Slew Rate Limiter

Does anybody have a circuit hanging around for a slew rate limiter? I'm looking for something that I can control the maximum slew rate via a pot. I found this in Hickman's Analog and RF Circuits, but once it heads into max-slew, the circuit settles exponentially and thus tends to turn everything above the maximum slew rate into a triangle wave. Anybody else have anything?

It seems there was a thread here a while back on exactly this subject.

There are several ways to make a slew rate limiter, but turning faster waveforms into triangle waves is exactly what slew limiting does. What kind  of output are you expecting to get?

I guess a VCA based comp, could detect slew rate in the side chain and reduce gain when slew rate exceeds a threshold level. This way instead of a triangle wave you would get progressively smaller amplitude sine waves... kind of like a hard limiter but with progressively smaller thresholds with  rising frequency. 

JR

In the 80's, I had a unit on the market named SRL (Slew Rate Limiter), aimed at the vinyl mastering market, which was exactly what JR describes (except it was FET based). Just a standard 1176-type FET compressor with a degenerated differentiator in the side-chain. Since it is aimed at processing transients, you have to make sure it has a fast enough attack time.

It strikes me that I was really thinking about the effect of slew rate limiting on a wave somewhat wrongly. I was thinking that when the direction of the wave changes, the resulting peak of the triangle wave switches direction so quickly. Somehow I had equated in my mind an instantaneous switch from a positive trending voltage to a negative trending one as too large a change, and that the resulting slew-rate limited wave should track the original wave a little more. But I wasn't really thinking about it in terms of total V/us, which is still 0 at the point where the wave switches direction.

A circuit that lowers the gain of the wave as it approaches faster slewing sounds just like a lowpass to me. Is there any practical difference I'm missing?

Nishmaster said:

A circuit that lowers the gain of the wave as it approaches faster slewing sounds just like a lowpass to me. Is there any practical difference I'm missing?

Click to expand...

Yes, it has a threshold for actuation that is neither pure amplitude or pure frequency but the product of the two.

One still must ask if it has much utility these post vinyl days. I will defer to abbey road d enfer, who has experience with one. Perhaps as a slightly smarter de-esser?

JR

Ah, yes. Missed that.

Well, the utility that I intend it for is even less glorious. I had an idea for a guitar fuzz pedal that would eschew the usual "tone" circuit (usually just a low-pass filter) and replace it with a slew rate limiting circuit. The idea would be that the usual ugly upper frequencies would be attenuated by the slew limiter, and as a bonus, the harder you played and the more you clipped, the lower the high frequencies are in comparison to the original signal.

Haven't mocked it up yet, so I don't know if this approach smooths out the usual hi-gain sound any (I hate that high frequency hash that shows up on most pedals). I'm sure turning sines into triangles doesn't decrease harmonics any, but hopefully the amplitude will be diminished enough.

We'll see. This is all a part of my new custom pedalboard experiment. I know guitar effects are somewhat low-brow for this particular forum, but you guys are the best, so bear with me while I poke and prod about various goofy looking circuit fragments and ideas.

-Matt

> guitar fuzz pedal

High-boost, clip, high-cut.

This will tend to want op-amp differentiator and integrator.

I've done this across a 400Hz-4KHz range on speech. The "hi-boost" was really just a too-too-short coupling cap ("bass-rolloff" near 4KHz) and added gain. Clipping was a 0.3W amp. Hi-cut was a wonky horn-speaker's throat-loss. The output is unnatural but very intelligible, and not grating.

This is in fact slew-rate limiting. Clipping voltage times integrator is the slew-limit.

Nishmaster said:

Ah, yes. Missed that.

Well, the utility that I intend it for is even less glorious. I had an idea for a guitar fuzz pedal that would eschew the usual "tone" circuit (usually just a low-pass filter) and replace it with a slew rate limiting circuit. The idea would be that the usual ugly upper frequencies would be attenuated by the slew limiter, and as a bonus, the harder you played and the more you clipped, the lower the high frequencies are in comparison to the original signal.

Haven't mocked it up yet, so I don't know if this approach smooths out the usual hi-gain sound any (I hate that high frequency hash that shows up on most pedals). I'm sure turning sines into triangles doesn't decrease harmonics any, but hopefully the amplitude will be diminished enough.

We'll see. This is all a part of my new custom pedalboard experiment. I know guitar effects are somewhat low-brow for this particular forum, but you guys are the best, so bear with me while I poke and prod about various goofy looking circuit fragments and ideas.

-Matt

Click to expand...

While filtering the fuzz out of a fuzz circuit may leave you with only....  clean signal?

Seriously one of my old patents (US05509080 Roberts) may be worth playing around with.

For a completely different reason (preventing transformers from saturating at LF),  I combined a simple diode clipper with a Baxandal tone control. By putting the diode clipper across just the bass boost part of the tone control circuit, I was able to clamp just the loud bass sounds, without the nasty "fuzz" of a full range clipper.

For your application, some variant on this where you effectively split the frequency band, and just fuzz (clip) the low frequency while passing the HF unmolested.

JR

PS: The patent issued in '96 so is still in effect (assigned to Peavey) , but I doubt making a fuzz effect derived from the concept would be considered infringement, especially since you would need to alter the transition frequency to something more useful, and wouldn't use boost/cut pots, etc. and it might not even work...

PPS: A variation on PRRs pre/de-emphasis trick could have the effect of de-emphasizing the clipper's HF distortion, without altering the response of the unclipped input. Kind of like EQ just on the distortion.

> de-emphasizing the clipper's HF distortion

You get a good lump of low-order distortion, both harmonic and intermodulation, without the tizzy high-order products.

And as the guitar decays, the distortion decays faster, an important part of e-gitar timbre.

Is it musical? I have doubts. Is it cheap? Somewhere between a one-BJT fizzbox and a two-4558 fuzzbox, cheap enuff to try.

While I sketched the differentiate/integrate as "ideal" (necessary for best fidelity), I suspect some shaping is wanted. Maybe let more/less <180Hz through so the lowest octave does/doesn't IM into the higher voice. Maybe not de-emphasize above 1.6KHz, let some tizz through. A job for someone with a good ear and a knack for 2-second 10-cent filter networks.

Nishmaster said:

A circuit that lowers the gain of the wave as it approaches faster slewing sounds just like a lowpass to me. Is there any practical difference I'm missing?

Click to expand...

If you draw the transfer function at a given level, it will look similar to a lopass, but the signal will be distorted in the lopass and clean in a slew-rate limiter, and obviously the impact on dynamics will be a major one. The transfer curve would be level dependent for the slew-rate limiter and non-dependent for the lopass.
I think you should experiment with a limiter with an agressive EQ in the side-chain, or even a preemphasis EQ at the input of a standard limiter and deemphasis at the output. That should bring you close to the famous clarinet sound of Brian May in Bohemian rhapsody, which was accomplished mainly with a treble boost, a compressor and a fuzz with the tone control turned down (and a couple hundred pieces of outboard equipment, it seems).

JohnRoberts said:

One still must ask if it has much utility these post vinyl days. I will defer to abbey road d enfer, who has experience with one. Perhaps as a slightly smarter de-esser?

Click to expand...

It's not post vinyl around here.

Hey Abbey, do you have one of those things laying around? I'd love to check it out.

Unfortunately no; these things were made 25 years ago and I'm not one who collects. But it was just standard FET limiter; only the side-chain was a little different.

I like the idea of de-emphasis of the high frequencies before clipping, and then boosting after.

How could I take the frequency response of the integrator and completely correct for it, though? Putting a differentiator after it will just give me another pole at the opposite end of the frequency spectrum, right?

Funnily enough, I had been looking into this as well.
It strikes me that a decent slew rate is not always a desirable thing and in fact a low slew might be advantageous in certain circumstances.
I recorded a lot of vocals with an LA 4 some time ago and was struck by how little 'essiness' there was.
The op-amp used in that particular unit is a 4136 - not very high slew rate.
I was also listening to some Beatles multitracks as part of a recent project and was struck by the same sort of sound where the lack of 'transience' on the vocals actually made them more 'listenable'.
(Not sure if they had used an Altec or an EMI unit - it was the Sgt Pepper sessions, perhaps someone here might know).
Anyway, food for thought!

Nishmaster said:

I like the idea of de-emphasis of the high frequencies before clipping, and then boosting after.

How could I take the frequency response of the integrator and completely correct for it, though? Putting a differentiator after it will just give me another pole at the opposite end of the frequency spectrum, right?

Click to expand...

Actually the concept was pre-emphasis before and de-emphasis after.. to reduce the harshness of the fuzz, but you can try it both ways. Baiscally the original signal passes roughly flat, while the interior circuitry receives 12dB of HF cut (or boost if you reverse the order). 

The circuit network is really simple, two resistors in series with a cap across the larger of of the two resistors. This gives a LF conductance equal to the two resistors in series, and a HF conductance of just the second resistor. The LF knee is the RC of the cap and resistor it's across, the HF knee is the RC formed by the cap and the two resistors in parallel. The amount of HF pre/de-emphasis is the ratio of the smaller resistor to the two resistors in series.  A 3:1 resistor ratio give a useful 12dB of pre/de emphasis. I often tuned for a LF knee around 500 Hz but you may want to tune differently for your application.  Using more than 12dB of pre/de is possible but ultimately this boost comes out of overall HF headroom.

Make two identical R-RC networks and place one in the input leg of an inverting opamp, before your fuzz, then the second in the feedback path of an inverting opamp after the fuzz.  Use precision parts for best tracking, but for a guitar effect response may not be that critical.

JR

John, PRR, gents, always appreciate the feedback.

My thinking with the emphasis order was that if I de-emphasize the highs prior to clipping, I should be less harshly clipping the higher harmonics and hopefully end up with something that sounds smoother by virtue of having harmonics generated that don't extend so high in the band. If I emphasize the highs prior to clipping, I'm really going to be squaring those off. I'll try both, however, see which one has a preferable sound. The de-emphasis/emphasis order may come at a noise penalty.

Or, perhaps, I can really think like an engineer and make it switchable!  8)

barclaycon said:

It strikes me that a decent slew rate is not always a desirable thing and in fact a low slew might be advantageous in certain circumstances.

I was also listening to some Beatles multitracks as part of a recent project and was struck by the same sort of sound where the lack of 'transience' on the vocals actually made them more 'listenable'.

Click to expand...

I think this is one of the "problems" with old recording systems that in hindsight was a benefit. That's why every one wants tubes and transformers before the DAW.

I started a thread a while back asking about a compressor with a "floating threshold". Meaning controlling the crest factor regardless of level. This sounds like it is in the same neighborhood as that. It may be even more useful for disk cutting because it is frequency dependent also. I will re read this thread very carefully. I would like to hear this concept in action.

barclaycon said:

Funnily enough, I had been looking into this as well.
It strikes me that a decent slew rate is not always a desirable thing and in fact a low slew might be advantageous in certain circumstances.
I recorded a lot of vocals with an LA 4 some time ago and was struck by how little 'essiness' there was.
The op-amp used in that particular unit is a 4136 - not very high slew rate.
I was also listening to some Beatles multitracks as part of a recent project and was struck by the same sort of sound where the lack of 'transience' on the vocals actually made them more 'listenable'.
(Not sure if they had used an Altec or an EMI unit - it was the Sgt Pepper sessions, perhaps someone here might know).
Anyway, food for thought!

Click to expand...

The 4136 was not exactly a slow part... 1.7 v/uSec will not slew limit that easily... And the sonic consequence of slew limiting if it does happen is probably IM hash on loud cymbals and such, not a 'softer" vocal sound.

JR

Buy a bag of LM4250, and a bag of 10Meg resistors for I-set switching and construct a "slewing machine".

Report back if it sings or farts...

The 4136 was not exactly a slow part... 1.7 v/uSec will not slew limit that easily... And the sonic consequence of slew limiting if it does happen is probably IM hash on loud cymbals and such, not a 'softer" vocal sound.

Click to expand...

That's your 'sonic' observation from recording and mixing music is it John ?