Slew Rate Filter?

[gcme93]

Hi all this is jus a quick thing that got me distracted whilst I was trying to revise:

Slew rate of an op amp basically defines how quickly the output of the Op amp can vary its voltage. This means an op amp can't respond to frequencies that are too high if they have high am

[brett]

Hi
the slew rate of modern op-amps is very fast. Some might say 'TOO fast' for audio.
It's quite popular in -ve feedback op-amp circuits to add a capacitor to the feedback path, which increases feedback at high frequencies and limits the slew rate.
The tubescreamer circuit has one (51pF I think). At a guess, I'd say it reduces the roll-off frequency and maximum slew rate by a factor of about 100.
cheers

[Keppy]

I recently built a dual-band distortion, and I noticed that on the high frequency band the diodes I used in the feedback loop had little effect. In fact, I had to use BAT41 Schottky diodes to notice much difference vs. no diodes at all. I believe bandwidth becomes more limited as gain increases, due to slew-rate related effects, but I was surprised to discover just how much distortion I got on high frequencies before the signal got anywhere near the opamp rails.

[gcme93]

I'm really not going to try and post off a phone again (without the actual app) as this post definitely did not work!

My original plan was to ask a LOT of gain from an LM741 (slew rate 0.5V) to boost the signal up to 12V P-P. This would then cause the op amp to be unable to respond to frequencies greater than 2kHz leaving me with (hopefully but almost certainly not) a nice smooth and dynamic low pass filter?

I guess it's a silly amount of gain for noise and obviously unusable until it's taken back down (basically a very power hungry active filter) but does it sound like a fair thing to experiment with?

Thanks again for responses so far that are impressively relevant despite only half my post appearing! 

George

[Mark Hammer]

Slew rate is certainly related to bandwidth, but the more relevant and informative spec, when it comes to distortion, is usually the open-loop gain or "gain bandwidth product". 

And yes, your instincts are pretty good on this one.  Some op-amps provide more options for the strategy you want to deploy.  The textbook case if the Proco Rat and its imitators.  These all use an uncompensated op-amp, like the LM308.  The gain-bandwidth product can be tinkered with by means of the compensation cap value, such that the "filtering" you want will occur at lower points in the spectrum.  Part of the magic of the Rat is that it is set up for ridiculous amounts of gain that far exceed the capabilities of the chip with 9VDC, and deliberately reduces those capabilities by means of a compensation cap.

Like any sort of filtering, remember that the gain-bandwidth product has a relatively shallow slope.  So yes, gain capability may be "down" 6 or 10db at 4khz (as an example), but there is still HF content , even if at a lower amplitude.  So do not have unrealistic expectations about how much filtering the limitations of the chip can provide you.  It mae still be wise to supplement it with a LPF-type tone control to keep things fizz-free.

[amptramp]

Slew rate limiting has been used in some communications receivers to eliminate pulse noise from ignition systems etc.  They switch the limiter on by adding a large compensation capacitor to an LM301 or LM308 audio amplifier stage so that it cannot follow the large slew of impulse noise.  It is said to sound better than diode clipping.  Slew rate limiting starts from the high frequency end of the spectrum since the slew rate depends on amplitude and frequency.

[WaveshapeIllusions]

There is always the fun idea of making a discrete opamp. The Miller Capacitor from collector to base on the Class A driver transistor has a strong effect on the gain bandwith and slew rate. Deliberately increasing it will have the effect you are talking about.

Another, simpler, method would would be to use an externally compensated opamp. You would just use a larger than necessary compensation cap. It would be about the same as the previous method.

An integrating setup would be similar as well. That would just be a cap in the feedback loop. I don't think that it would be the same as slew distortion though. I believe it would not vary based on signal voltage; a higher excursion signal would have the same cutoff as a lower one. With slew distortion lower voltage signals will have a higher cutoff.

[amptramp]

The difference between a cap in a feedback loop and slew rate limiting is that you generally do not get up to the output current limit in most amplifiers whereas the compensation capacitor works with a current source.  Using CV = Q = IT where:

C is capacitance
V is voltage
Q is charge
I is current
T is time

you can determine from the current in the current source, the change in voltage for a certain duration and voltage change.  The equation can be rearranged to:

V/T = I/C

where V/T is the slew rate, I is the current in the internal current source in the op amp and C is the compensation capacitor.

[gcme93]

This is such useful stuff!

Thanks a lot everyone, I love the idea of having a compensated op amp that you can deliberately reduce the slew rate of and that discrete op amp idea is one of those things I feel almost obliged to try out - we take a simple easy op amp for granted, so I feel I should know what's going on inside it a lot better. If that means I can then adjust the slew rate and other fun things that would be great. Maybe I could make the op amp have some trimmers to play around with the different parameters and get a more tweakable op amp (though there's scope for massive headaches if that doesn't work. I'll limit the varying to one or two parts)

I'll keep on with the revision but might focus slightly more on gain bandwidth products, and semiconductor circuits

[brett]

Hi again
if you are thinking about a discrete op-amp, you might want to consider using one of the boost/clip sections in the Big Muff Pi. Those single transistors act like very basic -ve input op-amps. ie the gain is the ratio of feedback resistor to input resistance and capacitors in the feedback loop reduce the gain-frequency product. The large value capacitors in BMPs trim high frequencies severly (fc~2kHz??), producing the creamy tones that BMPs are known for. Very cool.
cheers

[gcme93]

Thanks Brett, you've hit on another idea that I was considering based off this chart:



It's so perfect for a beginner with a bit of electronics knowledge like myself, as it dissects each stage so well. I guess from here I'll have a play around with both these ideas - discrete opamp (or BMP circuitry) and compensating op amps.