18 volts, compression, headroom and transients.

[chumbox]

Hi All

Just been schooling myself in 18v, reading all the threads on here and I have noticed a common thread that 18v has little difference on overdrive/distortion due to it's nature to be more compressed already.  I'm just wondering though whether running a drive at 18v and increasing headroom would still 'in theory' let more transients through before clipping them, lifting the level boundary at which the signal gets compressed (whether it's that audible or not)? Or do drive pedals get compressed due to the circuit and the headroom therefore does very little?  From my reading it appears 18v serves little 'obvious' purpose in drive pedals?

I guess this is more of a hi-fi theory question that was rattling around in head but don't get more wrong, if a pedal sounds good, it sounds good 9v or otherwise.  Now that I think of it I should just plug a couple of pedals into pro-tools with 9v and 18v and actually have a look at the transients.

[petemoore]

 Check that all components are rated above the applied supply voltage.
Most circuits designed for distortion use diodes or other fixed voltage signal clipping element, raising or lowering supply voltage will change'where they begin conducting on the waveform, changing it's shape and character.
It can be interesting to find the 'sweet spot' where different is 'novel', and set up an LM317 [so it doesn't overheat,etc.] and so a knob evenly covers that voltage range.

[samhay]

It will depend quite a lot on the type of circuit involved - more than one way to build an overdrive.

Something like a Tubescreamer will benefit from more headroom if you feed it a very hot signal.
Something like a Rat with diodes clipping to ground will have less/no benefit.
The FET-based circuits may or may not benefit, depending on the FETs used and the way they are biased.

[Transmogrifox]

+1 for samhay.  You need to consider specific circuits one at a time to determine whether 18V supply makes any difference.  There will be a lot of rule breakers if you make a generalization.

[garcho]

plug a couple of pedals into pro-tools with 9v and 18v and actually have a look at the transients.

if you can't hear it, why look at it? what sound are you trying to achieve that you're not getting right now?

[ashcat_lt]

Something like a Tubescreamer will benefit from more headroom if you feed it a very hot signal.

Though it will get cleaner and cleaner as you use more and more of that headroom.  By the time you get near 18V peak-to-peak, it'll be in and out of that "crossover" region that matters so quickly you won't much hear it.

Something like a Rat with diodes clipping to ground will have less/no benefit.

A lot of people think that the opamp clipping is important in the sound of a Rat, but the point remains.

I got to thinking at one point like "Why amplify it with 9V rails, so that it swings 7 or so volts, just to smack it back down to about 1V?  Why not just use smaller supply to begin with?"  The answer, of course, is that most of the opamps we use won't work on much less than 9V supply.  But I found that the LM324 will actually work down to 3V.  It will swing all the way down to 0, but will only get to within about diode drop from the top rail, so the total swing is only about 2.2V, a bit less than a pair of LEDS.  If you bias it to 1.5V, you get assymetrical clipping like you'd expect (but might not actually get) from a Si/LED pair.  And as much as PRR disapproves of the use of the LM324 in audio work, it actually sounds pretty damn good.

The FET-based circuits may or may not benefit, depending on the FETs used and the way they are biased.

I'm a bit unsure on this one.  It seems to me like the effective resistance of the S>D channel is affected (only?) by the difference in voltage between the S and G, and that resistance can only get so big or so small, and that doesn't really have much to do with the supply voltage.  Changing supply voltage should (?) change the maximum output voltage swing, but not necessarily where in the input swing it get's clipped.  Does that make sense?  Is it anywhere near correct?  Obviously it depends on bias conditions, etc, but am I just completely wrong on the theory?

[Groovenut]

IMO, unless you're hitting the rails, more voltage wont gain you much, if anything.

[Scruffie]

Something like a Rat with diodes clipping to ground will have less/no benefit.

Disagree with this, there's quite a bit of opamp clipping going on in the RAT, true it does take it in to more standard clipping territory with 18V but it does vary the sound quite a bit by doing so.

[PRR]

> PRR disapproves of the use of the LM324 in audio

If you are deliberately "distorting", I don't give a hoot.

It does add "something extra". Aside from slamming the rails, you get crossover distortion in the middle, rising with frequency. (And with both together, garbage I can't imagine.)

I've been very disappointed in LM324 for "clean" audio. It is possible, but you need luck or deep understanding (and extra parts). As there are less-treacherous chips for audio, I prefer to steer clear.

I love your other points but doubt they will sink-in with everybody.

[Keppy]

Something like a Rat with diodes clipping to ground will have less/no benefit.

Disagree with this, there's quite a bit of opamp clipping going on in the RAT, true it does take it in to more standard clipping territory with 18V but it does vary the sound quite a bit by doing so.

1) I've always been under the impression that the opamp clipping in the RAT (or OCD, or other hard-clipping circuit) is rather limited in its effect once the signal is re-clipped by the diodes that follow.

2) I've also been under the impression that the unique contribution of the 308 chip in the RAT is the distortion cause by its low slew rate, which is distinct from clipping.

3) Does slew rate vary with power supply voltage? I'd think not, but I don't actually know that.

4) Scruffie, you make it sound like you've tried a RAT at 18v. What was the effect?

Please don't take this post as a contradiction or challenge. I'm genuinely curious to see if you have observations that I can learn from. My understanding of theory has been contradicted by observation before and will be again, so your post has piqued my interest.

[Scruffie]

I'm not taking it as a challenge, no worries there (for fun I could pretend to be one of those members who takes questions as a personal affront if it would make you feel more comfortable? ) and i'll admit, it was a long time ago so it could be a false memory so perhaps I shouldn't have disagreed so vehemently but I do recall it had an affect on the sound above switching your usual screamer or DOD 250 to 18V, if I commented on sound it would only be buzz words like 'openess' etc. but I recall it was quite audible and that I felt it clipped less harshly.

There is a LOT of gain in that circuit compared to a standard OD so, my memory is not implausible.

It wasn't using an LM308 though, think it was an NE5534.

On to your first point... the big muff clips twice, the opamp can top and tail the signal and still be clipped again.

Slew rate vs power variation will be a datasheet question.

[PRR]

> Does slew rate vary with power supply voltage?

For most opamps, hardly-any, until supply gets so low it is about to quit working at all.

There are exceptions. But that makes the data-sheet confusing, and chip-sellers like KISS (don't confuse or disappoint chip buyers), so they flog the designers to make most parameters pretty stable, as much as possible.

[Mark Hammer]

It"s helpful to recognize that there can be multiple sources of clipping in a circuit; some of which will correspond to supply voltage, and some of which won't.  Diodes have a fixed forward voltage, such that whether you power the circuit with 6V, 9V or +/-15V, they will conduct at that voltage, and if their location is one that results in clipping, the clipping will begin around that forward voltage, despite whatever headroom is provided to other semiconductors in the circuit by a higher supply voltage.

Some op-amps can be more and less "resilient" to clipping than others, with that characteristic enhanced by changes in supply voltage.  Many folks who have built their very own MXR Distortion+ have noted that it still distorts, even without clipping diodes.  It still distorts more with them than without, but clipping in the chip itself contributes to the overall sound.  However, I would imagine that, whatever the puny 741 chip does would be altered by a different supply voltage.

For a little while, here, there was a spate of JFET-based overdrives, mostly using J201s and 2N5457s, and no diodes at all.  Now in those cases, supply voltage would change the characteristics of the circuit, if only because the biasing set up for a lower voltage would be unsuitable for a higher one.  I'm not knowledgeable enough to know if appropriately rebiasing would yield the identical tone at a higher (or lower) supply voltage, but I suspect the transition zone between sparkling clean and filthy dirty would be a little different.

Finally, it is probably also worth considering that, since clipping results in the addition of harmonics, which are higher-frequency components, the audibility of what supply-voltage changes to a pedal might do can depend onthe lowpass filtering in the circuit, and whatever lowpass filtering one's speakers provide on top of that.

[Frank_NH]

For a little while, here, there was a spate of JFET-based overdrives, mostly using J201s and 2N5457s, and no diodes at all.  Now in those cases, supply voltage would change the characteristics of the circuit, if only because the biasing set up for a lower voltage would be unsuitable for a higher one.  I'm not knowledgeable enough to know if appropriately rebiasing would yield the identical tone at a higher (or lower) supply voltage, but I suspect the transition zone between sparkling clean and filthy dirty would be a little different.

I was wondering about this.  For something like the Fetzer valve FET gain stage, you would want to bias the drain to suit the supply voltage (e.g. ~9V drain voltage for an 18V supply).  But FET performance is dependent on the device characteristics, namely Idss and Vp (pinch off voltage), which are fixed for a given FET.  You can increase the source voltage, but the FET will never provide more current than Idss.  Also, it would seem to me that the input could clip if the amplitude is much greater than Vp.  Again, this is independent of the the supply voltage.  I suppose that the clipping characteristics may be a little different, but if the 9V and 18V versions of a FET gain stage have drains which are biased to 1/2 supply, wouldn't they behave pretty similarly?

[ashcat_lt]

I was wondering about this.  For something like the Fetzer valve FET gain stage, you would want to bias the drain to suit the supply voltage (e.g. ~9V drain voltage for an 18V supply).  But FET performance is dependent on the device characteristics, namely Idss and Vp (pinch off voltage), which are fixed for a given FET.  You can increase the source voltage, but the FET will never provide more current than Idss.  Also, it would seem to me that the input could clip if the amplitude is much greater than Vp.  Again, this is independent of the the supply voltage.  I suppose that the clipping characteristics may be a little different, but if the 9V and 18V versions of a FET gain stage have drains which are biased to 1/2 supply, wouldn't they behave pretty similarly?

That's kinda what I said.  Given that we bias it about the same, I think the actual input>output transfer curve is the same, and power supply is like a constant multiplier after that.

[Frank_NH]

That's kinda what I said.  Given that we bias it about the same, I think the actual input>output transfer curve is the same, and power supply is like a constant multiplier after that.

One thing I thought about is, while a single stage maybe won't sound too different at 18V versus 9V (assuming similar biasing), certainly downstream stages would be different since the 18V signal from the output of the first stage would be twice as big and thus introduce clipping earlier.  To counteract this, you could use a resistor to drop the signal level going into the next stage.  This seems to be what ROG did with their 18V Supreaux Deux (http://www.runoffgroove.com/sd.html), which can be compared with their original 9V Supreaux (http://www.runoffgroove.com/supreaux.html).   

[ashcat_lt]

One thing I thought about is, while a single stage maybe won't sound too different at 18V versus 9V (assuming similar biasing), certainly downstream stages would be different since the 18V signal from the output of the first stage would be twice as big and thus introduce clipping earlier.  To counteract this, you could use a resistor to drop the signal level going into the next stage.  This seems to be what ROG did with their 18V Supreaux Deux (http://www.runoffgroove.com/sd.html), which can be compared with their original 9V Supreaux (http://www.runoffgroove.com/supreaux.html).

See, but what's the point in that?  Maybe if we expected massive loss along the way, or to pick up a bunch of noise, we might want to whack it up at the transmitter end and then pad as necessary on the receiver end, but if we're going 0.4" or whatever through a cap to the next transistor...  And if it really kind of is going to sound (look, be) about exactly the same...


Edit to add - I think we are agreeing that the curves will be the same either way.  When you said Idss it kind of concerns me.  This says that the transistor will only ever pass some maximum amount of current.  That means that its minimum reffective resistance depends on the supply voltage.  That means...something...but I'm still working on my first pot of coffee and have to get to the studio...

[Frank_NH]

Well, my main point is that if your input signal is larger with a higher supply voltage, then you WILL clip the next stage sooner than with a lower supply voltage, due to the fact that Vp is fixed for the JFET.

[Groovenut]

So this may not be here nor there, but I ran three sims with a center biased 2N5457 at 9, 18 and 36 volts V+. In the sim you can see that Vp pretty much determines where the input voltage will clip regardless of the V+. At each V+ I roughly center biased, correcting with the drain resistor while leaving the source resistance and input signal level the same. You can see, with adjusted center biasing, the output is primarily what changes scaled to the V+. This is all of course without a source resistor bypass cap. It all goes out the window when you add that 


Again may mean nothing but thought i would throw this up for everyone to look at as fodder for thought.

[ashcat_lt]

Well, my main point is that if your input signal is larger with a higher supply voltage, then you WILL clip the next stage sooner than with a lower supply voltage, due to the fact that Vp is fixed for the JFET.

I got that part.  Then you went on to describe how to "fix" it.  NP, we on the same page. 

Groovenut - that seems to confirm what we've been saying, but it would be a lot easier to see subtle differences if the images were somehow normalized and overlaid on one another..