confusion regarding treble response in dist+

[comfortably_numb]

Okay, not a dist+ but pretty close.  Here's the deal though.  I designed this thing, hoping to get a relatively flat response, but even with the tone control dimed, I'm losing some top end.  Now the 10pf cap should only roll off up tp 12057 Hz if I calculated correctly, and the tone control may even be useless. 

So what gives?  Where is my top end going? 

I'm not losing MUCH, but enough to make my distortion sound darker than my clean sound.

Also, I'm using this in a medium to low gain situation.

Thanks!

[WGTP]

It may be a matter of having too much bass in the circuit.  If your sure the tone control isn't doing much, try 1/2ing the .47uf cap to .22uf or .1uf.  This should thin it out some and reduce the gain some, which shouldn't be a problem since you are using it at medium gain.  To me that cap sets the "Character" of the whole pedal.   Of course switching is an option. 

Est. Roll Off Points

.47uf - 100Hz  Very low and liable to be Fuzz like
.22uf - 220Hz
.1uf - 470Hz
.047 - 1000Hz  More like I typically use for smoother distortion, although it may require some bass boost afterward.

1/2ing the 3.3K resistor is like 1/2 a cap, but raises the gain instead of lowering it. 

Between the 2 parts you can tweak to taste.   

[comfortably_numb]

Thanks very much.  I'll give that a try.  I think I'll go with the switching option, the bass boost I have now could come in handy, and who knows, maybe I'll end up prefering it.  I'd like to go with .22uf but all I've got is .1 at the moment.  I'll post my results when I get them.

[comfortably_numb]

Hmm...  Well, I put in the .1uF cap and heard the bass loss, which actually translated into some midrange honk - not what I wanted.  I still, however, am lacking the highs that I hear with the pedal bypassed.  It's subtle really, so I can live with it...I actually like the muscularity it has...but I'm wondering where they've gone, those tinkly upper harmonics.  Any other ideas?

[comfortably_numb]

The OA is a TL071 by the way.  I tried a 741 as per the original, and it had even less highs...so maybe this is just opamp choice?

[Elektrojänis]

That 0.001µF cap from input to ground might cut some treble. Guitar pickups are very sensitive for that kind of stuff as they are inductive. If you are feeding your guitar straight in to that pedal it might cut some treble.

741 probably made the circuits input impedance lower and loaded the pickup down and that caused even more treble cut. TL072 should have high enough input impedance.

So... Try making that RF-blocking cap on the input even smaller.

[comfortably_numb]

What is the smallest value I can use and still have an effective block on RF?

[PaulC]

The cap to block rf on the input will work much better if brought after the 10k resistor.  It'll set up a simple filter allowing you to use a much smaller cap.  The thing you have to watch out for is if you make the cap to big it'll interact with the volume pot on the guitar rolling off alot of highs as you turn down the volume.  If you make the cap small and use a larger series resistance it'll have less effect on the volume pot, but may add some noise due to the larger series resistance. 

In most guitar amps you'll see a series resistance of 34k with about 150pf to ground due to the miller effect of the input tube.  This will give you -3db @ 31Khz. 

It might look like you should be able to lower this freq point a good bit without effecting the highs, but it changes when you adjust the guitar pot, so it might be to dark sounding when you've got things turned down if you increase the cap to much. 

Hope this helps!  PaulC

[Mark Hammer]

What you have here is a .001uf cap situated somewhere along a 20k "pot" consisting of the 10k xied portion and the 10k variable portion.  The diodes are situated smack in the middle, but where the cap goes can change.  The rolloff produced by the cap is given by 1/(2*pi*[10k+Kp]*.001), where Kp is the resistance of the pot portion between the diodes and cap.  When the cap is situated where it normally is, in parallel with the diodes, that rolloff starts at: 1/(6.28*.010*.001) = 15.9khz.  When the wiper is moved over so that the cap now goes from the other end of the pot to ground, that rolloff starts at 1/(6.28*.020*.001) = 8khz.

The 10pf feedback cap is going to provide a rolloff around 15.9khz, so the tone control will essentially pass all the content the distortion is putting out and only take off an octave from that bandwidth.  Not going to be terribly audible.  When you consider as well that the typical guitar amp speaker tends to crap out above maybe 6-8khz, there is small wonder that you aren't hearing much effect of the tone control.

But the loss of top end is baffling, since you should get plenty of sizzle here.  Just out of curiosity, have you measured the caps in question?  Is there a possibility that, oh I dunno, the cap you think is 10pf might be 100pf?

[PaulC]

Mark's right about the tone control - there's not much play in it.  You could play around with the values a bit to make it work better.  You could try something like a 1K resistor for the fixed value, and a cap value of 10n.  That would give you -3db points at about 15.9khz and 1445hz.  Plus you'd have just a little bit more output because you'd have 11k total in series with the 100k output instead of 20k. 

I do think that what you're hearing as a high end loss is mostly due to the 1n cap on the input.  That's going to really suck out some highs as soon as you turn your volume knob on the guitar.

Good luck!  PaulC

[Transmogrifox]

You could add a 100k resistor in series with a 1 nF capacitor for a boost at 1.5kHz between the noninverting terminal of the opamp to ground.

For medium to low gain settings, the .47uF cap makes a time constant that's lower than any frequency in the guitar signal due to the portion of the pot that is put in series with the cap to ground.  It's not a wonder it tends to dull the higher frequency content.

You'd want to be changing that cap value to something around 1nF to 3.3nF or so to bring it out of the 'honkey' range for medium to low gain settings.  Unfortunately it would be very tinny and bad for high gain settings, thus I recommend adding the treble boost separately as I mentioned above.

[comfortably_numb]

What you're saying here would boost treble frequencies BEFORE the gainstage of the OA, correct?  I'm to take this RC network from before the positive pin of the OA.  So what effect would this have once the signal was amplified by the OA?  Also, how do you calculate that corner frequency?  I got 1591Hz using 1/2piRC, = 1/2pi x 0.1 x 0.001 = 1591.  Probably missing a decimal somewhere judging by the answer. 

Is resistor in series to ground with a cap the same as a cap after a resister in series with the sigal?  I suppose it isn't, since this is a treble boost, and the opposite would be a lowpass filter.  So how does this work, and how is it calculated?

Is there a similar way to achieve a bass boost pre-OA?

Just trying to learn this theoretical design stuff.  Thanks!

[Mark Hammer]

What comrade Trans is describing is the pre-emphasis of the Rat gain stage.  With a noninverting op-amp, you can have multiple paths to ground from the invertng pin, with each path creating a different gain result for different portions of the frequency spectrum.  I've never attempted it with an inductor, but in general this strategy only works (or rather, only gets used with) higher frequency content.

So, your treble enhancement options are to:

a) cut the bass at the op-amp input so treble gets preference during the boosting.
b) provide more boost for mids and highs than for the rest of the spectrum.

This is counteracted by the treble cut of the feedback cap at the input, in the op-amp feedback loop, and in parallel with the diodes.  Unless you specifically have a lot of treble cut, or insert some sort of scoop-like filtering after the clipping stage (which will necessitate some gain recovery), it will be difficult to have something that gives you a more substantive bottom end as well as top-end bit.

On the other hand, what I've never seen implemented for bass purposes is the T-filter network in the feedback path of an op-amp.  Normally this is used for treble enhancement.  Keep in mind that the way op-amps work is that the more negative feedback is conserved from the output back to the input, the lower the gain is.  Conversely, if you lose some portion of the feedback, that reduces the negative dampening effect of the feedback and the gain goes up.

So, imagine the following feedback path.  You have your 1M resistor, but in addition, you have two caps in series.  Now, normally, several things happen here.  First, caps in the feedback path allow for lots more negative feedback for higher frequencies to find its way back to the input, thus reducing gain for high-end and leading to huge treble cut.  Second, if the caps are in series you get an effective cap value equal to 1/Ca + 1/Cb, so a pair of 1000pf caps in series gets you 500pf.

Now, take a resistor, and run it from the junction of those two caps, to ground.  What do you have?  First off, the first cap coming back from the output, and the resistor to ground, form a highpass filter, shunting high-end in the feedback to ground.  Since you are now losing some of the dampening negative feedback signal, that means the gain for high end goes up, at the frequency set by the RC combination.  However, the second cap, in tandem with the first one also allows high end to pass smoothly from the output back to the input, resulting in more negative feedback at higher frequencies.

Altogether, though, they do something very interesting.  Let's say we have, in addition to our 1M resistor, a pair of caps and a 150k resistor to ground.  The two .0027uf caps in series = .00135uf.  With a 1M resistor, that results in a reduction in treble gain starting around 118hz; obviously a lot lower than what you get with 10pf.  The 150k resistor and first .0027uf cap, however, form a highpass filter that dumps content below 392hz to ground so that it tends not to reach the inverting input and dampen the gain of the op-amp.  That means that the gain of the op-amp in that region will be greater.  When the two are combined, there will be a small resonant peak introduced into the feedback loop that results in there being a bit more gain halfway between 118hz and 392hz (255hz) than for the rest of the spectrum.  How MUCH gain is going to depend on the resistance to ground from the inverting pin, as it usually does.

In fact, this sort of network is precisely what is used for a great many auto-wah circuits, like the Doctor Q, except that the fixed resistor is made variable, so that the centre frequency of that "bump" moves around.

You could, in principle, insert such a network in the feedback path, and you ought to, in theory, anyways, get a bass hump which would remain in place, regardless of where you set the gain.  Well, in theory, anyways.  Since you opted to use a gain pot arrangement that varies both the feedback and ground resistance (which is a GOOD thing), that will monkey around with what the two caps in series do.  Happily, that can be altered by retuning the hump-network via the resistor to ground.  The smaller that resistor gets, the lower the hump frequency will be.

That's about as much explaining as I can do at the moment.  hairyandy is in town and we're getting together after lunch to plug and talk gear.  I'll check back later tomorrow.

[comfortably_numb]

The Rat was what first came to mind when I read that response, but then I re-read, and he said the NONinverting terminal, so I assumed differently.  It makes a bit more sense now that it is clarified.

This is all very interesting however.  Once I get some treble back I intend to revise the tone control to get a bit more use out of it. 

The feedback cap hasn't been tested, but it SAYS 10pf.  It's a silver mica from smallbear.  Also, I initially tried a 22pf cap, as per your suggestion Mark, and the top end was DEAD.  It sounded dull and lifeless.  At this point, it's just puzzling me as to where the top end has gone, but actually sounds pleasant.  I just want to learn as much about this as possible, and also get the unit to a point where it has a usable tone control.  I want MORE treble, so I can roll it off.  Figures.  My fiance doesn't get it.

I really like the T-network idea, and will consider using it for my next design (I'm really liking Orman's new bmp tone clipping idea!) but for this particular pedal (which is already on vero and in a finished box) I'd like the keep the solution as simple and close to the original design as possible.

PaulC - thanks for your comments.  To clarify about the way this pedal is being used, it comes immediately after an ernieball jr. volume pedal.  This of course is passive, so may be interacting in the way you are specifying.  I wonder why the original distortion plus was made this way?  And why no one has addressed this so far?  I suppose because it sounds good as is. 

Mainly my concern was that the theoretical numbers didn't add up with the result.  So, I'll alter some things about the tone control to make it more usable.  Moving the input cap to ground AFTER the 10k resistor might not be feasible.  Any other options concerning that?  Can I just use a smaller value to get out RF?

Thanks everyone - and Mark, feel free to continue your explaination or ideas or whatever you want to post.  I always enjoy reading what you have to say. 

[comfortably_numb]

So I've been looking through schematics today and I've found that very few designs actually use that RF input cap to ground.  Can I just get rid of it?  My first notion was to just clip the lead and see, but then I thought, hey I like the sound, why screw up a good thing without an easy fix?  So I'll wait until tomorrow after I buy some more solder braid to find out.

I think I'll try this (given the go ahead) and the changes to the tone control suggested by PaulC when I get some freetime tomorrow (or whenever)

[Transmogrifox]

So I've been looking through schematics today and I've found that very few designs actually use that RF input cap to ground.  Can I just get rid of it?  My first notion was to just clip the lead and see, but then I thought, hey I like the sound, why screw up a good thing without an easy fix?  So I'll wait until tomorrow after I buy some more solder braid to find out.

I think I'll try this (given the go ahead) and the changes to the tone control suggested by PaulC when I get some freetime tomorrow (or whenever)

Often times you can get rid of it--particularly in a pedal such as this where you get some RF rejection due to the parasitic input capacitance and limited frequency response of the op amp.  If you hear an increase in noise or AM radio, then put something in there.

Here's a guideline that will help you decide on a good value:
You probably don't want a 3dB frequency much lower than 15 kHz since 3dB is 3dB...as in, there is significant loss on the corner frequency of a 1rst order filter, so though there may not be anything appreciable above 10 kHz in your guitar signal on the input, there would be some measureable loss down to perhaps 8kHz, or lower.  After 3 or 4 of such "RF blocking" pedals, you'd have some audible signal degradation.

So let's use 15kHz as our absolute Minimum corner frequency.

We can figure a guitar's ouput Volume pot is 500k and the impedance of the pickups is comparatively negligible.

1/2 volume is the worst case for output impedance.   In this case you have essentially two 250k resistors in parallel to ground.  This is a source impedance of about 125k.

if
f0=1/2*pi*R*C,
then,

C = 1/2*pi*f0*R

As determined above, f0 must be 15kHz worst case, and R=125k

C=85pF

Try to keep it between 47pF and 85pF.  Too low and it won't give you enough attenuation at AM frequencies to justify having it.

Above, about the recommendation to add the 100k resistor in series with the 1nF capacitor between the negative terminal and ground; you could read "negative terminal" the same as "noninverting terminal".  The one with the minus sign.  Anyway- you did the calculation right.  1.5 kHz is 1500 Hz.  The "k" moves the decimal point for you.

So the corner frequency for the boost is about 1592 Hz, which is 1.592 kHz, which is actually more accurately rounded to 1.6kHz (my bad), but I just called it 1.5kHz just because... 1.5 kHz is just as viable if you consider the 5% tolerance on your resistor and 10% tolerance on a capacitor.

[comfortably_numb]

Very cool of you to go through all that Trans - sorry about my bad math...or rather, my good math and bad attention. 

I thought the minus sign denoted the inverting input of an opamp.  Is this incorrect?  I do understand your suggestion now however.  Thank you.

Theoretically, with this concept in mind, I could use many separate RC networks to ground to make a complex pre-emphasis for this type of circuit, no?

I'll cut the cap out when I get a chance and then inform everyone of the results.  If necessary I'll order the replacement 100pf cap from SmallBear.  Assuming removing it completely has the proper effect, the 100pf should cut out any noise without the associated problem I'm having now (again, not really a problem, just a puzzlement.)

Thanks again!!!

[Mark Hammer]

I really like the T-network idea, and will consider using it for my next design (I'm really liking Orman's new bmp tone clipping idea!) but for this particular pedal (which is already on vero and in a finished box) I'd like the keep the solution as simple and close to the original design as possible.

Thanks everyone - and Mark, feel free to continue your explaination or ideas or whatever you want to post.  I always enjoy reading what you have to say. 

Well feel free to enjoy them, but exercise self-restraint when it comes to trying them out.  Just after I scooted off on the bus, I realized that what I had suggested would not work.  Well, let me modify that.  It would certainly work for creating the "bump".  Problem is there would be nothing outside of the bump.   I was so fixated on thinking of something that would produce a low-end emphasis that I completed neglected the remainder of the signal.  Too busy pruning the bushes to stand back and look at the garden!

What I described yesterday was essentially a very low bandpass response - NOT what you were after.

[WGTP]

When I'm messing with treble boost or cut, I sometimes crank the presence/treble and volume up on things till it hisses real good and then it is easier to hear what is going on when you try to boost or cut it.  Not very scientific, but it's easier to hear what is going on then.  Crank things up and clip the .001uf cap and see if you hear any difference.  As for using a smaller cap, say 470pf, I would think you would want the largest value possible that you can't "hear" doing anything.

The cap in the feedback loop, I think you were using a 500K pot, a 47pf would be barely audible at max gain, I would guess.

In looking at the original schematic, a .047uF cap is used rather than the .47uf.  That is why I suggested modifying it.   

If you want it Rat like, try the following values:

2.2K resistor - 47n cap
                &
22K resistor - 100n cap

If you really want to make it bark, these values will produce the same response curve, but with over 10db more gain:

1K resistor - 100n cap
              &
10K resistor - 220n cap

[Transmogrifox]

Above, about the recommendation to add the 100k resistor in series with the 1nF capacitor between the negative terminal and ground; you could read "negative terminal" the same as "noninverting terminal".  The one with the minus sign.

Sorry for the confusion.  I just caught the stupidity in this.  What I have meaning to say all along is "inverting terminal".  It was an airheaded mistake, and I can hardly believe that your confusion didn't make me see it. 

This statement should be,
"Above, about the recommendation to add the 100k resistor in series with the 1nF capacitor between the negative terminal and ground; you could read "negative terminal" the same as "inverting terminal".  The one with the minus sign."

Please disregard my confusion with "inverting" and "noninverting".