First tool [1] is for creating spice model from measured voltage and current values. You pass 30 or so measured voltages and currents it it will generate spice model, for example: .model D1N4007 D(IS=2.24679e-9 N=1.76869 RS=0.647905)
Second tool [2] is for finding similar diodes. Let's say you found your mojo diode and you want something similar. You measure 30 or so values from 1uA to 25mA and paste the voltages and currents into the tool, then click on "Similar 3" button and it will display the curves that are most similar to that. The only problem is that there are not that many diodes, so feel free to contribute your curves via github.
Nice visualization! But the optimizer could use some work. It wasn't really able to handle the data I had lying around for an 0402 hyperbright white LED (EAST1005WA0) even though this should have been in its ballpark. (Though this LED is particularly "soft" which might have been troublesome for the basic 3-parameter model?)
If you want to improve things the change that will probably get the biggest benefit for the least work is to have the optimizer work in log Is instead of Is directly. (At least, that helped me a lot when I did the super-basic 2-parameter version of this in Excel for somewhat cruder purposes.) If you're up for heavier lifting, because this is only a 3-parameter model, you can probably write out an explicit or approximate explicit solution for 3 input data points, then expand that into a least-squares solution for N data points, then use the actual heavy machinery optimizer as the final correction step. It will run very quickly if you give it a good initial starting parameter set.
That parameter set will also be a decent starting position for the more detailed models used by LTspice and friends, should you like to go there.
I’ve got a few blog writeups of the more complicated ones, too. The Tube Screamer is the most popular but the Boss Graphic EQ or the DynaComp compressor are my favorites.
This seems to be my recurring theme - reading the title and misinterpreting the meaning, as times have changed.
The first thought I had is that we have some metal-can germanium diodes [0] where I clip the top off, and make them photo responsive. This was a practice in the 70s and 80s, mostly with metal-can transistors, but it worked with various diodes too. The packaging was tiny metal can (aka cap), with the lid (aka baseplate) providing the support for the germanium and leads. If the can was removed, often the level of light can impact the gate's ability to transmit current. A real quick way to get yourself a photo-diode when it was in short supply.
If anybody is interested about clipping diodes please make sure to checkout the beautiful resource that are the lab notes of AMZ: http://www.muzique.com/lab/main.htm
This is one of the things I keep a downloaded copy of because it would be a shame if it ever got lost.
These explanations take the diode out of its context, view it in isolation from the overall circuit which is always a mistake, the frequency response of what is before and after the clipping has a greater affect. If you want to understand how different types of diodes affect the sound you need to look at the harmonics they create and how the input and output circuitry affect the harmonics created by the clipping.
> These explanations take the diode out of its context,
Which turns out to be one of the cornerstone of the scientific method: isolate a tiny part of a complex system, understand its behavior thoroughly and in depth if you ever want a shot at understanding how the complex system works and how the tiny parts contributes and interacts with it.
The scientific method is a very lousy way to learn something which is well established and understood. Go learn Maltese through the scientific method, I will get a book or teacher and be conversational in it while you are still identifying diphthongs. More importantly this article is a poor foundation for scientific understanding being built on very subjective language use instead of verifiable data.
PS: What I don't see on these pages is the variation of one or many diode pairs to ground ending with a variable resistor as a "softness" control. Can't recall the source of this idea, but it's probably R.G. Keen.
There are various projects around in which diodes are put in series with resistors, either alone or groups of them to achieve softer clip, plus asymmetric clipping when different number/nature of diodes are used; experimentation can lead to very interesting results. At one time I simulated using tunnel diodes too, which according to the simulation should have produced some unique waveforms due to their negative resistance, but when built my prototype didn't work at all. I had no experience with them however, so I may retry one day.
I was interested to see that asymmetric distortion is achieved by using different numbers or types of diode in the feedback path. I was expecting this to be achieved by adding a DC offset - i'm wondering why this isn't how it's achieved?
Offset is used sometimes but a diode is cheaper than the two resistors and cap needed for that offset voltage. More common on old transistor based pedals but their offset is really their bias, already there so might as well use it.
This is what bothers me with diode comparisons in general. The forward voltage is obviously the leading factor in the clipping but is actually the easiest to adjust through biasing the diode. It would be far more interesting to bias them all at the same forward voltage and then see how the other characteristics (leakage, capacitance, recover time, etc.) impact the waveform.
Yep. I opened the comments for this same reason. After you've spent a lot of time with DAWs looking at waveforms, there's a very strong connection between how something looks and how it sounds. This is very much missing that.
Perhaps even more obviously missing - sound samples of the same input through various circuits.
First tool [1] is for creating spice model from measured voltage and current values. You pass 30 or so measured voltages and currents it it will generate spice model, for example: .model D1N4007 D(IS=2.24679e-9 N=1.76869 RS=0.647905)
Second tool [2] is for finding similar diodes. Let's say you found your mojo diode and you want something similar. You measure 30 or so values from 1uA to 25mA and paste the voltages and currents into the tool, then click on "Similar 3" button and it will display the curves that are most similar to that. The only problem is that there are not that many diodes, so feel free to contribute your curves via github.
[1] https://dvhx.github.io/spice-diode-model-js/index.html
[2] https://dvhx.github.io/spice-diode-model-js/find.html