We've Sprung a Leak! Cloning the Very Precise Cornish G-2

It's been, what, three weeks since I posted a clone of a Big Muff Pi variant? Must be time for another one! This time around, I'm building the AionFX Cygnus, a clone of the Cornish G-2. This particular build has honestly been one of my more challenging builds for a variety of reasons. First off, it's a Cornish pedal, and Pete Cornish's designs are always absolute works of art with an amazing transistor-based buffer design. The component count is usually very high as well, and the circuits are typically extremely well balanced, and as long as you have high-quality components you can usually guarantee a good build from the outset. However, because the G-2 uses Germanium diodes for the clipping stages, that adds an additional level of complexity to the build, which I'll go into later. As you can't touch a real G-2 for under $500, this one screams for a good clone.

As with many of my recent projects, I've built two of these - one to keep and one to share. The parts count is typical for a Cornish pedal, much of which is actually not the effect itself, but the buffers! I used all 1% tolerance resistors, mostly Yageo. The capacitors are all high-end metal film, mostly WIMA and Kemet. The electrolytic capacitors are all Nichicon or equivalent. When it comes to the components that actually shape the pedal's tone like transistors and diodes, I always try and mimic the original as closely as possible. This has sometimes gotten me into trouble, and as it turns out this ended up being one of those times!

Rather than using the American 2N5088 transistors, I used BC549C transistors like the original. The pinout of the BC549 is reversed from the 2N5088, but as AionFX always labels the emitter, base, and collector for transistors on the silkscreen, I'm always able to get them aligned properly using my Peak Atlas analyzer. The real challenge on any G-2 build comes in the diode selection. The build notes for the Cygnus go into great detail on the electronic properties your germanium diodes need to function properly in the circuit. Exact values vary, but most germanium diodes have a forward voltage of 0.2V to 0.4V. The Cygnus / G-2 in original configuration needs diodes with a forward voltage of at least 0.3V. Otherwise, you will have to substitute the diodes out for silicon Schottky diodes (in which case, why build a G-2 - just build a P-2 or other silicon-based clone) or change the resistance of R18 and R19 in the first clipping stage so there is enough signal gain for proper distortion in the second clipping stage.

No problem (I think to my naive self), I can just check the forward voltage of the 1N34A germanium diodes using my digital multi-meter, hand select those above 0.3V, and Bob's your uncle - instant awesome G-2 clone! Well, it didn't quite work out that way. I got the pedal together, boxed and on to my spaghetti tester and I immediately noticed something wasn't quite right with it. I mean it worked, but the volume was incredibly low. I had to crank it all the way up to just get back to unity. Once I was there, the pedal actually sounded pretty good, though I also had to crank the distortion a fair amount to get a good signal. I watched a couple of videos of people playing actual G-2 pedals, and they tended to have both knobs past noon as well, but it was clear that there was something else going on limiting the output of my pedal.

It turns out germanium diodes have another property I wasn't entirely aware of which makes a huge impact in this particular circuit. It's something I'm aware of from my mundane life in the semiconductor industry call leakage current (a big shout out to Benny and Chuck from the PedalPCB forum for pointing me in the right direction!). It's actually a little ironic as I've spent a fair amount of my professional life working on high-k dielectrics for semiconductor devices to reduce leakage current and enable the continuation of Moore's Law, and leakage current is what was killing my G-2 builds.

Based on the instructions posted by Chuck on the PedalPCB forum, I was able to construct a simple diode leakage current tester using a 100K Ohm resistor, a 9V battery, and a breadboard (photo a couple of pictures above). As I couldn't test the diodes I'd already soldered into my pedals, I grabbed some from the same batch and checked their leakage current. Oh dearie... dearie my... Their leakage current was extremely high which was likely killing the volume of my pedals. So, I went for a deep dive into my collection of diodes and began looking for alternates.

After testing about 30 diodes I found that my collection of 1N270 germanium diodes had the best leakage current. So I selected eight (four of the small form-factor, and four of a larger vintage form factor), that all had leakage currents less than 0.5μA. I then tested the forward voltage on all of the ones with low leakage current. As it turns out, the were all to the low side with an average of about 0.25V. So now not only did I have to try and de-solder and replace four diodes per board, but two resistors as well.

As serial number 2 hadn't been boxed I started there. To be honest my de-solder and re-solder jobs were a bit of a dog's breakfast, but once together the pedal actually worked much better than serial number 1 with the original 1N34A diodes. I therefore performed the same replacements on serial number 1 (above) and it came out looking just great. I need to put them in line and see if there is any noticeable volume difference between the two because the leakage current on the vintage diodes on #1 is just a little higher than the leakage current on the new diodes on #2.

For these first two enclosures, they're a direct homage to the Pete Cornish originals - by Eggeaux Steggo of course. The tone of a G-2 is more bass heavy than most other Big Muff circuits. Honestly I sort of like it. AionFX provides alternative capacitance values for the tone stack to make it sound a bit more like a conventional muff, but again, if that's what you want, why are you building a G-2 in the first place? Now that I've worked out how to actually make the circuit come out properly, I may build up a few more of these because it isn't a build for the faint of heart!