LM13700 - Anyone familiar with prototyping VCOs using this chip please?

[frequencycentral]

Hi, I'm working on a new effect which requires an VCLFO using half a 13700.

I breadboarded one from the 13700 spec sheet. I'm using 9 volts, with one opamp from a LM324 providing bipolar supply. My diodes are 1n4148.

Here's the schematic:



I just changed a couple of values, using what I had available, so R2 is 1k8, and the 5k1 is 4k7 in mine.

Here is the waveform I'm getting:



I call it a 'sharkfin triangle'.

My question is this: is there any way to make a perfect triangle using this circuit? I've made VCOs/VCLFOs before with better symmetry using both OTAs of the 13700, but I really want to use just one in this application. Any suggestions, or is the imperfection inherant in a single OTA VCO? If that is the case, I'll just go with it anyway, it's close enough for jazz!

[gez]

I've breadboarded that circuit in the past.  Can't remember much about it as it was such a along time ago; I'd have to check my notes to remember how it works.

Anyway, it's not supposed to be a triangle oscillator (IIR).  Use the dual-amp version for that outcome.

[edit] I have a vague feeling that you buffer and amplify the + input for a square-wave. 

[snap]

I`d try a 2k5 trimpot for R2, or a 2k2 paralleled by a 10k or 25k trimpot for fine adjustment.
The 4k7 for 5k1 seems alright.

[gez]

I'm pretty sure that won't make a scrap of difference.  The diodes provide the thresholds at which the output flips over, and this is reflected in the scope readings (you can see those sudden transitions at the output).  That level shift is always going to be there.

This isn't meant to be a 'pure' triangle oscillator.  I'm afraid you're going to have to live with this fact and build the real thing.

[Paul Perry (Frostwave)]

Yeah, I never got that to a perfect triangle either!
But, depending what you are doing, you may not need an exact triangle.. if you run the lopsided triangle through a triangle to sine shaper, if you get the levels right, i guess you could make it look like an acceptable flattish sine.

[frequencycentral]

if you run the lopsided triangle through a triangle to sine shaper, if you get the levels right, i guess you could make it look like an acceptable flattish sine.

Can you point me in the right direction for a triangle to sine shaper?

[Paul Perry (Frostwave)]

Ironically, what I use is an OTA arranged as an amplifier, and overdrive the input by just enough to get the flattening.
But if you use that, you might as well use both halves of the LM13700!

here's another approach: http://mypeoplepc.com/members/scottnoanh/sitebuildercontent/sitebuilderfiles/th_sine_shape.jpg with a transistor pair

and one with diodes: http://electronicdesign.com/Files/29/2257/Figure_01.gif

Notice the rails are +-15v, standard for the old analog stuff. But you can do it at lower voltages like +-4.5, just tweak the resistors.
But even if you can't get what you want, you will definitely learn a few things while trying!

Now if anyone can show me how to get a proper triangle from a CD4046 with minimum extra parts, I will be much obliged..

[frequencycentral]

Ironically, what I use is an OTA arranged as an amplifier, and overdrive the input by just enough to get the flattening.
But if you use that, you might as well use both halves of the LM13700!

Yeah, I was thinking the same myself - why go to the bother of using extra circuitry - so I'll use a dual OTA 13700 VCO and use a 3094 or 3080 for the other OTA I need in the pedal.

Thanks everyone!

[gez]

if you run the lopsided triangle through a triangle to sine shaper, if you get the levels right, i guess you could make it look like an acceptable flattish sine.

Can you point me in the right direction for a triangle to sine shaper?

You can use the other half of the LM13700, though slightly differently from the way Paul was outlining.  Set it up as an integrator and use the same pot for both Iabc pins.  The integrator acts as a filter.  If you use a large enough cap value, it will round out a triangle into a very acceptable sinusoidal wave form.  There's a linear decrease of amplitude with frequency when it comes to integrators, but by tying everything to the same pot this is compensated for as the frequency rate is adjusted: higher frequencies receive more gain (less resistance on the pot) and therefore amplitude remains stable.

I usually set up the amp in non-inverting mode.  Try a 10k from 'buggered triangle' output to the - input of the integrator and then a 470R from the same input to your virtual, op-amp earth.  Use the integrated darlington buffer and hook another 10k from its emitter to the - input.  The + input is tied to op-amp earth.  Stick a cap at the output of the OTA (note: I mean the output of the OTA before the buffer).  Light blue touch paper and stand well clear.

Cap value?  Suck it and see.  I use 1u, but you'll get such an attenuation of the signal that you'll need a gain recovery stage (second half of your op-amp virtual earth?)  Than again, messing around with values might compensate for this (one of those 'hmmm, I should have done that differently' moments). 

Good luck and if you don't come back...you'll have served a good cause.

[frequencycentral]

Ok, I had a little play about after it occurred to me that maybe a cap in parallel with the diodes would smooth them out. this is what I got with a 0.033 cap:



I'm pretty please with that. There is a volume drop but I think it will be useable. Obviously the symmetry varies a little with the pitch though.

[gez]

Obviously the symmetry varies a little with the pitch though.

Symmetry or amplitude?

[Paul Perry (Frostwave)]

Symmetry or amplitude?

I would guess both, but that may not be a problem, depending on range. For a LFO it might pass.

[frequencycentral]

Obviously the symmetry varies a little with the pitch though.

Symmetry or amplitude?

Symmetry or amplitude?

I would guess both, but that may not be a problem, depending on range. For a LFO it might pass.

Yeah both - some really weird low amplitude shapes at extreme settings! So I'm going with a dual OTA VCO - at least it saves me an opamp as it already has square output, so I don't have to convert, so I can ditch the quad LM324 in favour of a dual - providing the circuit likes vref without an opamp buffer.

The circuit is going to be a trem with two rate controls, stomp controlled with a slew limiter to ramp between the two. I'm hoping to achieve bit crusher type sounds at faster speeds. So my chip count will be one 13700 for the VCO, one 3080 or 3094 for the VCA, one dual opamp for the slew limiter. I'm going to have a blend for the triangle/square, and switchable cap on the VCO range.

I'm even toying with using a 6111 tube in place of the VCA OTA - a la Tone God's Bullitt.

I'm sure its all been done before...........................?

[frequencycentral]

So it's breadboarded! And working!

Its the triangle/square dual OTA VCO from the 13700 spec sheet coupled with U3A and U2B from Tone God's Bullitt - but using a 6111 submini tube. I'm so pleased!! 



Very nice trem - just have to get the slew limiter breadboarded!

[Paul Perry (Frostwave)]

The circuit is going to be a trem with two rate controls, stomp controlled with a slew limiter to ramp between the two.

I'm not sure why you need a slew limiter.
If you have two pots, each selecting a control voltage to drive the LFO, and you switch between the two voltages from the pots, but run the selected voltage via a resistor, then if you have a cap from the other  end of the resistor to ground, that might be good enough to smooth the transition back and forward.

[frequencycentral]

The circuit is going to be a trem with two rate controls, stomp controlled with a slew limiter to ramp between the two.

I'm not sure why you need a slew limiter.
If you have two pots, each selecting a control voltage to drive the LFO, and you switch between the two voltages from the pots, but run the selected voltage via a resistor, then if you have a cap from the other  end of the resistor to ground, that might be good enough to smooth the transition back and forward.

I want to be able to define the rate of change between the two speed - like a Leslie. A slew limiter will give me a 'Ramp' pot, for slow or fast transisions.

[nelson]

Make the resistor variable and you have a variable ramp control - no slew limiter needed.

[frequencycentral]

The circuit is going to be a trem with two rate controls, stomp controlled with a slew limiter to ramp between the two.

I'm not sure why you need a slew limiter.
If you have two pots, each selecting a control voltage to drive the LFO, and you switch between the two voltages from the pots, but run the selected voltage via a resistor, then if you have a cap from the other  end of the resistor to ground, that might be good enough to smooth the transition back and forward.

Make the resistor variable and you have a variable ramp control - no slew limiter needed.

So what you're saying is that I take the common from the stompswitch to which the two rate pots connected, via a pot (1M?), then a cap to ground, into the VC input of the VCO. Simple as that?

Maybe I'm thinking I need an opamp buffer either side of the pot/cap because most of my experience is with (often over engineered) modular synths?

[nelson]

The circuit is going to be a trem with two rate controls, stomp controlled with a slew limiter to ramp between the two.

I'm not sure why you need a slew limiter.
If you have two pots, each selecting a control voltage to drive the LFO, and you switch between the two voltages from the pots, but run the selected voltage via a resistor, then if you have a cap from the other  end of the resistor to ground, that might be good enough to smooth the transition back and forward.

Make the resistor variable and you have a variable ramp control - no slew limiter needed.

So what you're saying is that I take the common from the stompswitch to which the two rate pots connected, via a pot (1M?), then a cap to ground, into the VC input of the VCO. Simple as that?

Maybe I'm thinking I need an opamp buffer either side of the pot/cap because most of my experience is with (often over engineered) modular synths?

Yep, as simple as that.  T=RC

[frequencycentral]

Wow! Thanks! That's great.

I've been at work all day - when I got home I started to put up a garden shed - this circuit is all I've thought about all day though.

With any luck I'll finish breadboarding it tomorrow.

If everything works as expected I'll be happy as a pig in shit!

Funny though - I was expecting a higher parts count. I guess it really is true what I said about modular synth circuits being over engineered. I'm constantly amazed what can be achieved in stompbox circuits with next to nothing.

Thank you to everyone who contributed to this thread - I've also got the OK from The Tone God use the part of 'Bullitt' I borrowed in the schematic when I 'publish' on this forum.

So give me a couple of weeks, to work up a completed pedal, and to persuade my girlfriend to let me use that great photo of her legs for the artwork!