Active Feedback

[SlashLP97]

Anyone ever messed around with using active components as feedback elements for opamps?  It seems like you would get a sort of dynamic gain characteristic that would smooth out some usually harsh overdrive sounds...

Also, my submission for FX-X



As for right now, PSpice tells me that this is the ticket (at least what I was going for) but the J201 model stinks so I'll have to play around on the breadboard tomorrow.  Hopefully I'll have it up and running and sound clips ready.

In the meantime, anyone have any input on this idea?  Any ideas/thoughts/comments/complaints would be greatly appreciated! 

[PRR]

> PSpice tells me

SPICE lies.

There's no DC feedback around U1A. The opamp's offset voltage will be multiplied by the opamp DC gain. Taking 1mV and 100,000, the output wants to go to +/-100V. In fact it will slam to zero or Vcc.

SPICE says different because the simple opamp modesl have exactly-zero offset error.

If you get real-world DC stability.... you get mostly the inverse of the curvature of J2. For small signals it will be simple 2nd harmonic distortion. Probably no audible difference from a single common-source JFET stage.

At the limit... JFET gain increases only so much, but decreases to zero, giving somewhat boosted positive peaks and a smooth cut-off. Taking the inverse: when JFET gain goes to zero, U1A gain goes to infinity, signal tries to swing to infinity, slams/clips against Vcc.

> using active components as feedback elements for opamps?

"Exponential Converter". See if you can find that 1970s technology on the WWW. (You may be overwhelmed by CPU-based exp schemes.)

[Brymus]

Im trying to understand what youve done
Feedback loops in op amps are one of the few things I actually have some grasp on.
If I understand you right you are trying to voltage control the negative feedback ?
Whatever the answer,and in spite of the fact it may not work I like the way your thinking outside the box.

[SlashLP97]

I still need to test it on the breadboard to find the "real world" problems, but the lack of DC feedback shouldn't be a huge problem as long as the JFet is biased correctly, because as the output starts to swing up the JFet will make the input change lower so it should cancel out.  I should get to breadboarding it tonight and I'll post my results then.  Thanks for the input!

I know that it might oscillate a little bit as the feedback is not "instant" but I figured that could be fixed with a real simple noise gate addition (2 anti-parallel 1n4148s with a 470k resistor to ground or something, just to knock off a few mV).

As for the gain of the Jfet going to zero, that is why I have such a large divider on the Jfet input, I want it to be in the linear region all the time and not to ever reach the cutoff voltage (roughly -.9, depending on the manufacturer).

[SlashLP97]

I built this beast on the breadboard and it was a little disappointing, but also a bit of a victory.  It doesn't sound HORRIBLE (well, you guys can judge that) but for a first attempt at something that I was just thinking outside of the box with the fact that it actually worked makes me happy 

You can find the soundclip http://public.me.com/swcline and the file is "active feedback.mp3" 

The recording method was guitar -> breadboarded effect -> firewire interface -> GarageBand, no other effects or anything, not even EQ.  I built it without the presence control or the master volume, just for breadboarding simplicity and also for a lack of pots I could easily breadboard with.  Also, I added a 220p negative feedback capacitor to the opamp, just to mellow out some of the highs, but still no DC feedback.

The first part of the clip is just guitar -> interface, no effect
The second part is with the gain control at about 9 o'clock
The third is about halfway
The fourth is all the way up, this is basically just the open loop gain of the opamp.

I would really appreciate any comments or possible suggestions for how to make this better, I think the active feedback idea has some real promise, as it could sort of reduce the gain of a circuit during the clipping times.

Thanks guys/girls!

[Brymus]

Wow to be honest that sounds really bad 

[CynicalMan]

My suggestion would be to record through an amp! I have recorded some clips that sound crappy, but are actually great-sounding circuits, because I couldn't mic or simulate an amp. Amps compress and roll off highs in ways that are difficult to copy using DI. At least get an amp sim, like:
www.voxengo.com/product/boogex/ (Free   But keep drive at 0 dB)

Brymus, I'm guessing that the clip made this pedal sound way harsher than it is, so it may be a bit early to judge.

Suggestions: At least try putting in DC feedback, it will likely change the tone. Also, if the pedal is really as harsh as it sounds in the sound clip, increase the 220p cap, or add in LED clipping at the end. LEDs can smooth out distortion without changing its character too much.

[Brymus]

Your right Alex,
I forgot that it was dry direct,having an amp and cab sim could make a world of difference.
Nice link on the Boogex I have other Voxengo VST's that I like,
You might try the Acme Bar Gig stuff I really like it > http://www.acmebargig.com/cab-enhancer/
And >http://www.acmebargig.com/razor-head/
Nearly all their amp sims sound pretty convincing,check em out

[SlashLP97]

Brymus, I'm guessing that the clip made this pedal sound way harsher than it is, so it may be a bit early to judge.

Even through an amp it is pretty harsh (Although I was using my HD-130, which sounds like crap when you push the input and I had no MV on the circuit).  I'll try it with a MV and if it makes a difference I'll mic that and post a new clip.

Suggestions: At least try putting in DC feedback, it will likely change the tone. Also, if the pedal is really as harsh as it sounds in the sound clip, increase the 220p cap, or add in LED clipping at the end. LEDs can smooth out distortion without changing its character too much.

The idea was to make the feedback circuit affect the actual signal input and pretty much be using the open loop gain of the TL072 the whole time, but I could try putting DC feedback and see if the JFet even does anything then.


And, once again, the idea was to see if you could use some active device to provide feedback instead of the typical passive components (which I have never seen done before, though that doesn't mean that it hasn't been).  So, now I am going to see what I can change to make it sound better, but I think the concept is proved.