TL072 Strikes Again / Unity Gain Buffer Hassles

[R O Tiree]

Am I just being dumb, here, or would a 1.8V Zener, say, inserted between the Ground pin of each OTA and true circuit GND provide the separation needed?

[R.G.]

Not dumb. This is one variation of the "give the TL072 a more negative power supply". The only difficulties are making it enough negative and stable enough, where "enough" varies from circuit to circuit.

[Paul Marossy]

To start with, Paul, IC1 is the one that needs to be swapped out, the one closest to the trimpot for mixing LP and BP. The other TL072 isn't affected (IC4). Until or unless a better workaround evolves...need to think about it.

I tried a TLC2252 for IC1 and a TL072 for IC2 and it did not work at all. I have TL072s made by three different manufacturers and none of them work. It will work with the TLC2252 and an LM1458, an LT1013 or an NE5532. So now it has the TLC2252 and an NE5532 in it. Those are only marginally quieter than the LM1458s are. Looks like the hiss comes from the CA3080s then. It's not that bad, was just trying to get it as quiet as humanly possible.

Everyone's favorite dual opamp seems to be the TL072, but I hate those things. I'll stay away from them from now on.

[Cliff Schecht]

So here's a caveat for you Paul. You are using one of the very standard OTA-based filter topologies (even if it's a variant, it's still OTA-->cap--buffer). The buffer circuit needs to be very high impedance for the input as to not do ANY loading on either the cap or the OTA output, both of which are sensitive to "stray" resistance. If this part of the circuit sees too low of an input impedance then your filter will start to fall off as it tries to force current into the capacitance. The easiest way to get a very high input impedance is to use a JFET or MOSFET based op amp (most MOSFET stuff is low voltage though) but these inherently have the problem of phase reversal. RG already mentioned going to a rail-to-rail op amp but he wasn't specific enough to help you narrow down your search through the sea of op amps designs that are out there.

The only op amps I know of the top of my head that can do what you want is the OPA131/2131/4131 series which are $3.50-$5 a pop (minimum) depending on what performance grade you buy. They are some badass Burr Brown parts and have worked great for me in the past but this is prohibitively expensive for DIY purposes. Maybe someone else knows a cheaper alternative that is guaranteed to work at the extremes of the common-mode input.

You should be able to run the TL0xx series op amps in a unity gain buffer situation and be fine though. What you need to do is make sure that the resonance circuit is limited so that it can't self-oscillate from rail to rail. Some limiting diodes in the feedback loop like what is done on the MS-20 filter would do the trick, or just make sure that there is enough gain to get positive feedback and resonance but not the crazy rail-to-rail resonance that will likely send the TL0xx series op amps into the rails. Or (as I think someone mentioned above) you can clamp the output of the resonance circuit so that it can't drive the filter buffers past their CM range.

[Ronan]

Thanks for the suggestions Cliff. Would these IC's be a good choice do you think?

TLC2262 $1.70 Small bear

http://www.ti.com/lit/ds/symlink/tlc2264a.pdf

TLC2272 $1.83 Digikey

http://www.ti.com/lit/ds/symlink/tlc2274a.pdf

In the schematic, IC4 (TL072) is a plain ordinary buffer and 4.5V supply, I don't understand how that IC (in that position) could latch up. The other TL072, IC1, uses both halves in the OTA circuit and I have duplicated the latchup in that one, but not in IC4. In my build, it will only latch up when the circuit is powered up with the pedal in the treble position, supplying max current to the OTA's bias pin, then unlatches as the pedal is moved to deliver less current into the OTA's bias pin. I do understand from RG that the latchup will vary from build to build, mine may well be a best case scenario. As far as I know, only me and Paul have built the circuit so far.

[Paul Marossy]

In the schematic, IC4 (TL072) is a plain ordinary buffer and 4.5V supply, I don't understand how that IC (in that position) could latch up.

Mine locks up with a TL072 anywhere in the circuit. Must be dependent on the OTAs that you use I guess. Mine are supposedly made by Intersil. They work OK, and I'm not sure how much of a factor this is (probably is due to wah pedal travel more than anything else), but with no active devices installed I couldn't get my heel down voltages that close to what you have listed here (I used method 1):

Here's some more info to help set up the two trimpots in the corner, this is the bit of info I hadn't written yet since I thought no-one would have a PCB for another week:

There's 2 methods to set the two trimpots next to each other in the corner of the PCB, these trimpots adjust the upper and lower frequency limits of the pedal travel.

Method 1: With NO IC's fitted to the circuit, power it up and measure the voltage on the middle lug of the 10K wah pot to ground.
Put the footpedal in the toe-down position and adjust the trimpot closest to the corner of the PCB for 3.0V.
Then put the footpedal in the heel-down position and adjust the other trimpot for 0.74V. Recheck the toe-down voltage as each trimpot affects the other. Then recheck heel-down voltage and so forth until the voltages are close.
Then remove power and refit the IC's. The voltages measured at the center lug of the wah pot should drop to around 1.2V for toe down and 0.68V for heel down with the IC's fitted. This should get you in the ball park. After that, tweak to get the sound you like.

Method 2: If you soldered the IC's directly into the PCB, then unsolder the wire from the middle lug of the wah pot, power the PCB up, and set the voltages to 3.0V and 0.74V as in method 1. Then remove power, resolder the wire back on the middle lug of the wah pot and check the voltages again for around 1.2V and 0.68V.

If the trimpots aren't set up like this, there is a possibility of damaging the CA3080E's with too much voltage on the bias pin.

[Cliff Schecht]

Thanks for the suggestions Cliff. Would these IC's be a good choice do you think?

TLC2262 $1.70 Small bear

http://www.ti.com/lit/ds/symlink/tlc2264a.pdf

TLC2272 $1.83 Digikey

http://www.ti.com/lit/ds/symlink/tlc2274a.pdf

In the schematic, IC4 (TL072) is a plain ordinary buffer and 4.5V supply, I don't understand how that IC (in that position) could latch up. The other TL072, IC1, uses both halves in the OTA circuit and I have duplicated the latchup in that one, but not in IC4. In my build, it will only latch up when the circuit is powered up with the pedal in the treble position, supplying max current to the OTA's bias pin, then unlatches as the pedal is moved to deliver less current into the OTA's bias pin. I do understand from RG that the latchup will vary from build to build, mine may well be a best case scenario. As far as I know, only me and Paul have built the circuit so far.

Those op amps both have input common-mode ranges that extend to the negative rail. They don't state anywhere that the op amps are latchup free but they probably would work. But the CM input range to the positive rail isn't very good and this could be problematic depending on the application. In a buffer application this means that the positive swing of the op amp is going to get clipped at 1V to 1.5V lower than the negative rail. There are probably better op amps for this application but I bet this one is good enough, these OTA based filters tend to not use the entire dynamic range that is available anyways (to avoid excessive noise/distortion from the OTA's). You could give it a try if you have some of these available.

As far as IC1 vs. IC4 in the schematic, IC1 would only latchup if you drove the crap out of the input. A typical 100mV to 1V input from guitar should not be causing any problems. Excessive drive from an external pedal or sig gen could cause this.. IC1 is the chip that is capable of latching up. The easy solution to this problem is to clip signals before they ever get to the point of locking up the chips, this is easy to do with a handful of diodes. You could also power the design off of +/-9V and completely avoid latchup issues..