Originally posted on the Freestompboxes forum
This is my version of the EVH overdrive by Bajaman. Just to be clear, I have nothing but respect for him, all the work he has freely shared and which has become popular in many ways such as between sold PCBs. I don't claim this version is better or worse than the original, I just liked the idea and wanted to get to the same goal through my own means, using the circuit as inspiration and with the goal of making it more compact and easier to build without compromising anything.
Let's go through it:
Removed the JFET input buffer. JFETs are expensive or tiny these days, and the 074 can do the job just as well. Input impedance is 1MΩ, input cap plenty large, input RF filter and current limiting have been scaled down for lower thermal noise while still avoiding pickup interaction. 68k was a common value for valve grid stoppers.
Input stage scaled for more E6, resulting in an error of maximum 0.2dB. Capacitors go to ground, resistors still go to Vref, because I like to save caps too :).
Gain control network: impedances scaled down, removed the resistor in parallel to the gain pot, pot made log taper, removed input cap. The bias for the next stage was already coming partly through the gain pot, but JFET op amps make this acceptable. The other values have been tweaked to get a very very close response across all the pot travel.
Here's my bold move: before hitting the LED, we have three stages. Of those, IC1b and IC2b have some frequency-dependent gain characteristics, while IC2a features a gain of 2, but it's after a 0.68x attenuator with flat frequency response (the low-pass before IC2a cuts in the hundreds of kHz). Attenuation is before amplification, so headroom isn't affected, only noise. I thought i could make those three stages into two, and so I did. At the same time I saved some components along the way (the 33k and 10n at the output of IC2a aren't doing anything that I know of) and re-scaled to my E6 taste.
You thought that was a bold move? The thing I found most interesting about the baja EVH was the idea of separately clipping the two half waves in different stages, but I really wanted to make this a 4 op amp circuit if I could. So I went looking at what's in between the two clippers. There's an attenuating low pass filter with an attenuation of 11 and corner frequency of 17kHz or so. I thought I could ignore the latter part. Following it there's IC3a, with a gain of 13 and flat response (low-pass at 49k). Combining this with the attenuation before it, since the series resistance is the same 2.2k and the diodes never conduct at the same time (I think we can say that ), the transistor diode is getting a signal 1.18 times bigger than the output of IC2b. In the realm of dB and gains in the neighborhood of 70dB, that's not much, so here's my fudge and maybe something you can lay blame on if the clipping (not the frequency response!) isn't the same: I just put the diode together with the led. Boring, but it works and seems reasonable to me. I've used a 4148 here, use a transistor if you want, personal choice. Running some simulations both for dc transfer and with transient, the visible difference is that the positive peak is limited to 1.6V instead of 1.88V, 1.18 times less, which is obviously because it doesn't get that little bit of amplification back. Since this difference is pretty close to the forward voltage of a schottky diode, I decided to do just that, and here's the result in DC transfer and transient with a 3V 1K sine, which is realistic given all the previous gain, and shows both clipping thresholds. Using a green LED instead of these two is also a viable option, and leads to the same results. This isn't the only way to go about compensating it, but I followed the dc transfer curves, which compare my clipping setup with the one with the attenuator and op amp in between, as main guide for both the clipping treshold and the output amplitude.
Going on, there's a “buffer” stage with 2x attenuation followed by 2x gain. This goes away. The JFET buffer goes away too, for two reasons: first, the output impedance driving the tone stack is in the worst case 2.2k, even ignoring the diodes, and it would be happy even with more. Second, unless I'm wrong, in the original amplifier, even considering some negative feedback, the tone stack is driven by a series 470K resistor. That's actually a bit nuts, given that we're told to drive tone stacks from low impedances. So who do I follow, Peavey or Baja? I went with Baja in giving the tone stack a low impedance. I went with a 25k pot for Mids because saving one cheap resistor makes me oh so happy, and taper is the same if not better. Of course if you have a 50k pot the parallel is the right thing to do.
Lastly, I went for tone stack “defeat” instead of bypass. The result is the same, but this allows me to use the second pole to switch off the make up gain in the last stage when the tone stack isn't active, so that the output stage can double as buffer. I was conservative in choosing the make up gain to avoid unintentional clipping, but since the loss depends on frequency and settings it's still approximate. Volume pot is 10K log, allowing you to plug this into a power amp for your own solid state 5150, or effects return, giving low output impedance at all settings avoiding unwanted tapering effects from low impedance inputs.
Here's a demo. I just had to try tapping, didn't I.
I had fun making this and I hope you'll enjoy it too. I wouldn't have gone through the effort if I thought the circuit didn't deserve the attention. I'm still waiting for my foreign friend to build the original version for a side-by-side trans-national comparison.