In most cases, people use diodes, and pose diode-related questions, for purposes of clipping and distortion.
This is getting to the point of needing an FAQ, since I've written this enough times, but here's the scoop (or at least part of it) on diodes.
You can, in principle, stick a Ge diode where an Si one goes, and vice versa, without fear of destroyng anything, but it will work differently. Whether this difference is more or less of what you want is another thing.
Here is a case in point.
The MXR Distortion+, and its various clones by DOD, Ross, etc., uses a pair of back to back Ge diodes to clip the signal it receives from the op-amp booster stage. Ge diodes will conduct in a manner so that anything in the signal greater than about 250mv will be "clipped". The clipping of signal is what we hear as distortion. Using a back to back pair means that both the negative and positive half-cycles will be clipped equally.
Two hundred and fifty millivolts isn't a very hefty signal, though, so the output of the Dist+ isn't that substantial, meaning that although you can get it to give you buzz and fuzz, you can't really overdrive your amp with it.
If you were to replace those Ge diodes with Si type diodes (1N4148, 1N914, etc.), you would hear the following differences.
1) Because Si diodes generally (more about this below) don't clip signals until about 500mv, the "ceiling" on the output of the Dist+ will be raised, and in this case, almost doubled. So, you will have more volume. I have a little Dist+ I made, with a switch to select between Ge and Si diode pairs, and the difference in output level is quite noticeable.
2) Because the "clipping threshold" is higher - i.e., the signal has to be above 500mv to be clipped - so the stock Dist+ circuit will result in less distortion if you replace the Ge's with Si's. Again, on my own, switch to the Si's and you get not only more volume but a less distorted sound and a little more dynamics (more spread between the quiest and loudest sound produceable).
3) The quality of clipping in Ge vs Si circuits is said to be different for equivalent volumes, because of HOW each type of diode clips. Quite frankly, I've never heard it, but then I've never run a comparison at sufficient volumes to be able to tell (I generally don't play above table radio volumes in my basement). My sense is, however, that this is a real phenomenon and not just one of those "golden ears" illusions. That being said, the differences that exist may be wiped out by one's speakers.
To some extent, some of the aspects of tube-type distortion can be emulated by use of asymmetrical clipping - one half cycle clips more than the other. One of the ways people often do this is to combine Ge and Si diodes in a back to back pair (each type oriented opposite to the other). Given that each diode has a different clipping threshold, one half cycle will clip more than the other.
The clipping threshold set by diodes is additive, meaning that if you were to put two Ge diodes of the same orientation in series, in parallel with two other Ge's of the opposite orientation, you would have symmetrical clipping, with Ge quality, but with a higher clipping threshold (double, or roughly 500mv). Similarly, stick a Ge and Si in series, and the threshold (for that half cycle) is now 500+250=750mv. You will sometimes see circuits that use pairs of 2 and 3 diodes in series.
This is why you will often see combinations of diodes in diode clipping circuits, and occasionally mixes (e.g., the Fulltone version of the TS-9 uses a 1N34, 1N914 and 1N4001).
Some time back, I introduced a term here: proximity to clip. Since what diodes do is prevent signals passing through them from being higher than a given level, they do very little to signals lower than that threshold. Normally, distortion devices provide boost that puts the input signal closer to the signal level the diodes will clip at - i.e., greater proximity to clip.
When you switch Ge to Si diodes, you reduce proximity to clip (less proximal = farther away) since you've raised the clipping point. You can restore this by introducing greater gain BEFORE the clipping stage.
Conversely, if you switch from Si to Ge, you increase proximity to clip. As mentioned above, the lower clipping threshold will mean lower output level, so you may want to stick another gain stage after the clipping stage.
BE AWARE that the stated clipping thresholds of different diode types are an average/typical value. Personally, I have found reasonable variation in voltage drop (clipping point) within diode types. You can buy a bubble pack of fifty 1N914's from Radio Shack, and some will clip at 470mv while others may clip as high as 650mv. Same is true of Ge's. In my experience, they range from 190mv to 400mv. Ideally, if you are planning out some combination of diodes to provide a certain flavour of distortion (e.g., asymmetrical) or degree of distortion (e.g., less overall proximity to clip, given the gain struture of the pedal), then you probably want to measure the diodes in question and work with knowledge of their values. For example, would putting two Ge diodes in series, and having them in parallel with a single reversed Si diode provide symmetrical clipping? Not necessarily. The Si could clip at 450mv, and the combined Ge's would clip at 600mv.
It is my view that sometimes, these tolerances and variations in diodes are part of what makes a given pedal sound better or worse than another copy of the same pedal. I doubt that manufacturers went to any effort to select diodes beyond the part number, but YOU can.
Hope this clarifies some things and gives you some ideas for
mods you might be interested in doing.