line of the stage to accomodate a wider range of devices. And all it costs is an extra connection needed to ground.

Regards,

Jay Doyle

[Doug_H]

Thanks Jay, yes I remember we talked about this before. And yes, that was the app note I was referring to. It sounds good although it may require more coupling caps since the "old method" has the gate referenced to ground- depends on the circuit. It almost sounds like a pseudo fixed-bias, except Vk > 0, so not quite.

One day I'll give it a try. It goes on the list...

[Jay Doyle]

Thanks Jay, yes I remember we talked about this before. And yes, that was the app note I was referring to. It sounds good although it may require more coupling caps since the "old method" has the gate referenced to ground- depends on the circuit. It almost sounds like a pseudo fixed-bias, except Vk > 0, so not quite.

Well assuming that you originally had a DC bypass cap in there, like you were supposed to  , you shouldn't have to add any. I say 'supposed to' because you don't want any previous DC to inject its bias onto the gate and take the gate AWAY from ground.

One day I'll give it a try. It goes on the list...

I think you are doing just fine without it. 

Tonally, the difference isn't really all that much, to me, JN might disagree 

BUT, the ability to keep the circuit relatively the same while biasing AND have it work across JFETs is priceless to me.

[brett]

Hi
Gus made a couple of excellent technical points.

I am suprised that even after all the years stuff has been on the web people keep building a Si FF with the Ge circuit.

Specifically: Rc1 = 33k and Rfeedback = 100k.
For modern transistors, in my opinion, Rc1 = 10k and Rf = 47k make sense.  Rf = 33k, maybe.  An early FF I made accidentally used a 33k feedback resistor, and it sounded good.  Even my Ge FF (hFE 100  and 140) uses a 68k feedback resistor.

I should qualify these comments by saying that the Axis Face avoids some of these issues by using an unusually low hFE for an Si Q1 (about 70 - the 2N/PN 2369A).  It also adds a simple filter that cuts excessive highs.
cheers

PS Although I'm a long-time fan of BD139s, I've been using a few TIP31s lately, and they are cool for both Q1 and Q2 in a fuzzface.  (hFE of 85 for my current batch of devices)

[Gus]

Yes med power transistors are fun I posted about them years ago.

Schematics has
http://www.diystompboxes.com/pedals/gusFuzzFace.gif

Been there for years The 10K 100 ohm first stage was done by math and ear drops the open loop gain of the first transistor. Note the lowpass filter 1K .01uf.  The rocket has the 10K 47K

Thanks for the feedback

[tcobretti]

I haven't really thought about it, but lowering the feedback resistor lowers the overall fuzz and makes the pedal more sensitive to changes in the guitar's volume knob - two traits that we expect from a Ge FF.

I built an Si FF with MPSA18s recently that sounded great with a 100k Rfeedback, but worked very much like my Roger Mayer Classic Fuzz with a 33k-47k instead of the 100k.

So, I think you guys are on to something.  Interesting stuff!