#bd-2

Here’s a schematic to follow along. And here’s a cleaner one that leaves out the wire routing and bypassing.

CLIPPING A lot of mods for this pedal basically amount to “upgrading” capacitors (worthless) and swapping values in the tone stack (will only change the tone control’s behavior). Sometimes you’ll see recommendations for changing diodes to other types, but in most cases this doesn’t work either. Why?

In short: clipping doesn't primarily happen at the shunt diode clipper (D7-D10). The BD-2 has two discrete ‘opamps’ in it, the first made up of Q9-Q11 and the second of Q12-Q14. Since the BD-2 has a dual-gang gain pot, gain changes are actually applied to both opamps simultaneously. Like an amp, the last stage before the master volume clips first -- so most of the actual distortion is the second opamp clipping. It hits the rails WAY before the first opamp reaches the diode clipper threshold. D7-D10 appear to be only for limiting the input to the 2nd opamp stage, so that its input FET gate is not overdriven. They do eventually clip, but only at extreme gain levels where the second opamp is already putting out a square-wave.

Some mods also advocate changing D1 and D3, but again, these are for protection and never clip in normal operation. They’re between the two inputs of an opamp stage that’s set up as a simple +8dB peaking filter (Q of 1.4) at 150Hz, but otherwise is always at unity gain, and it’s after the master volume. So the only way those diodes are EVER going to conduct is with the master volume all the way up, in which case, you’re not hearing the subtle tonal contributions of D1 and D3. They’re best left alone.

There are two ways to clean up the distortion and make it less harsh:

First, you can add some clipping diodes in the negative feedback path of the 2nd opamp (insert back to back, parallel to C25). This is a typical Tubescreamer-style ‘soft-clipping’ arrangement that limits the opamp’s gain once the output voltage reaches a certain threshold. The second opamp has about 8Vpp of headroom before it starts to distort, so ideally you want a clipping network that retains as much of that headroom as possible (otherwise distortion will happen much earlier on the gain pot travel, and you’ll lose all dynamics).

I used two germanium D9Es in series with a blue LED in each direction here, allowing about 6.5Vpp of voltage swing before a soft knee sets in -- this gets close to the previous headroom without allowing the opamp to clip to the supply. I didn’t opt to add in any asymmetry here, to preserve dynamics and a more amp-like feel. But you can use pretty much whatever you want, as long as the forward voltage drop is such that the diodes will hard-clip just before 8Vpp. Note that you’ll still have a lot of crunch and some fizz in the sound, but the harmonics will be much nicer in the mids, and the higher frequencies won’t be as spitty and harsh.

Second, you can lower the threshold of the D7-D10 diode clipping network, so that it actually starts compressing/limiting/clipping the signal around the same time as the second opamp starts to distort. The easiest way to lower the threshold is switching to germanium diodes, with a typical Vf of 0.25V versus 0.6V for the stock silicon diodes. Even fairly heavy clipping by germanium diodes still sounds quite mellow, due to their broad knee -- it’s more like compression/limiting, and it doesn't develop fizzy square-wave harmonics until well after the 2nd opamp is clipping.

However, and this is important: germanium diodes have loads of reverse leakage, which makes them look like resistors (even and especially below their conduction threshold and when reverse-biased) to the circuit driving them. The BD-2 has a ‘fixed Fender tone stack’ network after the first opamp, which basically copies an old Fender amp with the treble at minimum and the bass and mids at maximum. This is a high-impedance network, which is very prone to loading by the subsequent stage. The reverse-leakage of germanium diodes (depending on temperature and humidity) can be on the order of 50-100k, much lower than the values in the ‘tone stack’ circuit, so the diodes will load it down and cause loss of bass. Seriously, try it -- replace D7-D10 with all germanium, and you will notice the bass disappears. It definitely sounds good, but it compromises my initial goal of maintaining the useful stock boost settings. I found that it’s best to use a combination of silicon and germanium diodes to reach a compromise between germanium’s low, soft knee and silicon’s zero-leakage and maintaining bass response. I shot for an overall conduction threshold such that the 2nd opamp starts to clip in the soft middle part of the D7-D10 network's knee.

I like germanium OA9s for D7 and D9 because they have a very low Vf of 0.22V, with much less leakage than most germaniums. I used silicon 1N4001s for D8 and D10, because along with zero leakage, they have a slightly lower Vf than most silicons (0.55V), and I typically like their sound in diode shunt clippers.

Some other germanium choices:

1N270s - 0.25Vf, fairly low leakage

New-manufacture 'workalike' 1N34As - 0.26Vf (varies), very low leakage because they’re often not actually germanium

D9Es (0.26-0.28Vf) enormous soft knee but horrendous leakage (these are the Soviet diode that people like in Klon clones)

NOS 1N60s (0.3Vf) higher Vf and graceful clipping but also bad leakage

1N695 0.29Vf, leakage is ok but still worse than OA9s and 1N270s

The clipping threshold of the stock network is about 2.2Vpp (which is reasonable in a normal design), but the signal is especially low at that point because of the insertion loss of the high-impedance ‘tone stack’ circuit. It requires quite a bit of gain from the first opamp to reach the stock threshold, by which point the 2nd opamp is into fuzz territory. The 2x OA9-4001 network yields about 1.55Vpp, quite a bit lower -- starting to softly clip the peaks before the modified second opamp starts to clip more aggressively, but maintaining clean boost-style operation at lower gain levels.

I added a switchable ‘asymmetry mode’ that shorts out one of the 4001s so that the positive-going threshold is only the 0.22V of a single OA9, yielding a total headroom of about 1Vpp. In this mode, bass seems a little attenuated because of the heavier loading of the single OA9 with no silicon beneath it, and perceived gain drops somewhat because of the limiting before the 2nd opamp clips. But it has a beautiful compressed feel to it, that preserves more of the guitar’s clean harmonics -- with an even softer response than the symmetrical network, and virtually none of the spitty decay of the stock pedal.

BASS RESPONSE Even the stock circuit attenuates bass a little. If this bothers you, or if you’ve opted to go with all germanium in the clipping stage and there’s no longer enough bass, you can modify the low end. A lot of mods suggest changing the input coupler or the tone stack caps, but these have almost nothing to do with the bass response of the pedal. C27 is the only coupler with any significant roll-off: a highpass at 72Hz, which would only affect downtuned or baritone instruments. You can change it to 47nF to make it full range.

The large majority of the bass loss happens via gainstaging. The biggest attenuation is in the negative feedback ground reference of the first opamp (R31-C22). The stock network here is 1.5k-150nF, which creates a highpass at 700Hz! That’s necessary to keep the distortion from turning into a farty mess, but it’s nice to have a LITTLE fundamental in there, especially if you play a baritone instrument. The idea here is not to change the corner frequency of the filter, but to shelve up the lows a little so that they’re not entirely rolled off.

I changed R31-C22 to 3.3k-68nF, which maintains roughly the same corner frequency but drops the gain by about 7dB. I then added a switchable 10k-10uF R-C series network in parallel with them, which when in-circuit shelves up the lows below 150Hz and returns the gain to stock levels (remember that high frequencies will pass through both networks). The same treble shape is maintained in either mode. With this network, turning the gain all the way down and setting the volume and tone knobs to noon results in a sound indistinguishable from bypass. Turning up the gain does get unruly much faster than stock (make it switchable!), but there are beautiful edge-of-breakup clean tones to be had in the middle ranges, very touch sensitive for clean amps that can’t quite get there at a reasonable volume, and useful for a different color of boost to push other overdrive pedals.

Here’s a gain-matched graph of stock and modded responses in opamp 1′s NFb loop (keep in mind this is NEGATIVE feedback, so the graph is upside down compared to output frequency response):

Here’s the same thing inverted, so you can see what the actual output response would look like. Notice that in bass boost mode, there’s about 6dB more output at 100Hz than in the stock setup. At 50Hz, 11dB! Note fundamentals on low-tuned instruments stay intact much better in this mode.

APPLICATIONS My BD-2 ended up with a 3-way on-on-on DPDT switch, set up like this:

Up: asymmetric mode

Middle: normal

Down: bass boost

I find that asymmetric mode is compressed, with more midrange harmonics, and tighter bass. Good for saturated overdrive as well as ‘squishy’ boosting.

Normal mode still has some Ge softness to it from the OA9s in the clipping network (and may still be a tiny bit bass-light vs stock as a result), but can be used as a clear boost just like the stock pedal, fulfilling my first goal. In overdrive, it's much more dynamic and touch-sensitive than the stock pedal, with more harmonic complexity in the mids and less of the spitty decay that bothered me before, which satisfies my second goal.