It's cute, but would it work?

[phaeton]

C D E D B D Ampwifwyers?



I don't have the ability to breadboard anything right now, but I can't seem to *not* think about schematics and designs and ideas and stuff.  Tried, can't do it, and it burns me up to no end.  So now I'm going to gain a small bit of relief by thrusting something half %^&*ed into everyone's face and convincing myself that i'm still "participating"!     The question here, is if this will work, and specifically if this will work as intended. I didn't cut/paste sections off of other pedals this time, I drew this one out of my head from my own understanding, then revised a few items from a review of the Natsemi datasheet, so there might be some glaringly obvious mistakes here.  I haven't seen anything like this (at least not implemented this way) in other circuits, but if this *does* work, i can't imagine that it's never been done before.  For clarity, I left out the typical stuff like DPDT switch, 1Meg resistors to ground, and filtered wallwart power supply.  They're there, you just can't see them .

Intended operation:

Use one side of a reasonably powerful opamp to drive the hell out of the other side of same reasonably powerful opamp

Do you think:

1) It will distort?
2) A lot? 
3) The "Input" control will act as a reasonably useful "Drive"?
4) The "Output" control will act as a reasonably useful "Master Volume"?
5) There are better choices than the TL082?
6) It will have a pretty flat frequency response? (in the guitar range, normal tuning)
7) This has been done before and it toadilly sucks canal water @ 15gpm?
9) I've used the term "reasonably" reasonably enough?
10) You have a better name for it?

Thanks for any and all comments, flames, and suggestions.

[Transmogrifox]

The circuit itself is not "out there".  For what it looks like it's intended to do, it's just a 2 stage op amp amplifier that would clip a little when you cranked the 1Meg pot on the input, and would distort a little more with the pot from the output cranked.

Unfortunately, the two noninverting inputs are connected to ground, thus rendering the op amps inoperable.  At best you would hear a blatty gated sound if anything at all in this configuration.  You would have a hard time getting much for a signal on the output.

You need to take those 2 noninverting inputs and connect them to a "V/2" reference.  A couple 10k resistors would work just fine, and you would thus have the op amp biased into its operating range.  Pinning the outputs to Vee or to Vcc (for that matter) puts the opamp into saturation before it even gets an AC input signal.

I did this with a 741 once, but I injected the signal on the noninverting terminal so it brought the opamp out of saturation for certain parts of the waveform, thus making a bad version of a half-wave rectifier.  It sounded terrible.  Wasn't long before I learned what my issue was and so I was off to go make some other dumb mistakes such as getting the pinout wrong on transistors and reverse connecting power supplies and things like that.  I still usually make magic smoke when I'm messing with power supply conditioning and voltage regulation.  I keep trying to get too creative without sitting down to learn on paper before pushing the plug into the wall. It's more fun that way, anyway.  You don't get to see big plumes of smoke and faces blowing off transistor packages when you discover a mistake on paper.  What fun is that?  Plug the thing in and see what happens

[phaeton]

The circuit itself is not "out there".  For what it looks like it's intended to do, it's just a 2 stage op amp amplifier that would clip a little when you cranked the 1Meg pot on the input, and would distort a little more with the pot from the output cranked.

Naw, not "out there".  The TS is similar, but it has a lot of other stuff in between to shape the clipping and tone and stuff.  I just want str8 up raw, ugly distortion here

Clip a little... hmm... Would adding a series resistor to the 1M pot create any more gain, or is this the typical limit for this opamp, you think?

Unfortunately, the two noninverting inputs are connected to ground, thus rendering the op amps inoperable.  At best you would hear a blatty gated sound if anything at all in this configuration.  You would have a hard time getting much for a signal on the output.  You need to take those 2 noninverting inputs and connect them to a "V/2" reference.  A couple 10k resistors would work just fine, and you would thus have the op amp biased into its operating range.  Pinning the outputs to Vee or to Vcc (for that matter) puts the opamp into saturation before it even gets an AC input signal.

I had a feeling about that.  I wasn't sure if there was the V/2 setup inside the opamp or if it was up to implementation.  While we're on the subject, you mention "10K resistors" for the divider, but i've seen other variations- 1meg, 100K, etc, all with a 9V supply.  Surely there's a rule of thumb for determining the values in each case?

I did this with a 741 once, but I injected the signal on the noninverting terminal so it brought the opamp out of saturation for certain parts of the waveform, thus making a bad version of a half-wave rectifier.  It sounded terrible.  Wasn't long before I learned what my issue was and so I was off to go make some other dumb mistakes such as getting the pinout wrong on transistors and reverse connecting power supplies and things like that.  I still usually make magic smoke when I'm messing with power supply conditioning and voltage regulation.  I keep trying to get too creative without sitting down to learn on paper before pushing the plug into the wall. It's more fun that way, anyway.  You don't get to see big plumes of smoke and faces blowing off transistor packages when you discover a mistake on paper.  What fun is that?  Plug the thing in and see what happens

I haven't destroyed a whole lot of things yet, but you almost make it sound fun.  About the best i've done is kill a couple of transistors in a non-spectacular way, and melt an LED.  I seem to remember once many years ago (non-guitar circuit) powering up something with an 18V supply, and then later realizing that my caps were rated for 16V.  Nothing bad happened, but it's something to think about, i guess.

Thanks for the comments!

[brett]

Hi.
First, connect the +ve op-amp inputs to Vbias.  To make Vbias, connect 2 x 10k resistors in series from supply to ground: the mid-point is Vbias (4.5V).  You'll also need to clip the signal somehow. 

Three well-established methods are:
1. Use a TL072 and connect the output to a small value resistor (e.g. 47 ohms).  You'll need a BIG DC-blocking cap out of the op-amp to keep the low frequencies flowing through this setup (try 100uF).  The op-amp suffers some serious stress when wired this way.  Tap your output from between the cap and the 47 ohm resistor.
2. Somewhat like a tubescreamer, use a couple (or even better 3) of diodes antiparallel in the feedback loop of the first op-amp.  3 x 1N4148s sound excellent.
3.  Like Rats, Distortion+, etc etc, use diodes to ground (again, 3 x 1N4148), between the gain stages.  You might need to filter some high frequencies after these diodes, because they often sound harsh.  (ie not as smooth as as diodes in the feedback loop)

keep working on it, there's a million magic circuits that nobody's ever dreamed of yet!

[phaeton]

Hi.
First, connect the +ve op-amp inputs to Vbias.  To make Vbias, connect 2 x 10k resistors in series from supply to ground: the mid-point is Vbias (4.5V).  You'll also need to clip the signal somehow. 

Those two non-inverting inputs won't have to be isolated from each other in any way, will they?  I could just wire them together and then connect that to the voltage divider, right?

However, do you feel that this chip (or this setup with this chip) is incapable of using one side to overdrive the other side?  I was hoping for lots of raw, ugly distortion without diodes, this time.  Fuzzfaces overdrive stage 2 and they use some (weak?) transistors, and I always thought that opamps would have a little more oomph than a single transistor would...

[MartyMart]

Phaeton, check out the recent Tech 21 XXL layout I did, you'll see its opamp
only distortion..... no diodes in feedback loop
That'l give you some ideas, it can be built with a TL072 or TL082 .....

Marty.

[phaeton]

Phaeton, check out the recent Tech 21 XXL layout I did, you'll see its opamp
only distortion..... no diodes in feedback loop
That'l give you some ideas, it can be built with a TL072 or TL082 .....

Marty.

Cool!  and Thanks!....   I'll fix the biasing of the non-inverting inputs (read up on voltage dividers, et al) and that should do it, eh?  Counting the days down to where I can breadboard this and play with it.

Thanks again.

[Gus]

http://www.montagar.com/~patj/mufffuzz.gif


The link is a fun circuit if you go to the schematic section I have a mod to it.   It is fun stock and fun to change parts.

[Transmogrifox]

Hi.
First, connect the +ve op-amp inputs to Vbias.  To make Vbias, connect 2 x 10k resistors in series from supply to ground: the mid-point is Vbias (4.5V).  You'll also need to clip the signal somehow. 

Those two non-inverting inputs won't have to be isolated from each other in any way, will they?  I could just wire them together and then connect that to the voltage divider, right?

However, do you feel that this chip (or this setup with this chip) is incapable of using one side to overdrive the other side?  I was hoping for lots of raw, ugly distortion without diodes, this time.  Fuzzfaces overdrive stage 2 and they use some (weak?) transistors, and I always thought that opamps would have a little more oomph than a single transistor would...

Back to your question about a rule of thumb when choosing V/2 reference divider resistors:

The main issues are noise and DC voltage drift.  You ideally want those resistors very small to reduce noise and make the parasitic temperature dependent bias current from the transistors a very small fraction of what is flowing through the pair so that it doesn't greatly effect offset--particularly in a configuration like you have where the op amps are DC coupled.  This offset gets amplified and may set the second stage into saturation.  For getting big nasty str8 up distortion, this isn't a big worry, since you're probably just going to go assymetrical, but not go too near saturation.

The conflicting issue with small resistor values is current draw from the pedal: Don't want to drain your battery in 5 minutes in the name of reducing noise.  It's a distortion pedal!

!0k just happens to be a very convenient value that keeps current draw acceptable, and doesn't degrade performance.

Now the next thing about gain:

You have a maximum gain of about 45.  This is pretty low gain for a distortion pedal.  It will just distort "a little" like an OD or something.  The Ibanez tubescreamer maximum gain is about 100, for reference.

I think to get a real raunchy fuzz you'll want a gain adjustable to at least 1000.  Lower the input resistor to about 10k--that'll load down your guitar pick-ups, but will give you a gain of about 50 (assuming about 10k output Z of a pickup) for the first stage.  Then lower the 22k down to about 5k1, and replace the pot with a fixed 250k resistor, and then you'll have a gain of 50 again...for a total maximum of 2500.  When you start getting toward a gain of 1000, the DC offset is going to be a pretty serious issue, so it will sound particularly raunchy as you krank it up.

Finally, yes, you can just connect the junction of the V/2 divider to the two + terminals just like that.  If your circuit goes unstable and oscillates, then you may want to try some things like adding a cap to ground from V/2, and maybe a small (33pF-47pF) cap in the op amp feedback loops.

In the end, you're ready to try it on a breadboard if you just add the V/2 to the inverting terminals.  It will distort, just not like big raunch fuzz, really...unless it's driving your amp into it.  It will be VERY loud on the output.  You may want the volume pot going from output to ground with the output taken off the wiper.

[R.G.]

While we're on the subject, you mention "10K resistors" for the divider, but i've seen other variations- 1meg, 100K, etc, all with a 9V supply.  Surely there's a rule of thumb for determining the values in each case?

There's a whole article on it at GEO - see "Designing bias networks to provide Vbias in effects" at http://geofex.com/circuits/Biasnet.htm.

Clip a little... hmm... Would adding a series resistor to the 1M pot create any more gain, or is this the typical limit for this opamp, you think?

Typical opamps have gains of 100,000 to a few million open loop. The feedback network tames this back down.

The real problem you have at very high gains is that you can't get a big enough feedback resistor to let it really rip. So you have to change to a feedback network that is not just a single series resistor. The simple way is with a T-network. Use two resistors in series from the output to the inverting input, and another resistor from the middle of those two to ground. In this circuit, put a big capacitor in series with the middle resistor to ground to avoid upsetting the bias. The resistor at the output and the resistor to signal ground divide the signal at the output down, and then the resistor from those to the inverting input feeds the divided down signal back to the inverted input. You can get huge gains this way with ordinary-value resistors.

The actual gain of an opamp circuit in the inverting mode approaches one divided by the attenuation of the feedback network. The T network lets you do a bunch of attenuation with low value resistors.

Those two non-inverting inputs won't have to be isolated from each other in any way, will they?  I could just wire them together and then connect that to the voltage divider, right?

Right - if the voltage divider is decoupled to ground by a cap. If you want it to have some self oscillation, you can leave off the cap, or even bleed a little signal into that node.

However, do you feel that this chip (or this setup with this chip) is incapable of using one side to overdrive the other side?  I was hoping for lots of raw, ugly distortion without diodes, this time.  Fuzzfaces overdrive stage 2 and they use some (weak?) transistors, and I always thought that opamps would have a little more oomph than a single transistor would...

Most people have no idea how much gain is available in a typical opamp. You don't even need the second side - your guitar all by itself will overdrive one side of the chip. This is the basis of Craig Anderton's "Optimum Fuzz Adapter" from his Guitar Player series. You can find it on the web, I'm sure. It was just one opamp set up with no feedback to run as a comparator. It did a LOT of fuzz. Playing a guitar with this thing on it was much like playing an organ.

It will be VERY loud on the output.  You may want the volume pot going from output to ground with the output taken off the wiper.

It will. It would be a good idea to put a large resistor in series with the volume pot to divide the output down by about ten at the max setting of the volume pot.

[phaeton]

Alrighty folks... in anticipation, i dredged this thread back up and read back through everything.  In the meantime i've studied up on things like voltage dividers and stuff.  Studied a bunch of other schems that do similar things and worked out an understanding of those sorts of things.  I've answered some questions, but now I've got new ones

1) I'm good on straight-line series voltage dividers, thanks to the geofex article and my used EE book.  I know how to do the math and figure out what I need and what makes what.  However, I haven't found a set of equations to address the T-network.  What set of laws or rules should I be googling for those?  I know I can just spew out the arbitrary "best guess" values and someone will tell me what I've made, but I'd like to know how to do that myself

2) About decoupling the voltage divider from ground with a cap.  I've often seen a lot of 10uFs and a lot of 100uFs used like that.  Is there *really* any importance to this value, or is it a matter of "10uF or more is all the same"?

3) You are right.  I had no idea that an op amp had that kind of gain capability.  I think I'll still use both sides of this one for this circuit, but I'll be sure to dig up the Optimum Fuzz Adapter.  It sounds kickin;

4) Thanks for the links, Gus and MartyMart.  I've actually put both circuits on the docket for "future project"

I've redrawn a new schematic based on the one above plus everyone's suggestions.  I for a number of things such as T-networks, oscillation stabilizers and such, I've chosen what I *think* are good values, based upon what I've seen in other circuits (some audio-related, some not), but I don't really know how to "do the math" to ensure I've got the right thing for this application.  I would appreciate any pointers to articles or suggestions of what to google on. 

When I get those parts ironed out, I'll dig out my scanner and post the new schematic for all to see.

Thanks in advance!

[phaeton]

I guess that not having a schematic too look at makes it tough.  I'll see if I can scan it tonight.

[Transmogrifox]

1) I'm good on straight-line series voltage dividers, thanks to the geofex article and my used EE book.  I know how to do the math and figure out what I need and what makes what.  However, I haven't found a set of equations to address the T-network.  What set of laws or rules should I be googling for those?  I know I can just spew out the arbitrary "best guess" values and someone will tell me what I've made, but I'd like to know how to do that myself

Look for "Kirchoff's current law".  Between that and your voltage divider equations you can figure out the ideal op amp.  The main thing to remember is that the op amp always tries to adjust the (-) input to match the (+) input.  Furthermore, the (-) input can be assumed *infinite* impedance, thus no current flows into there. 

If you feed a current through a resistor to the (-) input.   It has to flow from the (-) input through the feed back network to ground and to the output of the op amp.  The output of the opamp then responds by changing its voltage until the reactionary current flow puts the voltage on the (-) terminal back to equal with the (+) terminal...or until the op amp saturates.  For approximating closed loop gain, you assume that the op amp has an infinite voltage supply and does not saturate.

By that means, you can figure out the voltage gain even with a T network.