Simple DOD FX17 wah circuit spinoff.

[radio]

I'm not done with it ,because I got too complicated . I first choose to build it in a Schaller Wah

enclosure, then decided not to gut it ,and instead build it into a EV-5 expression pedal after

seeing this link : http://www.noystoise.com/2011/02/hartman-envelope-filter-mod-job.html

Now I think I'd better had a separated enclosure,to see all the controls;a seperate footswitch

would have been needed anyway to preserve the entire range of the 10k pot ,and keep the

EV-5 unchanged. Thank you for the refund ,hope to post some report later.

[Paul Marossy]

Well Ronan, I think you should keep pursuing this project, it's a good sounding wah if you ask me. I like that there is no inductor to pick up EMI or to be microphonic. Maybe you could do a new design with the easier to obtain LM13700?

[artifus]

i built this from a kit when it came out: http://img.photobucket.com/albums/v517/uncle_boko/Psycho_wah/psych_wah.gif

finally found the schematic via this post: http://www.diystompboxes.com/smfforum/index.php?topic=37073.0 having long since lost the article - thank you stephen! may be of interest.

[R.G.]

 First time I saw this one. Some comments.

- the schemo says IC1 can latch up. This is much more unlikely if you terminate the + input at Vr for both halves. The OTA doesn't care what voltage its output is tied to as long as it's within the power supplied to it, and this will keep the input common mode for the TL072 within bounds. Value is noncritical as long as it's high enough to not mess with the cap rolloff. Try 1M. As noted, an opamp with a rail-to-rail input cures it as well. The output of an OTA is a high-compliance current, so the DC voltage it holds is determined by stuff outside the chip.

- The hiss is probably the ca3080s. They're known for noise. The LM13700 is better, the CA3280 and NE5517 better still.

- Companding works, but since the wah will always cut highs anyway, you might as well put a second filter on the thing after the buffer and cut everything above 3kHz anyway. Anything up there is pretty much guaranteed to be OTA hiss, as both LP and BP will have cut most of the highs from the original signal.

[Ronan]

- the schemo says IC1 can latch up. This is much more unlikely if you terminate the + input at Vr for both halves. The OTA doesn't care what voltage its output is tied to as long as it's within the power supplied to it, and this will keep the input common mode for the TL072 within bounds. Value is noncritical as long as it's high enough to not mess with the cap rolloff. Try 1M.

I did try a 1M tacked underneath the pcb on both +ve inputs to the IC1 TL072 but it still latched up. I will try 470k tomorrow night.

As noted, an opamp with a rail-to-rail input cures it as well. The output of an OTA is a high-compliance current, so the DC voltage it holds is determined by stuff outside the chip.

This is where I am confused. In the case of an LM13700, I don't understand how the output of the OTA can work into the darlington emitter-follower stage if there is no bias voltage present to bias the emitter-follower. An OTA having a current output vs a voltage output suggests there may be an output voltage present depending on the load of the output. Maybe the darlington arrangement does present some load, if not, the capacitor. On the NE5517 it mentions a "constant impedance buffer" maybe the LM13700 has that too? I'm sort of lost as to what's causing the problem, AC or DC conditions? Maybe this is something I should experiment with and find out myself, if I am capable.

Some extra detail on the symptoms - in my build, with TL072 in IC1 position, the circuit suffers latching at switch on of power, only if the pedal is in the treble position, and if the pedal is then moved to the bass position, it will unlatch and work fine from there on. If power is applied with the pedal in the bass position, there is no latch-up. And as noted, no problem in my build with TLC2262 opamp under any conditions that I've tried.

- The hiss is probably the ca3080s. They're known for noise. The LM13700 is better, the CA3280 and NE5517 better still.

- Companding works, but since the wah will always cut highs anyway, you might as well put a second filter on the thing after the buffer and cut everything above 3kHz anyway. Anything up there is pretty much guaranteed to be OTA hiss, as both LP and BP will have cut most of the highs from the original signal.

Along with Paul's encouragement, I think I should give it another go, the NE5517 looks reasonably easy to get at a reasonable price. And add a 3kHz LP filter. My build has less noise (and less hiss) than my favorite inductor wah, but if it can be improved on then no reason to hold back.

I have one more question, is there any advantage to use a TL072 buffer over the inbuilt darlington buffers? I guess I should use my ears to answer that question.

Thanks for your comments RG, I really appreciate it, I went too far out on a limb with this one, beyond a full understanding of what I was dealing with, which is usually a good recipe for some sort of disaster or setback, and then bounce back again.

[Paul Marossy]

- The hiss is probably the ca3080s. They're known for noise.

Yeah, I can't make this wah circuit by Ronan any quieter and had concluded it was because of the CA3080s. And I came to that conclusion by previously building Craig Anderton's "volume pedal retrofit". While that circuit is cool because it gets rid of the scratchy pot problem, it creates a new one - lots of hiss. Maybe when he designed that circuit it was OK as people were still doing home recording on tape and stuff which is inherently hissy, but these days it's just unacceptably noisey to me.

In the case of Ronan's circuit, it's pretty quiet in comparison to Anderton's circuit. I've got a little bit of hiss in a high gain scenario when not playing anything, but it's better than a vintage wah pedal. Not bad enough for me to deem it unusable.

[R.G.]

I did try a 1M tacked underneath the pcb on both +ve inputs to the IC1 TL072 but it still latched up. I will try 470k tomorrow night.

Hmmm. There goes another good theory, all shot to blazes. 

This is where I am confused. In the case of an LM13700, I don't understand how the output of the OTA can work into the darlington emitter-follower stage if there is no bias voltage present to bias the emitter-follower. An OTA having a current output vs a voltage output suggests there may be an output voltage present depending on the load of the output. Maybe the darlington arrangement does present some load, if not, the capacitor. On the NE5517 it mentions a "constant impedance buffer" maybe the LM13700 has that too? I'm sort of lost as to what's causing the problem, AC or DC conditions? Maybe this is something I should experiment with and find out myself, if I am capable.

The output of all the OTAs I know of is the junction of one NPN and one PNP collector. The impedance is deliberately as high as it can be made, so that the output closely approximates a current source. It has no "natural" output voltage other than being between the V+ and V- on the chip. The darlington buffers on these things have an internal bias source that tends to put the *voltage* on the base of the darlington near the middle of the power supply, while still preserving a high impedance at its input. This state of affairs lets the voltage at the output of the OTA be determined by leakages, transients, and other such black magic. Generally, you put a resistor to whatever DC voltage you want the output to sit at, and often make the resistor be the item that determines the voltage gain of the OTA stage overall.

That seems not to be working in this setup. I can speculate some more, though.

Some extra detail on the symptoms - in my build, with TL072 in IC1 position, the circuit suffers latching at switch on of power, only if the pedal is in the treble position, and if the pedal is then moved to the bass position, it will unlatch and work fine from there on. If power is applied with the pedal in the bass position, there is no latch-up. And as noted, no problem in my build with TLC2262 opamp under any conditions that I've tried.

I think that kind of makes sense from what little I know of the TL072 latchup mechanism. I think it may be that the capacitors which load the outputs of the OTA are initially at 0V at power up. They stay that way until current from the OTA output changes them. The TL072 input is the gate of a JFET, so it is held at the cap's voltage as it charges. If power comes up fairly quickly, the input is held at ground on the TL072 while power comes up, and it goes into latch. Flipping into bass position may cure this. As I say, I'm speculating.

But I can think of some things to try to correct the issue. If my guess is right, it's the difference between power supply and the "ground" end of the load capacitor that's getting you. If you can make that capacitor's initial voltage of 0 track the reference voltage at power up, the TL072 inputs will not see a difference between the cap load and a resistor divider bias setup like we normally use. So you might try this, just as an experiment until the actual failure can be found out.
- set up your reference voltage divider to put Vref into the middle of the power supply range.
- instead of a single cap to ground bypassing the Vref string, use two capacitors, one from V+ to Vref, on from Vref to ground, of equal values. Try 22uF just for grins if you have them. Now the Vref both statically and dynamically tracks 1/2 V+ as V+ rises, and the caps force it to try to stay in the middle on the way up.
- connect the "ground" end of your filter cap to this dynamic Vref. Now the cap will be trying to force the TL072 input to *follow* Vref, instead of holding it to ground.
- a largish resistor from TL072 input to Vref may still be needed; I'm a belt and suspenders kind of guy where imponderables are involved.

I have one more question, is there any advantage to use a TL072 buffer over the inbuilt darlington buffers? I guess I should use my ears to answer that question.

There are some minor ones. The 072 (or other FET input opamp) does a better job in some situations. The synth community universally uses an opamp - usually an 072! - for setups like this instead of the internal darlington. The darlington is usually good enough for musical stuff, except that it has a more limited input range and output range, and certain minor input issues. Small stuff, usually. The opamp lets the OTA work more like the classical equations say, and where you're trying to get the 1000-to-1 range of audio, that can be important. Secondary stuff. The darlington is not where the hiss is the problem.

Hiss in OTAs is often an artifact of the way you have to treat the signal. OTAs without distortion precompensation can only handle signals of about 25mV. So you take your 100mV to 1V guitar signal and run it through a resistor divider to get it to be small, run it through the OTA and amplify it up again, perhaps with the OTA. This is the wrong way to get low noise. For low noise, you never attenuate a perfectly good signal, the amplify it back up unless you attenuate noise in the signal more than you add with more amplification. Using the distortion precompensation diodes in the LM13700 - and I think in the 3280 and 5517 as well - let it handle signals up in the 1V range without objectionable distortion, so you don't have to pad the signal down as much, and may get better noise performance too. I have not delved all that deeply into this, just know what I've read about them. (That translates as "I haven't tried this in an apples-to-apples comparison.")

Thanks for your comments RG, I really appreciate it, I went too far out on a limb with this one, beyond a full understanding of what I was dealing with, which is usually a good recipe for some sort of disaster or setback, and then bounce back again.

However, you have and are still learning a lot. Education is always expensive.

I once heard successful parenting defined as making sure the little dears' mistakes were nonfatal.






[/quote]

[Ronan]

I have put the 2 x 22uF caps across the voltage reference, and left the existing 100n to ground
in there, but no change.

I then tied the filter caps to the Vref instead of ground. No change in stability, but the first
impression I got was that the sound quality seemed to improve...I'll do some more A/B testing
on that this weekend.

Tomorrow night I will fit the resistors from TL072 input to Vref, I have run out of time tonight.

Thanks for the help, spare time during the week is short, but doing my best here 

[ORK]

How about some output caps on IC1a and IC1b, and if not so, connecting those 22k resistors at their outputs to Vr instead of to ground?

[Ronan]

I added 470k resistors from the TL072 input to Vref, and also tried connecting those 22k resistors at the TL072 outputs to Vr instead of to ground, but still it latches up. I also tried removing the BP/LP mix pot, a different power supply (analogue instead of switchmode), a 1n cap on the input, and various amounts of feedback, but no joy.

Time to go back to the breadboard, and start again, I believe. With all the insights on how OTA's work, the datasheets are making more sense, which will be a big help.

In the meantime, I'll just keep playing the wah with the TLC2262 in position IC1 (next to the BP/LP mix pot). That seems to be the simplest unconditionally stable fix at this stage...

And thanks again for all the help!