Slew Rat

[Vivek]

I want to do a study on LTSPICE re effect of slew rate in the Rat circuit

I will make one schematic with slow slew rate Opamp and another with high speed Opamp, and try to analyse the differences.

So far, I thought these analyses could be useful:

A) Frequency response
(which will be part of the story since slow slew rate Opamp will change Wave shape along with Wave Amplitude)

B) Waveshape at few different frequencies and amplitudes

Something like:

For slow and fast slew rate Opamp,
   for 500, 1K, 8k frequency
       for 100, 500, 1000 mvp input
             Analyse
       next
   next
next

C) FFT analysis at few different freq/amplitude combinations to find out effect of slew rate

slow Slew rate will increase harmonics
but decrease amplitude
so how do these two contradictory forces interact ?


D) Does final hard clipping remove any goodness created by the earlier stages ? Does any slewiness show through after hard clipping ?



Please propose any other tests that could be meaningful.



They are already a slew of Rat analyses on the web, for example, ELECTROSMASH really slew it !!

but I feel above (critical) questions still remain unanswered. I hope that DIY community are slewed and slewed by these new analyses.

[amz-fx]

A) Frequency response
(which will be part of the story since slow slew rate Opamp will change Wave shape along with Wave Amplitude)

Slew limiting will produce a low pass filter effect.

B) Waveshape at few different frequencies and amplitudes

Something like:

For slow and fast slew rate Opamp,
   for 500, 1K, 8k frequency
       for 100, 500, 1000 mvp input
             Analyse
       next
   next
next

A guitar cannot make an 8k Hz signal, and even the harmonics of a high fret string will be very llow at 8k.  The 20th fret of the high E string is C6, which is 1046.5 Hz. (barely above 1k). The harmonics of the high note are quite small.

C) FFT analysis at few different freq/amplitude combinations to find out effect of slew rate

slow Slew rate will increase harmonics
but decrease amplitude
so how do these two contradictory forces interact ?

The questions is: does it happen to a real signal from a  string with the highest pitch barely above 1kHz?

D) Does final hard clipping remove any goodness created by the earlier stages ? Does any slewiness show through after hard clipping ?

The opamp signal is going to be about 8v p-p and will be hard clipped by the grounded diodes. I suspect that not much from the previous stage will impact the final sound. A way to test this is to run the pedal on 18v and see if there is a difference from the one powered by 9v.

Best regards, Jack

[Vivek]

Thanks for your comments and guidance, Jack !!!

[Vivek]

@Rob Strand,

What is your gut feel ? Will hard clipping grounding diodes remove any effect of Slew rate of earlier Opamp ?

[amptramp]

One test would be to increase the value of C8, the compensation capacitor, from the 30 pF maximum required value for unity-gain stability to higher values which would slow the lethargic LM308 slew rate down even further.  There are two other compensation schemes for the LM308 with one using feedforward (if you want to increase the slew rate) and another with a 100 pF cap to ground from pin 8 that improves rejection of power supply noise.

The effect of feeding in a signal that is fast enough to keep the op amp slew-rate limited is to generate a lot of triangle waves.  Since a triangle wave can be considered to be a square wave subjected to mathematical integration, the Fourier analysis shows the harmonics to be all odd like that of a square wave but instead of the amplitude being the reciprocal of the order of the harmonic, it is the reciprocal of the square of the harmonic, in other words where a square wave has the fundamental, 1/3 of the third harmonic, 1/5 of the fifth harmonic etc., the triangular wave has 1/9 of the third harmonic, 1/25 of the fifth harmonic etc.

One interesting use of an overcompensated LM308 was as a noise limiter for impulse noise in ham radio receivers.  To get limiting, an larger compensation capacitor was switched in in parallel with the normal one.  This meant a normal spike would not be clipped at some multiple of the average signal as with usual limiters but would generate a rising waveform that wouldn't get very far before the noise stopped and the amplifier went back to following the audio input.

[Rob Strand]

@Rob Strand,

What is your gut feel ? Will hard clipping grounding diodes remove any effect of Slew rate of earlier Opamp ?

It can't change the part of the waveform between -Vd and +Vd.

This old trick of loading the opamp works.   A first glance you would think the low
impedance load is limiting the swing like the diodes.   Well it does.  But the idea
is it doesn't swing so far.  Here the voltage gain of the transistors boosts the voltage
and the larger swing at the output is the point which has the higher slew rate.
The opamp output is still limited it's just the swing is now less.

[ElectricDruid]

The effect of feeding in a signal that is fast enough to keep the op amp slew-rate limited is to generate a lot of triangle waves.  Since a triangle wave can be considered to be a square wave subjected to mathematical integration, the Fourier analysis shows the harmonics to be all odd like that of a square wave but instead of the amplitude being the reciprocal of the order of the harmonic, it is the reciprocal of the square of the harmonic, in other words where a square wave has the fundamental, 1/3 of the third harmonic, 1/5 of the fifth harmonic etc., the triangular wave has 1/9 of the third harmonic, 1/25 of the fifth harmonic etc.

One other difference between triangle waves and square waves beyond the differing roll-off of the harmonics is that the phases are different. The harmonics of square waves are all the same phase. The harmonics of triangles switch back and forth between 0 and 180 degrees.
I have no idea how that plays out in something like the RAT, but I thought it was worth mentioning!

[Steben]

The most intriguing effect of slew rate in theory is dynamic low pass. But as we have settled in other threads this effect is limited by the impossible dynamic range of the player. Which renders any effect less dynamic. In other words: in the best case we have a situation comparable to a fuzz face.

[iainpunk]

i have experimented with slewrate limiters in the past. they are great for dynamic filtering, but they kind of take the edge away from the attack. my experiments did involve an expander up front and a compressor after, this exaggerated the effect of the waveshaping and filtering even more. my general take away from those experiments is that its mainly usefull on bass.

cheers

[Vivek]

its mainly usefull on bass.

I will add 80 Hz to the list of test frequencies.

[iainpunk]

its mainly usefull on bass.

I will add 80 Hz to the list of test frequencies.

how about 41 and a quarter Hz for the low E

cheers

[Vivek]

Done Iain !

The first step will be to make meaningful frequency response curves, since LTSPICE handles nonlinear components in an interesting way during these calculations,

Plus, It can be meaningless to plot Vp at different frequencies when the wave changes shape at each of these different frequencies.

Any ideas on how to present meaningful frequency response curves for nonlinear, dynamic systems ?

[iainpunk]

Any ideas on how to present meaningful frequency response curves for nonlinear, dynamic systems ?

3d graph. like one of those audio visualisation graphs using faux heatmaps but instead of having music/audio as input, you sweep a sine through that spectrum. this gives both a view of fundamental treatment as well as insight over which frequencies distort more.

[Digital Larry]

3d graph. like one of those audio visualisation graphs using faux heatmaps but instead of having music/audio as input, you sweep a sine through that spectrum. this gives both a view of fundamental treatment as well as insight over which frequencies distort more.

I'd suggest something similar, but instead of sweeping the frequency, show what happens to the spectrum for e.g. a sine wave with an exponential decay.  I think this would be great for anything whose behavior depends on the signal level.  Not talking about "everything" which will clip when driven hard enough.