Trying to understand Slew Rate calculations here

[PRR]

Open-loop gain-bandwidth product is not ....

No, that essay discusses Slew Rate. Which keeps getting confused for GBW product.

[ashcat_lt]

Not that I actually know anything, but I think I agree with the OP.  We're talking about limiting the slope of the wave.  Below the point that it actually clips, that slope should be the same as without clipping.  A wave that is trying to go to 100V peak, but gets clipped at 7.5V has the same slope below 7.5V as if it just was allowed to go all the way to 100V without clipping.  It doesn't know it's going to clip until it does.  Before it clips, yes the slew rate limiting will be based on that slope.  The peak level you want to look at is the unclipped peak, and the maximum frequency without slew distortion will depend on that unclipped "target peak" voltage.  It can't be otherwise.  That just doesn't even make sense.  Now, how the slew rate limiting affects it in the clipped region is maybe a different thing, but I don't think that's what they are talking about.

From everything I've seen, GBW is just a single pole low pass.  Essentially linear, not distortion.

[iainpunk]

GBP is not a normal filter, but a gain dependent filter. at higher gains, the break frequency lowers.

what im trying to say is that you are mostly right, in a theoretical way.
in the most extreme situation when the slewrate makes the most impact is when a signal gets the full gain of an opamp, that difference in slope would be heard if the full waveform was able to come out without clipping, but because the opamp clips, we can consider the max gain signal to be a squarewave, the slew rate only has influence of a small portion of the wave all of a sudden, and the amount of slew distortion in this low of a headroom only affects frequencies above our hearing spectrum.

if we were to make an exact LM308, with the same GBP, but with an extremely high slewrate, we wouldn't hear the differences.

cheers

[ashcat_lt]

I didn't think the question was whether or not we would hear a difference, but just plain whether we should calculate the highest undistorted frequency based on the "attempted" peak level or the actual clipped peak level.  It definitely wants to be the "attempted" peak level.

But I do think that a trapezoidal wave at least theoretically can sound noticeably different than an actual square wave.

[iainpunk]

I didn't think the question was whether or not we would hear a difference, but just plain whether we should calculate the highest undistorted frequency based on the "attempted" peak level or the actual clipped peak level.  It definitely wants to be the "attempted" peak level.

But I do think that a trapezoidal wave at least theoretically can sound noticeably different than an actual square wave.

viewing the question that way, yes youre right

yes, a trapezoid does sound vastly different from a square, but the slopes of the trapezoid an LM308 , for instance, can reproduce is so steep that we don't really hear a difference. there are simple circuits out there to limit the slew rate significantly lower than the 0.3v/us of the venerable LM308. im currently researching/experimenting with a slewrate limiting circuit in a comparator fuzz.

cheers

[zbt]

Only one thing in my mind, practically how low can it go?

[Vivek]

there are simple circuits out there to limit the slew rate significantly lower than the 0.3v/us of the venerable LM308.
cheers

Suppose we have 1Vp guitar signal at 1Khz
and we send it through a stage which has a gain of 200
Then we need a Opamp of slew rate better than 1.25 V/μS

Suppose we have 1Vp Guiar signal at 1Khz
and we send it through an Opamp with slew rate 0.3V/μS
The maximum gain without slew rate limitation will be 48x


and at 5Khz
Suppose we have 1Vp guitar signal at 5Khz
and we send it through a stage which has a gain of 200
Then we need a Opamp of slew rate better than 6.4 V/μS

Suppose we have 1Vp Guitar signal at 5Khz
and we send it through an Opamp with slew rate 0.3V/μS
The maximum gain without slew rate limitation will be 9.6x

Hence an Opamp with slew rate 0.3 V/μS will have significant slew rate limitation at guitar frequencies, with normal gains seen in distortion devices.

[zbt]

I try change the 30pF to 100pF and 500pF  maybe too much.

So the LM308 can make it softer the high, but the output is so low.  May be I can combine it with LM741 for tighter low 

I should try first, thank you Sir, Vivek.

[iainpunk]

Only one thing in my mind, practically how low can it go?

with dedicated slew rate limiting circuits, we can go very slow, millivolt per minute kind of ranges if you really want to.

I try change the 30pF to 100pF and 500pF  maybe too much.

So the LM308 can make it softer the high, but the output is so low.  May be I can combine it with LM741 for tighter low 

I should try first, thank you Sir, Vivek.

changing this capacitor does not only make the slew rate slower, it also lowers the Gain Bandwidth Product, diminishing the actual gain at higher frequencies.

@Vivek
although the theoretically huge wave will be impacted and shaped by this slew limitation, we won't hear that in the real world.

in a 9v system, the harmonic content that differs between 20v/us and 0.3v/us lies above 16kHz, so a diode-less rat would indeed benefit slightly from such a slew rate limitation, were it not that most guitar speakers have a treble roll of around 5 to 8 kHz. when we clip this wave with diodes, the difference becomes even smaller, the harmonics that actually differ between the two slew rates are all above the 90kHz region, inaudible.

you can easily simulate a slew rate limiting circuit and look at the spectrum.

i simulated a 200Vp sine wave, being clipped at + and - 10v, (to allow the impact of SR limiting to be the most it can be, while still being a real-world example) and run through a slewrate limiter that limits slew to 0.1v/us, and plot the spectrum both between the clipper and the SR limiter and after the SR limiter. note that there are 2 plots in this graph, although you only see one due to them being the same in the audible range.

edit: graph is wrong due to falstad not being quite working the way i want, still; heres the difference:
(i shifted them up and down to best showcase the difference)

:10v 0.3v/us


:20v 0.1v/us


you see in the most extreme version, that the difference in frequency response only really starts at 7kHz, and an extremely common guitar speaker, the V30, already has dropped off 20dB at that point. other speakers are often even darker than the V30, diminishing the effect even more

cheers