Rat opamp distortion question

Started by add4, February 10, 2014, 02:31:03 AM

Previous topic - Next topic

add4

Hello guys,

I've got one question about the design of the rat.
Let's start talking from the generalguitargadgets schematic :
http://generalguitargadgets.com/pdf/ggg_rodent_sc.pdf?phpMyAdmin=78482479fd7e7fc3768044a841b3e85a

As i understand it:
usually, the (inverter or non inverter) pin of the op amp that gets the input guitar signal is usually forced in the middle of the voltage range fed to the op amp (here : 0 - 9V) -> we would usually create a 4.5V bus using a voltage divider (R15 and R16) and then connect the input signal to that 4.5V source so the opamp gets (4.5V + - guitar signal oscillations). This allows to have the biggest possible signal amplification from the op amp without clipping (clipping happens when the output signal > 9V or < 0 V)

In the rat, the idea is to use the op amp clipping in addition to classical diode clipping -> R1 is added in series to the 4.5 source, effectively redicuing the baseling voltage of the guitar signal fed to the op amp, making it (much) closer to 0.-> with a bit of amplification from the op amp, the output signal hits 0V, gets clipped, distortion happen.

If that is so, wouldn't that be interesting to replace R1 by a 1M pot so we could effectively control the amount of opamp distortion happening? with a dual gang pot and a resistor in series to the 2 diodes, one could also control the mix between opamp and diode distortion.
Is that correct?

Thanks for you input


kingswayguitar

i just tried something like this with the tube screamer topology. didn't work too well. the distortion i was getting with lower value resistors (a.k.a. R1) wasn't pretty. by the time i got down around 39k it was just some bad farting sounds. at 10k the distortion was so bad it was just cutting out.

dwmorrin

Quote from: add4 on February 10, 2014, 02:31:03 AM
Hello guys,

I've got one question about the design of the rat.
Let's start talking from the generalguitargadgets schematic :
http://generalguitargadgets.com/pdf/ggg_rodent_sc.pdf?phpMyAdmin=78482479fd7e7fc3768044a841b3e85a

As i understand it:
usually, the inverter or non inverting pin of the op amp that gets the input guitar signal is usually forced in the middle of the voltage range fed to the op amp (here : 0 - 9V) -> we would usually create a 4.5V bus using a voltage divider (R15 and R16) and then connect the input signal to that 4.5V source so the opamp gets (4.5V + - guitar signal oscillations). This allows to have the biggest possible signal amplification from the op amp without clipping (clipping happens when the output signal > 9V or < 0 V)

???In the rat, the idea is to use the op amp clipping in addition to classical diode clipping -> R1 is added in series to the 4.5 source, effectively reducing the baseling (baseline?) voltage of the guitar signal fed to the op amp, making it (much) closer to 0.-> with a bit of amplification from the op amp, the output signal hits 0V, gets clipped, distortion happen. ???

If that is so, wouldn't that be interesting to replace R1 by a 1M pot so we could effectively control the amount of opamp distortion happening? with a dual gang pot and a resistor in series to the 2 diodes (R6?), one could also control the mix between opamp and diode distortion.
Is that correct?

R1 is a 1M resistor that biases the non-inverting input to 4.5V.
You can make it as small as you like, and the op amp will still happily accept that it is receiving 4.5Vdc.
Your signal, however, will simply be getting more and more attenuated by C12, R15, and R16.  Lowering R1 lowers the input impedance.  Generally, we avoid that.

The gain of the op amp is set by R3, R4, and R5.  Having R3 variable controls the amount of op amp distortion.  Having R1 variable would control guitar input attenuation - and probably not a very good version of that control, but YMMV.  R6 variable I can't say speaks to me either.
Try your ideas on the breadboard and let us know your results.

JFace

The op amp distortion is caused by the low slew rate. Changing the input bias will probably not have a desirable effect.

ashcat_lt

The voltage dropped across R1 is completely dependent on the current draw of the opamp input.  The positive input of this opamp draws very very little current.  To have a significant impact on the bias voltage, R1 would have to be very large indeed.  If you really want to change the bias voltage, then change either R15 or R16, which are the divider that actually sets the bias!

But that isn't necessarily going to cause any more clipping, it'll just make the clipping more assymetrical.  That might be worth pursuing, but it's not quite what you said.  If you want to clip the opamp harder, you need more gain in the circuit itself.  Increasing R3 is probably the best way to do that.

In the end I'm really not sure how much it matters.  The diodes end up smashing the signal so hard that it's difficult to say that the opamp clipping makes any difference.  Extreme misbiasing will of course lead to gating and farting, and of course you should try it yourself just to see, but idk that it's going to blow your socks off.

And no, the distortion is not primarily caused by slew rate.  This circuit will cause distortion no matter what opamp you put into it because it's asking for hundreds of volts out of a 9V supply.  Every opamp will clip in these conditions independently of slew rate.  A very slow slew might contribute to some of the tonality quality.  Essentially the higher frequencies will distort a bit more than the lower if all other things are equal.  Except, of course, they're not equal.  The high frequencies which might be affected by slew rate don't get as much gain as the lower frequencies.  So even that is a bit questionable.

HTH

add4

Quote from: ashcat_lt on February 10, 2014, 11:52:27 AM
The voltage dropped across R1 is completely dependent on the current draw of the opamp input.  The positive input of this opamp draws very very little current.  To have a significant impact on the bias voltage, R1 would have to be very large indeed.  If you really want to change the bias voltage, then change either R15 or R16, which are the divider that actually sets the bias!

But that isn't necessarily going to cause any more clipping, it'll just make the clipping more assymetrical.  That might be worth pursuing, but it's not quite what you said.  If you want to clip the opamp harder, you need more gain in the circuit itself.  Increasing R3 is probably the best way to do that.

In the end I'm really not sure how much it matters.  The diodes end up smashing the signal so hard that it's difficult to say that the opamp clipping makes any difference.  Extreme misbiasing will of course lead to gating and farting, and of course you should try it yourself just to see, but idk that it's going to blow your socks off.

And no, the distortion is not primarily caused by slew rate.  This circuit will cause distortion no matter what opamp you put into it because it's asking for hundreds of volts out of a 9V supply.  Every opamp will clip in these conditions independently of slew rate.  A very slow slew might contribute to some of the tonality quality.  Essentially the higher frequencies will distort a bit more than the lower if all other things are equal.  Except, of course, they're not equal.  The high frequencies which might be affected by slew rate don't get as much gain as the lower frequencies.  So even that is a bit questionable.

HTH

Very interesting .. and that's exactly what i'm asking, except that i want LESS opamp distortion!
What makes that opamp ask for so many volts then? I don't see many changes between this design and say a classical YATS distortion.. except than then, if you remove the diodes, you get a 100% clean boost.

dwmorrin

Quote from: add4 on February 11, 2014, 08:11:25 AM
Very interesting .. and that's exactly what i'm asking, except that i want LESS opamp distortion!
What makes that opamp ask for so many volts then? I don't see many changes between this design and say a classical YATS distortion.. except than then, if you remove the diodes, you get a 100% clean boost.

On your schematic, the gain can be figured by dividing R3 by R4||R5 (for frequencies that see the caps C5, C6 as low impedance) and adding 1.  The extra 1 doesn't make much difference in these high gain circuits.

For simplicities sake, just divide R3 (max = 100k) by R5 (47Ω).   The equation works out to a gain of 2128.

This means that the op amp is designed to take whatever ac voltage wiggle is at pin 3, and "try" to produce the same wiggle 2128 times bigger at the output.
If you input a 1mV (that's 0.001V) wiggle, a 2V wiggle should come out.

If your guitar is more like 100mV, then the op amp "wants" to produce a 212V output.  However, you don't have a >212V power supply... you are limited to a 9V output at best.  The op amp clips as your signal approaches 0V on the negative swings, and 9V on the positive swings.

Comparing to a TS-808 schematic (that's a YATS, right?), the same gain equation works out to (551k/4.7k) + 1 = 118.   118 is WAY less gain than 2128, so comparing tube screamer and RAT gains is apples and oranges.

Increase the values of R4 and R5 if you want to tame the RAT's gain.

ashcat_lt

#7
Quote from: dwmorrin on February 11, 2014, 09:47:54 AM
Increase the values of R4 and R5 if you want to tame the RAT's gain.
...or decrease the pot value.  Or do what we did in the old days and turn the knob!  ;)

Note that anything we do to values of these resistors also changes the cutoff frequency for the filter caused by those caps around them.  You can re-calculate their values based on the new values of whatever you're changing, I suppose, but you can't really get the same frequency response as the original with the gain wide open, because of the way that those filters sort of shelf off.  

Reducing the value of the gain pot acts exactly the same as turning the knob down, and if that works for you than the stuff about the filters isn't an issue.  You could also put a resistor across the gain pot to reduce the value, though it will change the taper a bit.

Edit -
Yeah, of course, reducing the gain of the opamp stage will end up causing less opamp clipping, but it also reduces the amount of diode clipping. 

So, you could replace those diodes with lower forward voltage to compensate, but then you end up with less output level, so you end up wanting to add some makeup gain after.  Could probably modify the output buffer...

...Or you could leave everything else exactly as normal and increase the supply voltage.  You can get somewhere around 34V swings out of most modern opamps at their power supply limits.  It maintains all of the filter action and the diodes still hit at the same point of the waveform.  You'd want to check the maximum voltage ratings on the 2n5458 and all the caps in the circuit, but I don't believe it needs any other modifications to run off a bigger supply.

You could also diode clip the input itself to make sure that input x gain never equals more than you want.

I don't expect you'll hear much different in the clipping itself, though.  If you could take out all the filters (which we know will change if we change much) and really just listen to the distortion, I don't think you'll hear much difference with any of the methods mentioned, as long as the diodes are still clipping it to less than 0.001 of what it wants to be.  Perhaps there will be some subtle difference in the lower ranges of the gain knob, but I'm not sure it's worth the trouble.

TubeScreamer and Rat are arguably the definitive examples of opposite ways of distorting a guitar signal.  They Rat that we're looking at clips off the tops and bottoms of the wave, the TS leaves the tops and makes the sides steeper.  The opamp there almost can't clip given reasonable input levels.  I'm pretty sure that increasing the gain in a TS to the levels we have in the Rat would actually cause it to clean up.  You can't compare the distortion sound between the two devices.

But you can combine their properties...