Comparison of Asymmetrical Diodes with Regard to Clipping

[frets]

Hi Guys,

I want to offer switching on my various diode related overdrives on a simple on/on switch.  This would give the guitarist two diode options for crunch on an overdrive.  In order to find diodes with distinct audio differences, I set up a 23 diode pair test.
This by no means is scientific and just my audio observations.  My aim was to find asymmetrical clipping diodes that have distinct sound differences.

There is some reference to this subject on the internet; but, like everything guitar related, it is very subjective.  People will hear subtle differences between different types.  I was looking for very clear differences.

So, I set up the test using a simple  transitor based amplifier and a 23 position rotary. I used diodes that I felt would be decent clippers - subjectively.
Here is the list of diodes compared:
1N34A, MA856, Red LED’s, 1N914, 1N916, 1S2473, 1S133, 1S1588, 1N4001, 1N60P, 1S4733, T5, 1N4148, 1N5819, 1N5817, 1N4007, 1N5918, BAT46, 4V7, 8V2, D9E, BAT41, 1N5231.

Overall, 20 of the 23 sound the same to me in terms of clipping level, tone, decay, and volume level.  3 pairs stood out from the rest.  These distinctively different paired diodes were:
1N4148
1N5817
8V2

The 1N4148’s are brighter than 22 of the diode pairings.  They also have a very slight volume loss. Finally, there level of clipping is distinctly “crunchier” than 22 of the diode pairs.

The 1N5817’s have a deeper tone with slightly less crunch. Volume level is the same as 21 of the pairs (21 pairs were deemed the same in terms of volume with the guitar volume on 7.) 

The 8V2’s were by far the most distinct from the other pairs.  They were the brightest, the crunchiest; but, they have a bit of a volume drop.  Not bad but a bit more than the 4148’s.  There crunch was such that they verged on distortion.

From this unscientific testing, I most likely will using the 1N34A pair and the 1N4148 pair.  I might experiment with the configuration of the 8V2’s and may use those on one side of the switch, with 4148’s on the other side for a more aggressive overdrive.

Those in this group will most likely have good reasons for these perceived differences - forward voltages, dissipation, etc.

It was a worthy exercise as it clarified the pairs I can use on Overdrives to give the guitarist distinct differences.

It must be stated that this little study is wrought with bias and confounds.  But I think if anyone else in the group repeated the test, they would hear the differences in the 3 pairs from the more traditionally utilized diode pairs; I.e., 34A’s, 914’s, etc.

[sumgai]

Cindi,


This is difficult to start, because I need to ask you a question, one that will make me look like I'm peeing in your Cheerios - nothing could be further from the truth!

You are not some dumb blonde who happens to be handy with her hands. You have the ultimate gift, and that is curiosity. In everything we've seen you do, you literally need to know what's going on, and how you can harness that for your benefit. I can only wish that more people in general had that mindset.

So, getting down to it.... do you know how a diode distorts an AC signal?  I ask that because you are on the right track (do a lot of bench testing), but you could ultimately save some time, and possibly money spent on items for testing, if you knew what's really happening to the signal as it passes through the diode.

At this point, I don't want to influence you any further, so no hints.   I'd rather that you not 'read between the lines' of my question and then focus on whatever you think I'm after. Instead, I want you to craft your answer from your own knowledge base and your experiences so far. (However, you are allowed, even encouraged, to Bingle for help, should you so wish.)

And finally, do NOT think I'm putting a damper your enthusiasm.  In my mind, I'm attempting to guide you to an understanding of a crucial portion of guitar-related electronics, one that won't melt your brain.  Trust me on that last point.




sumgai

[frets]

Hi Sumgai😸

I hope I’m answering your question.

Whenever one of the diodes starts to conduct, the operating voltage gets clamped away from ground; then, the input reverses direction at which point the signal is taken up to its current level. Then the waveform distorts. That’s how I understand it.

[asmith]

Is there any way you can make a video/audio recording of your results? Maybe a friend has a good mic, if you don't? Because I would be so interested to hear the side-by-side comparisons.

Thank you very, very much for doing all this testing and relaying your findings to us.

[unreg]

Apr 17, 2023 14:36:24 GMT -5 frets said:

I most likely will using the 1N34A pair and the 1N4148 pair.  I might experiment with the configuration of the 8V2’s and may use those on one side of the switch, with 4148’s on the other side for a more aggressive overdrive.

Those in this group will most likely have good reasons for these perceived differences - forward voltages, dissipation, etc.

It was a worthy exercise as it clarified the pairs I can use on Overdrives to give the guitarist distinct differences.

It must be stated that this little study is wrought with bias and confounds.  But I think if anyone else in the group repeated the test, they would hear the differences in the 3 pairs from the more traditionally utilized diode pairs; I.e., 34A’s, 914’s, etc.

1N34A! Great choice frets! I believe I’ve shared this video with you, but:



Fast forward to the 15 minutes 48 second mark where part 2 starts… he explains diodes there excellently, to novice me at least; maybe sir sumgai will add more. Perhaps you’ve already seen this mam Finnegan… or frets.



edit: His oscilloscope use in part 3 is really cool! Maybe an oscilloscope would be a great tool to do your testing with?

[thetragichero]

i definitely encourage such experiments (as such experiments were encouraged of me... somewhere there is a thread where i opened up a metal zone and tried all sorts of clipping diode and op amp swaps, with direct-in audio clips included), even though to me they're super tedious and i generally choose based on Vf and how cool a diode looks

one thing that's important for this sort of post is to include a schematic of the circuit under test so we can determine whether you're testing "big muff"-style clipping arrangement or "electra"-style clipping

for what it's worth, i have sold dozens of modified proco rat pedals that use a three-way on-off-on spdt (although I'll use dpdt of that's all i have) clipping switch to choose between stock silicon (1n4148/1n914, which is generally the same part), no clipping diodes, and one red and one green led. the choices are fairly intentional (also born out of economy, since the leds were dirt cheap in bulk and reusing stock parts), but I'm not going to give away sumgai's riddle

[sumgai]

Per trag's depiction of my previous query as a 'riddle' (I like that!), I want to finish with the punch line.

Sadly, for various health reasons, I seem to be lacking in mental clarity this morning. I'll try again in a few hours.

But that's for the deep-down details. On the bright side of things, I fully agree with trag about experimenting. In fact, I'd be almighty surprised if I were to learn for a fact that somebody got through Electrical Engineering school without having ever so much as touched a soldering iron at least 3 or 4 years before entering said school. It takes a certain mindset, and yes, a deep curiosity about all manner of things electrical, to want to suffer through 4 years of higher learning just to satisfy an itch. So to speak....

I'd lay good money that every EE started out by either being fascinated by a demonstration of some sort of electrical/electronic gizmo, or else he/she got a shock from touching something that's not supposed to be touched while live. (Would it surprise anyone to learn that I'm in the latter category?) From there, it's tinker, tinker, tinker, until the light bulb over one's head starts staying on longer and longer.

Aaargh, the fog is setting in.....

[frets]

Hi Guys,

To ASmith, I will try and make some recordings. I can’t promise when though.
To Trag, it was a basic Electra style circuit. A little different but basically the same concept.

BTW - The diodes I used - the most of them I inherited from an Uncle. I have more but I think I’m done with this experiment.

I appreciate the interest.

[Yogi B]

Apr 17, 2023 14:36:24 GMT -5 frets said:

Asymmetrical

Given a symmetrical signal, an alike pair of diodes connected anti-parallel results in symmetrical clipping. (Within component tolerances, which as with everything else can be rather iffy with older/rarer components.) Asymmetrical clipping would typically involve a mismatched pair or unbalanced combinations of diodes (or a single Zener diode). However that doesn't mean it's impossible to get an asymmetrical signal from a symmetrical pair of diodes, as this can still happen in two scenarios: the signal is already asymmetrically clipped; or the signal is symmetrical about some voltage, but not the voltage to which the diodes are connected (i.e. the signal has some additional DC offset).

A common emitter amplifier inherently does not clip symmetrically, even if biased optimally: R_C & R_E set hard cap on collector current allowing for a much more abrupt transition to saturation than to cutoff. Another issue related to saturation is that with I_C unable to increase further it can no longer regulate I_B, and as such the combination of the input cap with the B-E junction effectively forms a charge pump. Each cycle pulling V_B towards V_E, temporarily altering the bias (this is the same principle as is behind 'blocking distortion' in valve amps).

A final thing to potentially watch out for is that the output will still have a DC offset whilst the output cap is charging, and on your image the electrolytic (which I'm assuming is the output cap based on its position) appears to be 100uF which seems excessive — though it should be fine as long as you keep the power connected and/or give sufficient charging time. While on the subject, I'll also note the seeming lack of an input cap.

---

Overall, 20 of the 23 sound the same to me in terms of clipping level, tone, decay, and volume level. 3 pairs stood out from the rest. These distinctively different paired diodes were: 1N4148, 1N5817, 8V2

I find this difficult to believe.

Those in this group will most likely have good reasons for these perceived differences - forward voltages, dissipation, etc.

Well, good reasons for finding the reported perceptions highly questionable. Without needing to go overboard on maths — which is complicated by the reasons above (plus not knowing the exact specifics of the circuit) — we still can make some rough estimations based upon the forward voltages (V_F) of the diodes.

V_F isn't constant and depends upon the current though the diode, and irritatingly the values quoted/graphed in a datasheet are often on the high side (measured at 100mA and above), whereas the absolute maximum we can expect in pedals is around 10mA. (An 18Vpp signal being driven through an impedance of at least 1kΩ before being clipped by an ideal diode about the mid-voltage equates to 9V / 1kΩ = 9mA.) Real world cases are subject to further reductions: inabilities to achieve a full rail-to-rail voltage swing (and the fact that those rails are likely only 9V apart, rather than 18V); the diodes being driven from a larger output impedance; and the existence of a non-zero voltage drop across any non-ideal diode.

Thankfully, however, the very action which makes diodes useful to us as clippers also alleviates the need to know the current to a great level of precision. Clipping occurs because (within limits) the current through the diode relates exponentially to the voltage across it, allowing only a small increase in voltage to vastly increase current flow. Or, from the inverse perspective, a large change in current can occur with only a relatively small change in voltage. Thus, although V_F does vary with current (and so cannot directly tell you the exact clipping threshold of any particular circuit), the variation is small enough that it can work as an effective proxy.

I'd categorise your diodes into roughly four categories:

• Schottky diodes — in particular 1N5817/18/19, very low V_F (around 150mV @ 1mA);
• Ge diodes & Other Schottkys — 1N34A/1N60p/etc. & BAT41/54, lowish V_F (250—400mV @ 1mA) additionally the relationship between I & V_F for Ge diodes is somewhat closer to linear than exponential meaning a softer transition into clipping;
• 'normal' Si diodes — 1N914/1N4148, the 1N400x series, etc. plus forwards biased Zener diodes (1N5231, 4V7 & 8V2 — also, FYI the latter two are their Zener voltage, not a specific part number), 'standard' V_F of around 600mV @ 1mA;
• LEDs — highest V_F (excepting reverse biased Zeners) starting around 1.2V up to around 4V, often tracking along with the frequency of their colour i.e. (infra) red being lowest V_F & (ultra) violet being highest V_F;
• Zener diodes — having been mentioned twice already, are technically a fifth type, having typical Si behaviour in one direction, but also having a comparatively well defined reverse breakdown voltage/behaviour in the other direction (ranging anywhere from about 800mV up into the hundreds of volts), though N.B. these reverse characteristics are irrelevant when two are placed anti-parallel (before reverse conduction can occur, the other diode begins to conduct forwards).

As such, I struggle to see how the diodes with the largest V_F supposedly sound the same as two of the three diodes with approximately ten times smaller V_F, yet somehow the third sounds markedly different.

[frets]

Those were my results as I experienced them. Go figure.

[gckelloch]

I would like to hear recordings of any future tests like this. If using a guitar, it's a good idea to record a DI clip so the same thing is used for each example.

[sumgai]

frets,



Cindi, I'm sorry but my (computer) life is not going too well just now. I need to finish a series of drawings to post for your edification, but AutoCAD is giving me fits. I'm probably going to have to either upgrade, or switch to something else, before all of this is done. Please, hang loose, I'm getting there.....




sumgai

[frets]

Sumgai😻

Don’t worry, you take care. I’ll be here.

[mr22]

I love this!

I have an idea. Is there some way to program a microcontroller or a Raspberry Pi to randomly select the pairs? If so, then it could be a really awesome setup for A/B testing different diode pairs to hear if there is an audible difference.

If a single knob on something like a JHS Rat pack can switch several sets of components at once, then I think something like this could be possible.

Unfortunately, I don't know much about microcontroller programming, Arduino, or Raspberry Pi.

Thanks for sharing.