Newbie; help with diodes

[BadIdeas]

OK, so I am new to electronics and forums and I was hoping to get some help from the community. I wanted to know about different diodes used for clipping in overdrive/distortion. I understand what they do and that the forward voltage affects how they sound. Lower voltage clips more of the wave, sounding more distorted but allowing less volume, higher voltage allows for more voltage but obviously requires a bigger signal to clip very much. I know germaniums have low voltage, leds have a higher voltage, and silicons are in between. I have heard it said that germaniums clip "softer" than others. I have also read that junction capacitance(?) affects the sound, but don't know why. Can anyone provide me with more information? Where to find datasheets? The learning process has been slow for me due to my employment situation, but hopefully that will change soon. I'll just trade in my spare time for cash. 
Any help will be appreciated.

[PRR]

> junction capacitance(?) affects the sound, but don't know why.

Learn "filters", especially R-C filters.

A capacitor is infinite ohms at zero frequency and zero ohms at infinite frequency.... in-between it is in between zero and infinity ohms.

Paired with a resistor, you can contrive a filter which is flat to some freuency, then rolls-off the highs. (And the opposite and many other variations.)

Paired with a diode... the diode is near-infinite ohms at zero voltage and near-zero ohms at several volts. So the total resistance (impedance) varies both with voltage and frequency.

However, unless circuit impedances are VERY high, or you use fat rectifiers, I kinda doubt junction capacitance matters much in audio.

And 5pFd, 50pFd, and 500pFd capacitors are a dime each in bulk, not much more buying one-sies/two-sies, so you can add "fake junction capacitance" easy/cheap.

[Hides-His-Eyes]

I too would very much like to know the electrical reason for germanium diodes clipping 'softly'...

[BadIdeas]

I understand how basic hi-pass and lo-pass filters work. So junction capacitance just means that a diode functions as a capacitor as well as a diode? Datasheets I have found show capacitance to be around .8-1pF for 1N34a, .8pF for 1N270, and 4 pF for 1N4148, so in a basic distortion circuit a 4148 would form a lo-pass filter below 120Hz if there was an impedance of 330K. Is this correct? Of course, these same datasheets show a forward voltage of 1V for the germaniums, but I have read elsewhere that it is more like a quarter or a third of a volt, so I don't know if I am finding reliable information.

I also don't know how to find impedances for caps and resistors in parallel, I just sort of guess based on how to do resistors in parallel. I've learned most of what I know from Electronics for Dummies, but it is a bit limited. I have my eye on another book, but am putting it off until I get some practical experience. Otherwise, it's just a bunch of theory and I would probably lose interest.

[merlinb]

4 pF for 1N4148, so in a basic distortion circuit a 4148 would form a lo-pass filter below 120Hz if there was an impedance of 330K. Is this correct?

Not quite!
4pF and 330k form a 120 kilohertz filter! Looks like you got your decimal point wrong somewhere.

Diode capacitance is so small that it can be ignored- it does not have any significant effect in a clipping circuit. Only the shape of the diode conductance curve matters.
You can see for the 1N270 that it is quite gentle, and the straighter part of the curve is around 0.15V. Ge diodes are usually estimated around 0.2V drop, so it's not far off.


A silicon diode looks more like this. It has a sharper 'knee', and a forward drop around 0.6 to 0.7V. (Actually this one is more than 0.7V, so it's probably a power diode like a 1N4001. Best I could find on Google I'm afraid)

[PRR]

Ya, 120,000Hz.

But note that a diode's capacitance varies with voltage and current. They like to quote with significant reverse voltage (where it matters for rectification). As it starts to conduct the capacitance rises a lot, but this is normally masked by the resistance dropping.

Here's the caveman argument. Crystal diodes are good rectifiers for MegaHertz. While you can show some error at 120KHz, down in the audio band they are "nearly perfect". Yes, common 1N4001 can blow-up used in 25KHz switchers, but not because they don't deliver most of the power, only because they hold-back just a little more than they can stand.

> more than 0.7V, so it's probably a power diode

0.7V at 10mA, 0.84V at 100mA, would be a very _small_ diode. (Or a made-up graph.)

The slope of the hi-I portion is (0.84V-0.7V)/(90mA-10mA) or 1.5 ohms. As a "1N4001" rated for 1 Ampere that would give 2V drop which IIRC is about twice what it should be.

Ge is a wheezy semiconductor. When "off", it still conducts a lot. When "on", most commercial Ge doesn't conduct as well as we like. This softens the transistion from off to on.

The bandgap for Ge is 0.67V; for Si 1.1V. Useful conduction tends to start at half these voltages. This is fairly true on Si, but Ge tends to be conducting a lot sooner (as seen in Merlin's graph).

There's also fabrication. You can make a Ge diode from coal scum (coal impurity is still a major source of raw Ge). Si has to be REALLY-REALLY pure before you dope it.

A near-Ideal Ge should clip as sharp as a Si (at half the voltage). I think what you are hearing is the sloppy fabrication on most older Ge diodes.

[BadIdeas]

Yeah, I think I got my picos and nanos confused.
Cool graphs, Merlin. It looks like there is some "swing" in the voltage drop depending on how much current is moving. I believe the datasheets for the germs were telling the maximum voltage drop was, or what it would be at a certain current or something. I am still not sure about leds. Does their curve resemble silicon or germanium more?

I've got an assortment of leds, an assortment of silicons, and I just got paid so I think I'll order some 1N270s and some 1N34as, then start comparing them in my DS-1.

[merlinb]

I am still not sure about leds. Does their curve resemble silicon or germanium more?

And if you want a really tubey overdrive then you can use a tube:

[BadIdeas]

So it looks like the lower voltage leds tend to clip harder than higher voltage leds. I have an idea in my head of a sort of "backwards tubescreamer" type of stage driving a clipping stage. Instead of placing the diodes across the gain control, I want to put a 1N4148 against a 1N34a, and place them across the resistor and cap going to the reference voltage, in series with a pot and resistor. In theory, the diodes turning on would actually increase the gain rather than decrease it, making more of a sawtoothish wave. Then I can put this wave across an asymmetrical clipping stage after the opamp output. Supposedly, if I adjust the values just right, it will allow for tons of touch sensitivity. Does that sound like something that would work?
It will be a while before I get to try it. I'll get some more experience with advanced mods, then maybe a homemade fuzz, then I'll see. 
By the way, my leds don't say if they are superbright, kindabright, or what. The package says Velleman, and has 3mm reds & greens, and 5mm reds, greens, & yellows. Oh, and the contents list says they are all "TD Universal". Is there a way to find more specific information? Do the size or type even matter?
Thanks for all your help so far, guys.

[merlinb]

I decided to measure some diodes today. In particular, we are interested in the low-current region, which is rarely shown in most graphs.

You can see that the shape of all the silicon/LED curves are basically the same. These diodes types basically produce the same sound, the only difference being how much headroom you get before clipping.

The Ge diode has a slightly softer switch on, but perhaps not as much as the snake-oil peddlers like to suggest.

The biggest difference, naturally, is between solid state the valves. It's pretty easy to see why creating a convincing valve distortion with solid-state diodes is an uphill waste of time struggle.

[BadIdeas]

Wow, you did all that yourself? That's a good-looking graph. Curious about the tubes, though. I thought they were used differently.
I've got some germs coming in the mail, as well as parts for a TB effects loop and stuff to hopefully turn my phaser into a (sort of) EVH model. Looks like I will be having some fun shorly. 

[Hides-His-Eyes]

definite different slope on the Ge, phew...