I like to take circuits I find interesting and rework them according to my taste, following some guidelines, which can be summed up as:
reducing the number of components as much as possible. I think there's beauty and elegance in simple things, and if I can get away with it or have a good reason to, I like to use less components;
standard values. I like to stick to E6 values for the same reasons, and also for accessibility, if I can. Call it a challenge, but if I can get to the same results sticking to standard values, and you usually can, even for the half of the values or less that are actually relevant, so other than the ones that are "plenty big". If the accuracy penalty is too big though, I don't;
noise. I like to apply simple concepts like low impedance design and avoiding attenuation before amplification to improve noise performance. Noise is usually neglected in guitar pedals, but with the high gains sometimes involved, I don't see why that should be the case if it can be helped;
faithfulness. Before all the rest, I strive to keep the circuit sounding as it is unless I have a good reason. People like the sound of these circuits, and I'm not here to argue with that, just to suggest a different way to get to the same goal. This means same gain over headroom and frequency response throughout.
So, the Klon. This isn't addressed to people for which only the original will do, or only an exact copy, I don't wanna argue with those. This is for everyone else who like the sound of the pedal or new circuits. I also have some interesting insights I've collected on the circuit though, and some circuit analysis sprinkled throughout.
Let's go step by step:
Input stage. This is just a buffer, nothing to see here, although the input resistor (maybe ESD protection) was removed according to all the points above.
The three branches. This is the tricky part. As some might know, the effected signal is a mix of three paths by the U2A summing amplifier, with a dual gang pot controlling gain of two of them. Well, that's only part of the story, as I found that R3 and C4 (in my schem) also form and high pass filter with C2 for the clipping path! Of course there's also clean attenuated signal tapped off there by R5, and this part is mostly unchanged, except for R3 being a standard value. Here's the FR of the top branch:
The bottom branch. This one carries clean signal of which the volume is adjusted by half of the gain pot. In the original, the signal to the gain pot comes from a fixed attenuator formed by a 5.1K and a 1.5K resistor, to which are added a 68n shelving highpass (bright cap) and a shelving lowpass cap. After the volume, the signal goes again through a shelving lowpass formed by the 27K in parallel with the 12K and 27n into the summing amp. In parallel to this there's a 22K and a 2.2n. Unlike the other one, this network is not at virtual ground and its output isn't summed directly. Here's an assumption, and it is that this network isn't relevant. In the simulation of the original it's been included and it doesn't contribute significantly to the signal coming from this branch. Another possibility is that the diodes conducting will pull this side to ground, but even with it grounded there's no difference. Maybe it's the other way around and it affects the signal coming from the clipping stage? I don't think so, as the source impedance there is much lower, so I think it's a safe assumption. I find this branch quite messy, with shelving filters which partially cancel each other, and indeed as you can see in the album, the frequency response of this branch in the original is less interesting that it would seem. I decided to rework it into a simple volume control, which is tapered so that the volume closely follows that of the original (important for balancing with the distorted signal), followed by a low pass filter. The frequency response is close enough, at least for me.
and here's both clean branches together:
The clipping stage. I decided to keep the shelving highpass filter of C3 and R4 because it was necessary to get an accurate response. Other changes are the use of standard values and the removal of the stopper resistor. Overall gain and frequency response are very close. If you're interested, the gain pot is pulling double duty: half of it attenuates the signal with R4, the other half is the grounded leg of the feedback divider, setting the gain and also frequency response of the op amp in series with R6 and C5, the latter making an highpass shelving filter especially at high gain.
Clipping. I'm not using germanium diodes here because I don't consider them commonly available, but you can do as you want. I'm a fan of BAT41 diodes because we've proven them to be very close, at least in the VI curve. Another change is to C7 and R8. Their cutoff stays the same, I just like to avoid those big 1u caps if I can. A side effect is that the current to the diodes is further limited, leading to a softer clipping (I've messed with this concept in simulation with transfer functions). You're free to go back to 1u and 1k here. The gain of the summing amplifier shouldn't be affected because when the diodes are clipping (and they will, with their low Vf and the high gain), the source impedance will be much lower with respect to the 47K. C8 is 1u but if you don't care about losing about 2dB at 30hz (and you can, especially with guitar, especially for the distorted signal) it can be 100n.
Here's also, just for curiosity, all three branches together. This isn't particularly meaningful because it doesn't account for the clipping diodes, but it's interesting to see they're close anyway.
Summing amplifier. This has the same frequency response except that the gain is lower and it uses standard values. I guess it's time to talk about...
Headroom. Yes, the charge pump is not there. No 24Vp-p output from this, "only" about 7V, which is still tens of times more than a guitar signal in gain terms, while in dB terms this means just 10dB lower than the original. And we're talking with everything absolutely cranked, a big boosted signal coming in, to even get there, and ignoring that anything after this will get beat to hell anyway. That's the raw voltage output. Regarding headroom, I've made sure that there's no unintentional clipping, especially on the summing and tone stages. The two clean paths add up to a slightly attenuated version of the clean signal at the summing amplifier output, so no trouble there at any gain setting. The clipping stage gets amplfied by 2 and its output is in the worst case 0,6vpp with schottky diodes, so we're far from hitting the rails. All this goes into the tone control, which at most has 5.6x gain at high frequencies. This is actually a modest amount for a clean signal, even at max boost, while for the clean signal it means the output will only reach 3V even if we consider it a full range boost, still far from the headroom of a 072 running at 9V. Slew rate isn't a concern either, since even the slowest op-amps won't struggle to output the full audio bandwidth at the amplitudes allowed at 9V, and the 072 isn't slow either.
Tone control. This one can be considered basically the treble half of the baxandall tone control, and it's pretty much as the original, but with the resistor values scaled down because lower Johnson noise always makes me happy. I've fiddled around with cap and stopper values to get a response very similar to the original.
Output and bypass. The volume stopper has been removed because unnecessary, the volume pot made log for a smoother sweep, caps made 10u which is my standard go to value for low output impedance (nothing wrong with 4.7u). The "anti pop" resistor on the bypass switch have been removed, because R12 and the volume pot already prevent all pops, the 68K and the 100K on the output jack might only help to prevent pops with a broken pedal leaking DC plugged after it. Fix that pedal instead. The bleed of effected signal in bypass, although tiny, seems like a flaw to me and I'm not surely adding components for that. The series resistance on the bypass output is kept, isolates capacitive loads and is good for stability.
Demo
PCB layout
Gerber courtesy of my friend G7D30N. It inherits the CC license of the schematic it's based on according to SA.
Comments