slew rate and playing dynamics

[Steben]

It seems a lower slew rate acts as a low-pass filter.
Yet, I don't seem to understand the relation between the slew rate and the input amplitude. Bandwith is always connected to gain in literature.
AFAIK slew rate is the capacity to change the voltage. Playing the guitar slower will make for more flat rises in the signal wave. So it seems to me, the softer you play, the less treble cut, which is something different to the set-bandwith style in theory, no?

[alanlan]

As I see it, slew rate is the maximum rate of change available on the output.  So, if you can achieve 1V/uSec, then you could output a nice triangle wave 1Vpk-pk at 500KHz, however, you would only be able to achieve 50KHz with a 10Vpk-pk triangle wave.  So although a high slew rate op-amp is likely to have higher bandwidth (or gain-bandwidth product), slew rate impinges on bandwidth at a given output level.

[Steben]

true.

yet I was talking about dynamincs. If you hit the guitar in a full E chord and the gain is at max, you could have a real loss of treble as long as the slew rate is small enough.
If you play gentle, though, the wave is far less steep and there is less slew rate needed to follow. This way, comparing small signal lows and highs there will be a point where both rise speeds are below the slew rate, resulting in no loss at all. Backing down volume and playing dynamics would give very creamy feel.
I guess the succes of the rat (LM308 chip) and "old" opamps lies in this, not in an "allround" treble loss.

By the way: how to control slew rate of a circuit/IC? What determines it?

[Mark Hammer]

Whenever I have attempted the math, the inherent slew rate of op-amps tends to make little practical difference, even amongst the worst cases like 0.5V/usec.  If anything, the most common instances where slew rate matters is when limitations actually work to our advantage.  For instance, the Proco Rat deliberately uses a chip with limited slew rate and then imposes further restrictions on its gain/bandwidth product.

So, why do I say this?  Well, you need to ask yourself what slew rate IS and DOES first.  It is an index of what you might call "acceleration".  In other words, how much instantaneous voltage swing is this chip capable of producing in a given period of time?  Note that this is not simply how much voltage swing it can produce, but how quickly it can produce that swing.  So, if you are seeking the maximum possible voltage swing, then as long as you are seeking it for frequencies lower than that assumed by the measure (and 1 microsecond to achieve a peak is a very high frequency, as has been noted by alanlan), you'll be able to achieve it.  If you make your voltage swing needs more modest (by designing for less gain per device), then you can achieve the needed voltage swing for frequencies even higher than that.

So, this prompts the question "How much voltage swing do guitarists typically need, and for what frequencies?".  The answer is that we typically are only interested in a fairly modest range of frequencies (content above 12khz tends to be of much less need/interest/relevance unless playing electrified acoustic), and unless it is a distortion that demands high gain, only a fairly modest bit of voltage swing.  ironically, in those instances where we do want  lots of gain (distortion circuits or envelope-detector circuits), we tend to find too much upper frequency content irritating and will use a number of deliberate (filtering, averaging caps) and incidental (speaker bandwidth limitations, slower LDRs) means of keeping such content to a minimum.

Does this mean that slew doesn't matter?  No.  A big part of the audiophile interest in slew rate really started to emerge when op-amp chips started to become the basis of things like phono cartridge preamps.  There, one had a need for a chip that could aim for lots of instantaneous voltage swing at fairly high frequencies, so as to take the teensy signal from a cartridge (MUCH lower than even a weak single-coil pickup, and even lower if it was a moving coil, rather than moving-magnet, cartridge) and boost it plenty (usually 40-60db) so as to assure both optimum S/N ratio and make sure that the full frequency range found its way to the power amp and speakers (which, if they were good, would have no problem reproducing content well out to 25khz).  Note that this context is characterized by two principal needs: much wider bandwidth than guitar or bass, much higher average gain requirements per-device at input.  So, if you are building a mixer or mic preamp, certainly pay attention to slew rate specs.  There are probably other very rare instances, but these are the most typical.  If you are building a fuzzbox or chorus, or EQ, etc., you can feel VERY safe living with 1V/sec and often even less. 

Note that better slew-rate specs often tend to come with other beneficial things, which is why I think we tend to get side-tracked by slew-rate.  For example, many bi-fet op-amps have much better slew rate specs, but the real reason we want them is because they tend to have much higher input impedances, and lower noise specs, both of which have a much greater impact on our sound than slew rate does...in the typical application.

[Steben]

Mark, thx for the input.
I still haven't got a yes or a no, concerning playing loud or gentle.
Do the limitations follow the gain (where gain is like a reversed treble control), or do they follow the specific signal, processed by the gain.
I would think the latter. If that is, the advantage of treble cut is only there with the mix of large signals and high gain. So, with the effect's level on full, rolling back the volume could still make a flat booster.

[Mark Hammer]

You are correct in assuming that there is a distinction between how much gain the circuit itself is set for, and how much voltage swing is "demanded" from the op-amp by the current input signal.  As most of us here can testify, feed a very high-gain circuit with very soft picking and you hear no clipping.  Similarly, feed a circuit set for very modest gain with a powerful input signal and you will get clipping even though the circuit possesses no clipping diodes etc.  So, signal level matters, just as much as gain-setting does.

The larger question, though, is whether it will be "flatter" if a very modest signal is used.  The answer to this, I think, is found more in the 9v battery than in any properties of the chip itself.  In other words, clipping will set in long before any slew rate limitations do, simply because the circuit itself is not permitted to swing to its full voltage anyway.  If you were powering an op-amp with its recommended optimum supply voltage (which in many cases is +/-15v, well over 3x what a 9v battery provides), then maybe it would be profitable to ask the question you are asking.  In the land of 9v batteries, though, slew-rate and even input signal amplitude matters much less.

[Steben]

Aha! Thx.

Do I smell a call to 2x9V supply?

[Mark Hammer]

For some things, yes you are correct, a bipolar supply of +/-9v or even +/-15v would be nice...if your goal is clear sound.  If your goal is clipping of some kind, though, sometimes starving the circuit in some manner ends up getting you what you want.  Remember that distortion is quintessentially an inability of the circuit to deliver a linear response.  Consequently, distortion circuits are deliberately designed to place them at a disadvantage as far as meeting the goal of linearity.  That disadvantage can include requesting more voltage swing of the chip than is feasible given the supply voltage and/or current.  Use of a 9v battery is bad enough, but some designs will place the device under an even greater disadvantage by restricting current, and many have found that some fuzzes sound "better" when used with a battery that is weak or a power supply that tries to simulate such a battery.

If you are building an EQ or a "clean" booster, or even a compressor, a higher supply voltage would probably be a nice thing to have.  The principal reason why you don't see this in products is because a bipolar supply tends to be a bit of a nuisance to implement.  You DO save some parts costs, but those savings are probably offset by the need to allot more space for a second 9V battery, resulting in a bigger chassis, higher shipping/packaging costs, etc.  Now, it IS possible to forego that and have pedals that use an AC wallwart, with on-board conversion from AC to bi-polar DC with regulation.  Many commercial pedals do this, and the space required for such power-conditioning is minimal, relative to the many advantages it provides.  But the allure of being able to stick a 9v battery in there is too strong a temptation to resist, particularly when so many pedals that might benefit from higher supply voltages can still "work" with a 9v battery.

[Steben]

 
Yeah, sometimes you want it both, no?
+/- 9V would be a compromise. Larger headroom, steeper signal rise, but still in distortion range...
I just did some graphs. slew rate of 0.3V/µs gives 300V/ms (LM308, LM10, OP07). If you draw that, you see that just below 10 kHz (1s/10000) the +/- 9V sinus wave crosses the slew line. Just before clipping! Not bad as a quick exercise. So clean you would have almost to 10kHz bandwith. Once the opamp starts clipping, the bandwith goes down. Very nice, especially if you use post-IC clipping zeners around 8V.

[JDoyle]

Yeah, sometimes you want it both, no?
+/- 9V would be a compromise. Larger headroom, steeper signal rise, but still in distortion range...
I just did some graphs. slew rate of 0.3V/µs gives 300V/ms (LM308, LM10, OP07). If you draw that, you see that just below 10 kHz (1s/10000) the +/- 9V sinus wave crosses the slew line. Just before clipping! Not bad as a quick exercise. So clean you would have almost to 10kHz bandwith. Once the opamp starts clipping, the bandwith goes down. Very nice, especially if you use post-IC clipping zeners around 8V.

10kHz is WAY up in the hearing range - and about 2x beyond the highest output frequency of an electric guitar, including most of the harmonics, not to mention: the frequency response of guitar+effects+cables+amp, especially the amp, which rolls of above 5k anyway, 10kHz is asking a lot out of that mess. You will be hard pressed to hear it, and to do so you'll need to crank the treble on your amp, and even then...Plus, when an opamp is clipping it screws everything, including the compensation network, so it's not really possible to clip and slew limit at the same time. Remember that slewing isn't level dependent, but frequency dependent, and vice versa for clipping.

...as an aside, one of the quick and easy ways to get really close to what your are suggesting is to take a LM324 quad op amp and pass your signal through four of the amp stages in series, all four set up as non-inverting buffers - the noninverting buffer puts the highest demands on any opamp's slewing ability.

The downside of all of this? Not only do you get slew limiting, which could be cool on it's own, but when an opamp slew limits it means that the input differential amplifier of the op amp can't supply the needed current to the internal compensation cap around the voltage amplifier stage - the cap is used to maintain the needed compensation to keep the opamp stable; this is how the Rat does it - the compensation cap is external and the designers 'over-compensated' the op amp, bringing the slew limit down to within the audio range. So one half of the diff amp is current limited, but not 'clipping', the current is getting sucked up by the cap, so the negative input still 'works' as a mixing point, though now the feedback network is compromised and so is the balance of the input diff pair - 'cats and dogs living together, mass hysteria!', sorry quick movie quote...

The problem with slew limiting is it introduces what used to be called Transient Intermodulation Distortion (now it is just lumped in with slew limiting, because, well, that's exactly what it is), which is a mixing of the output's distortion products with the input signal (at the inverting input of the opamp) which creates harmonics that are completely unrelated, musically at least, to the input signal.

I don't think it sounds good, and the 'HiFi' crowd goes to amazing lengths to avoid it; and I guess that's why I never really liked the Rat - but hey, you might, but to hear it I think you will need to shoot a little lower in frequency. And you'll need a single opamp with compensation pins to do it.

Regards,

Jay Doyle

[iainpunk]

sorry for this zombie thread revival.

i was wondering if there are distortion pedals / projects with deliberate slew rate limiting, like with a slew rate limiter circuit pictured below?


cheers

[idiot savant]

I seem to remember way back in the day Tim Escobedo was playing around with slewing distortion.

I would look at old "programmable" opamps like the LM4250, or LM776. They use a current set resistor that directly affects slew rate and GBP.

[Rob Strand]

For circuits you have to search through the archives.  It has come up a few times.

I would look at old "programmable" opamps like the LM4250, or LM776. They use a current set resistor that directly affects slew rate and GBP.

Another angle is to use uncompensated opamps with frequency compensation pins and use an excessively large caps.

In principle it's not too different iainpunk's schematic.  Here the first opamp saturates and ramps
the integrator up with a current Vsat / R3.  Whereas in the uncompensated opamp case the opamp's first stage
differential-pair provides the "integrator" current.

The advantage of the programmable opamps and iainpunk's schematic is the ramp can be adjusted with a resistor whereas the uncompensated opamp you have to switch the caps.

A diff amp driving a cap or an OTA driving a cap are some other options, all conceptually similar to the above.

[287m]

That drawing look like timmy with feedback?

[amptramp]

One of the items you will find on a communications receiver is a limiter that usually uses a diode biased to a fraction of the average signal so that any appreciable excursion beyond the input is clipped.  But some recent designs use an op amp with an external compensation capacitor so that large spikes (like lightning strikes superimposed on the signal) will only be responded to at the slew rate which is deliberately reduced.

Slew rate limits will make the output follow a straight ramp because you have a current source feeding a capacitor and CV = IT = Q where:

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

[iainpunk]

That drawing look like timmy with feedback?

yes and no.
timmy's variable resistor determines a filter, this changes the slew rate. the inputs are flipped on the 2nd opamp over and no clipping diodes, the 2nd opamp is an integrator instead of a gain stage, and so much more is different in function, despite the schematics looking alike.

I seem to remember way back in the day Tim Escobedo was playing around with slewing distortion.

I would look at old "programmable" opamps like the LM4250, or LM776. They use a current set resistor that directly affects slew rate and GBP.

can't find those opamps for less than 1,80 per IC.
the distortion i'm working on already has a specific opamp choice due to its exploitable nature, so adding another normal opamp is the easiest option, ill probably use a double, so i can replace the transistor buffer for less real estate use.

cheers

[Mark Hammer]

Without wishing to reread the many interminably long posts I had in this thread, I will simply note that gain-bandwidth product is the more important and relevant spec when it comes to drive pedals.

Where a drive pedal uses the feedback resistance of an op-amp to determine gain, that feedback loop often has a small-value capacitor that will restrict bandwidth as gain is increased.  If the op-amp itself has a modest gain-bandwidth product (AKA "open loop gain"), its own limitations in providing higher-frequency content at the gains aimed for adds another layer of lowpass filtering to sweeten the result.

But for all those contexts where one isn't attempting gains of 10x or more, slew rate has no discernible impact.

[anotherjim]

Programmable opamps seem to be something the IC industry doesn't think we need anymore. After all, there are plenty of micropower chips around that don't need the extra Iset input to lower power consumption. Using the Iset to control frequency response isn't something they were meant for although it was the tuning method in the Soviet era Polivoks synths.
Findable/affordable POA chips may be flat-pack/SMD only. The TS271 CMOS example might be an interesting one to experiment with.

[moid]

sorry for this zombie thread revival.

i was wondering if there are distortion pedals / projects with deliberate slew rate limiting, like with a slew rate limiter circuit pictured below?
[/img]

cheers

I might be wrong about the below, because I'm not entirely sure if I understand what slew rate is (I think it means how long it will take some audio to be amplified?) but I once built two modified Seppuku FX octave drones and, in the spirit of wild carefree fun, plugged them into each other and let my son go nuts on them with a guitar. At two points in the below video, the settings I found made the guitar audio fade out to sputtering crackles as he was playing really hard and as he either stopped playing to let the notes ring out, or he played more gently, the guitar audio faded back in to overwhelm the background noise. You can hear those bits at 3:14 and about 5:00. I warn you this is loud and mostly quite horrible noise, so don't watch all of it! (one pedal was a commission for a friend who wanted something utterly horrible that he could use to terrify an audience with - so I built one for myself as well because I wanted to find out if I liked that sort of thing; I still have both of them - he liked them so much he wanted two of them in one enclosure!)

If this effect is what you are talking about, I'd love to build a circuit that had an adjustable slew rate - so I could set how long it takes the audio to fade in (a bit like an EHX Attack Decay I guess?). It would be even more awesome if it could fade bass notes and treble notes in at different rates, like a filter I guess? And possibly not so incredibly distorted!

I don't have a schematic for my mod, I think I took someone else's vero layout and redrew my version from that. Mine is below if that helps.




The person who runs this site https://effectslayouts.blogspot.com/2015/04/seppuku-octave-drone.html says that this pedal is based on Tim Escobedo's Rambler if that's of any use to explore.

[iainpunk]

slew rate has to do with the waves, not the volume level.
what slew rate determines is the speed the output voltage can rise or fall. if you have a square wave, and you were to lower the slew rate of that signal, it would turn into a trapezoid wave. a low pass filter is somewhat similar, but has rounded edges, while limited slew rate still gives hard straight lines, it gives a filtering effect without it being to dark all of a sudden, and it chages the character.

i tried to put a slew rate limiter together today, but it has horrible oscillations. i guess i will find the source tomorrow, as i was interrupted today.

cheers