LM13600/700 as VCA... square wave troubles.

[ExpAnonColin]

Has anyone else given LM13600/700's a try for tremolo-like circuits for guitar? I have been for quite some time and always end up with the same problem. I'm always able to null the LFO signal completely, but when the guitar is being input, the square wave is just to FAST... so if you think of a sine wave, it brings the sine wave from whatever position to zero so fast that you can hear it. The bigger the change, the louder the click... so for example, if at that moment the sine wave input is at a peak, the click will be loud, but if it's at one of it's nodes, there is no click... creating a pseudorandom series of clicks while there is input. Putting capacitors in various places as lowpass filters does fix the problem, but it creates funkiness with the OTA... I think that the capacitor's way of charging and discharging makes the OTA pop. I've tried to null this new pop in a variety of ways, but can't... adjusting the input bias, and the standard LM13700 method of shifting the bias between the +/- inputs, both of them won't get rid of this "capacitor pop"... which is there even if the cap is before a buffer.

Has anyone encountered this problem? The only LM13700 based guitar pedal/tremolo design I have seen is the ring frobnicator (though there are infinite synth VCAs using OTAs), which has no square wave at all... I imagine he found the same problem. Does anyone know of another LM13700 trem? Or perhaps any more ideas?

Otherwise I am going to give up and go back to a light-based approach... it just feels so dirty! H11F3's will be my best friend...

-Colin

[R.G.]

Solve it the same way we solve popping with series JFET switches. Ramp the control voltage up/down over a period of about 10ms.

Even a series resistor/shunt cap on the control line should help.

[ExpAnonColin]

Solve it the same way we solve popping with series JFET switches. Ramp the control voltage up/down over a period of about 10ms.

Even a series resistor/shunt cap on the control line should help.

By control line, you mean the line from the LFO to the current control of the OTA?  There is already a 20k resistor to current limit... any more would change the control current.  A capacitor shunted to ground (or Vb) on the control line, as I sad, creates a new, un-nullable popping.

-Colin

[gez]

Yup, been there.  Solved it by converting the square into a trapezoid:

http://www.diystompboxes.com/smfforum/index.php?topic=44065.0

Scroll down for details.

[ExpAnonColin]

Ah, that clipping integrator is an interesting solution.  I will give it a try.

-Colin

[gez]

Ah, that clipping integrator is an interesting solution.  I will give it a try.

-Colin

It's easier to just use the other half of a LM13700 (as outlined on the first page), unless you're going for stereo? 

Whatever you use, don't be afraid to slope the sides a fair bit (I tweaked my circuit empirically till clicking went and it sounds like on/off stutter despite having gentle slope).  Easy to incorporate a pot to do this.

[ExpAnonColin]

I am doing stereo, yeah, so I want both sides for modulation... the final design will be a device that pans between effects loops, and sums them, or simply pans outputs.  The hardest part, as it always seems in amplitude modulation, is the square wave.

-Colin

[A.S.P.]

 
I see some (lots of..) future foreheadslappers in my imaginary eyes 

[gez]

I see some (lots of..) future foreheadslappers in my imaginary eyes 

Well, I suppose 'panning' with a square wave is just switching in my books...

[A.S.P.]

 
the slapping just concerned the "square-ticking" in general, in OTA VCAs...

[ExpAnonColin]

OK, tried the diode integrator, it introduced "pops" just like the capacitors did.  Looks like I'm going opto...

-Colin

[gez]

OK, tried the diode integrator, it introduced "pops" just like the capacitors did.  Looks like I'm going opto...

Did you try increasing the value of the cap/resistor (with a scope to check you're not ending up with a triangle)?

Also, if you're feeding the Iabc pins from the output of an amp that swings to ground, try raising the 'floor' by 2 diode drops (connect from Vss pin to ground).

[ExpAnonColin]

OK, tried the diode integrator, it introduced "pops" just like the capacitors did.  Looks like I'm going opto...

Did you try increasing the value of the cap/resistor (with a scope to check you're not ending up with a triangle)?

Also, if you're feeding the Iabc pins from the output of an amp that swings to ground, try raising the 'floor' by 2 diode drops (connect from Vss pin to ground).

Yeah-I was measuring with scope and ear.  The only situation that would entirely remove the click was with upwards of a .22uf cap.  This made it triangle at higher speeds for sure, and nulled it out completely at speeds I wanted.  But it did remove the pop.  The output doesn't swing to ground, it swings to 1V.  Were you able to get an entirely nulled pop with this type of integrator and a LM13xxx?  At this point, I can get a better square wave using an H11F3 and op amps, which can be nulled with lowpass filters...  What is it with OTAs...

-Colin

[gez]

The only situation that would entirely remove the click was with upwards of a .22uf cap.  This made it triangle at higher speeds for sure, and nulled it out completely at speeds I wanted.  But it did remove the pop. 

What frequency range are we talking about here?

Were you able to get an entirely nulled pop with this type of integrator and a LM13xxx?  At this point, I can get a better square wave using an H11F3 and op amps, which can be nulled with lowpass filters...  What is it with OTAs...

The op-amp integrator outlined in that post?  I don't normally use this, it was just posted as a 'fix' (only worked at the expense of sacrificing the PW control) to RG's circuit.  The OTA idea I posted works, though maybe not if you're going for a large frequency range as you'll just end up with a triangle of diminishing amplitude at some point.

Can't remember the top rate of the trem I made.  Round about 16 - 18Hz I'd say from listening to it.

[ExpAnonColin]

My top rate is just about 30hz, so it is a bit higher, I suppose.  Not that I wouldn't be willing to make a 15 hz sacrifice.  The max I found at a .22uf cap that it could go without completely nulling was probably 15hz, but that was quite triangular pulsish.

-Colin

[gez]

The max I found at a .22uf cap that it could go without completely nulling was probably 15hz, but that was quite triangular pulsish.

The waveform I got at top rate was still trapezoid, though admittedly it was heading towards triangle.  The thing is though, at around 18 Hz plus I doubt your ear is going to perceive subtle differences in waveforms (so long as they're modulating to the point of switch off).  It will make a difference at frequencies below this, but then the waveform definitely isn't heading for anything resembling a triangle and sounds like on/off stutter to my ears.

[ExpAnonColin]

Right right-I wasn't worried about teh wave being really triangular at higher speeds,  just so long as it's there!  And with the higher cap value, it wasn't.

-Colin

[Dolly Parton]

Could be a layout problem?

If this were me, I'd shift the floor of the op-amp up a little so that the OTA is never turned fully off (or at least test to see if this is causing the problem)...perhaps divide down the output slightly (unlikely you'll need rail-to-rail to get the modulation you require - it's how low you go that makes more of a difference with 'square wave' modulation).

gez Dolly

[ExpAnonColin]

It is on breadboard, so I am allowing some error for popping...  But the problem can easily be isolated by looking at the LM13700.  With the square wave "trace" moved around, closer to this point or that point, you don't hear popping, but once it touches the OTA, the popping appears...  Using an op amp as a simple buffer instead, there is no popping.  I have to assume that something about the capacitor vs OTA is making it pop like that.  I will try dividing the output though...

Well, it works!  No more popping.  My max freq has been trimmed down to 12.5hz, which is still fairly fast.  I'm using a .1uf cap now...  If the trap wave is shifted up a little, it stops popping.  I can get the square entirely nulled.  The problem is that the square wave now swings from 2V to about 8.5V, while the triangle only goes from 2 to about 6.5... but I can correct that and remove some of the signal in the "off" state by adding a series resistance.

While working on the opto design yesterday, I started trying to make an adjustable trap wave by clipping the triangle wave... I think I will try that again now, too, because that way the triangle can be trimmed to max amplitude via the amplifier... then you can just amplify more and more, adjustable wave shape...  problem is people will be more tempted to turn it higher at higher speeds, as the slope will be less noticeable early on.  I also want to cmoe up with a way to use 100k linear, not audio...

-Colin

[gez]

My max freq has been trimmed down to 12.5hz, which is still fairly fast.  I'm using a .1uf cap now...  If the trap wave is shifted up a little, it stops popping.  I can get the square entirely nulled.  The problem is that the square wave now swings from 2V to about 8.5V, while the triangle only goes from 2 to about 6.5...

Perhaps shift the op-amp a little higher and use a larger resistor for Iabc?  This might allow you to adjust things for a steeper slope/higher frequency. 

This is why I prefer the OTA approach I outlined in that thread.  You can set a very definite upper threshold.  Anyway, you're on the right track by the sound of things.  Stick with it and keep experimenting.

PS  I managed to do this and without the need for a nulling pot, so it's doable.