Op-amp And Diode Clipping Stages Revisited
A look at op-amps, diodes and distortion mechanisms in the lights of a few commercial pedals
last update: May 14, 2011

Copyright 2010-20 by H. Gragger. All Rights Reserved. All information provided herein is destined for educational and D.I.Y. purposes only. Commercial re-sale, distribution or usage of artwork without explicit written permission of the author is strictly prohibited.The original units  with their associated  trade-names are subject to the copyright of the individual copyright owner. The Author is by no means affiliated with any of those companies. References to trade names are made for educational purposes only.By reading the information provided here you agree to the Terms of Use.
MAIN PAGE>MUSIC STUFF>DIODE_CLIPPERS

Index


The Birth Of Component Mojo

The KoTīs Spice Ingredients
Diodes In Clipping Stages
Diode Bounding - A Perfect Approach To Tube Tone
More Tube Tone
Slew Rate Limiting And Psycho-Acoustics
More Possible Sources Of OPA Distortion
Reference



 
The Birth Of Component Mojo

I do not want to add to the list of the guys talking about the virtues of a certain semiconductor "sound" and where to get that maybe long obsoleted device.

Since everybody rants about those mojo components, they are readily available these days and I would be willing to buy them just for convenience, if there were not this uncomfortable feeling remaining, that I do not really know why they are preferable over some other type.

Sometimes the choice of a certain component in a device happened accidentally, for no particular conscious reason. It was later that it turned out that the choice was crucial. The often quoted TS series op-amps for example were initially chosen for profane reasons only: availability and price. Later it is hard, if not impossible, to find out what made the device so special.
But we can look at a few milestones at least.

In the verge of building a King of Tone clone, I looked at the diodes used there and what allegedly makes them so special.

Back To Index

The KoTīs Spice Ingredients


Diode Type
Diode Capacitance
Switching Speed
Remark
1s1588 3 pF
4 ns
high speed switching diode
1N4148/1N914 4 pF
4 ns
high speed switching diode
equivalent to 1N914[1]
1N400x
15 pF
30 ĩs[2] std. rectifier diode

Diode Type
Diode Capacitance
Dynamic
Resistance
Remark
MA856 (=MA2C856) 2 pF
0.85 Ohm
band switching diode
BA482 (483, 484) 1-1.6 pF
0.7-1.2 Ohm
band switching diode
BA278
1.2 pF
0.7 Ohm
band switching diode[3]

It becomes obvious that they are not very special at all and it may thus appear a waste of time and money to pursue those exotic types. As the story evolved for the opaīs used in the tube screamers it very likely did for those diodes - they just were around in millions at the time (or place) the circuit was developed. (Note: I do not say the specific type of diode was chosen accidentally).

Back To Index

Diodes In Clipping Stages

Diode clipping circuits can be roughly divided into two kinds of circuit:
  • the op-amp feedback clipping type and
  • the series shunt type
The original TS falls into the first category. Read about all its secrets in R.G.Keens article. A pair of anti-parallel diodes in the feedback loop limits the amount of voltage the output of an operational amplifier can produce. This interacts with the opaīs gain. Allegedly it is even important if the diodes are in the feedback path of a non inverting amplifier rather than an inverting as some article claims.

A series shunt type clipping circuit follows a driving stage (like an opa) and is fed by a current limiting resistor. This does not interact with the opaīs gain. The Fulltone OCD  incorporates exactly that topology, and, indeed, the distortion section of the KoT does too.


At the time of writing it is not known to the author, if there is any tonal difference between those two methods provided they are set up comparably and the headroom is sufficient. More investigation is in order.

Both types of clippers on their own will limit the voltage across them to their threshold voltage. For a silicon diode, this is 0.6 to 0.7 Volts. In this region the magic happens and thatīs where all the mojo comes in, because diodes will differ in the way they start to cross into conduction.

Somebody suggested the usage of the intrinsic (parasitic) diode of a MOSFET. While this basically works, they are comparably slow, although their on-resistance is low. Although it might sound good to somebodyīs ears, it will certainly not serve well to replace the exotic diodes asked for in this circuit. Similar things apply for standard rectifier diodes like the 1N400x series and even more so for LEDīs. I vastly appreciate Madbeanīs work, which was the basis for my tone queen project, but technically I disagree on his suggestions for diode subbing for the above mentioned reasons.

It turned out that MOSFETS do sound good for the series shunt clipping part despite their obvious technical difference.

Shottky diodes are fast too (even faster) and may work, but more of them would have to be stacked for experimentation, since they have a voltage threshold that is about 1/3 of a standard rectifier diode, otherwise the voltage balance will be impaired which may in turn impair the dynamic behaviour that makes this stompbox so unique.

I am not claiming, that using those diodes is essential to the sound, but assumed it is, any of the listed replacement diodes will perform exactly the same and are readily available. And I am sure, there are lots more with the same specs. Anything beyond that... is mojo...

[In retrospect, the 1N4148 do not sound overwhelming in a stock series shunt configuration as the KoT has it. The MOSFETS sound much better. See my article dedicated to the tone queen...]

Back To Index

Diode Bounding - A Perfect Approach To Tube Tone

After all, with tube screamers and all the clones in its wake, we are after tube tone.
A tube clips much more graceful and gradual than a solid state rectifier, even germanium. Hard limiting also means heavy compression. Any amount of input signal is limited to a certain value. This is not tube like.

Letīs look at the way a tube diode conducts.



chapter 5: Preamp Modifications - Distortion Generating Techniques - Signal Bounding

"Signal Bounding is simply the restricting of the signal excursion. If you tie a string around the middle of a balloon, then inflate it, the string will limit the balloon diameter at one point. We can do the same thing with electronic signals."

- Kevin OīConnor, 1995, The Ultimate Tone, Power Press Publishing, London, Ontario, p. 5-12.



chapter 5: Preamp Modifications - Diode Bounding - Semiconductor Diodes

"These voltages [where solid state diodes conduct] vary only slightly with the forward current, as the internal resistance of the diode is very low."

- Kevin OīConnor, 1995, The Ultimate Tone, Power Press Publishing, London, Ontario, p. 5-13.

If we apply this method, we can therefore expect some hefty compression.



chapter 5: Preamp Modifications - Diode Bounding - Vacuum Tube Diodes

"We can also use vacuum tube diodes to bo8und the signals in our tube preamp. [...]
However, the internal resistance of the tube is much higher than that for a semiconductor, so the forward voltage across the tube rectifier will vary with the signal. The signal-dependent diode voltage makes the tube diode clipper output less square than the solid state output. The tube method will sound softer, or less harsh or crisp, compared to the metal rectifiers."

- Kevin OīConnor, 1995, The Ultimate Tone, Power Press Publishing, London, Ontario, p. 5-15.



chapter 5: Preamp Modifications - Using Solid State Diodes to Mimic Vacuum Diodes

"It would be ideal if we could configure these inexpensive components to mimic the soft bounding that the vacuum devices exhibit. And fortunately, we can.
The primary difference between the performance of the two technologies is the internal resistance to the signal current. The obvious solution is to add series resistance to the solid state diode circuit [...]
By adding a simple resistor, we have taken the edge off the squared output. There is now a rounding of the wave, with the variation from true-square dependent on R+Is . The larger R is made, the more peaked the output will be. At some high value of resistance, no bounding will occur [...].

At London Power, we call this signal-dependent portion of the output wave "compliance"

- Kevin OīConnor, 1995, The Ultimate Tone, Power Press Publishing, London, Ontario, p. 5-16.


To give an example how much signal variation this method introduces, look at the following table that has been experimentally determined in a test conducted by Kevin OīConnor:

Zener Voltage (V)
Series Resistance (Ohm)
Output Voltage (Vo)
open open 80
5 0 5
5
3k6
11
5
100k
58
12
0
12
12
3k6
18
12
33k
42
12
100k
62
36
0
36
36
5k1
43
36
100k
66

Source: Kevin OīConnor, 1995, The Ultimate Tone, Power Press Publishing, London, Ontario, p. 5-16

It can easily be seen that with rising series resistance the characteristic of the diode gets increasingly swamped, even although the threshold voltages are so vastly different.

And just for completeness:



chapter 5: Preamp Modifications - Capacitive Compliance

"Another method used to soften the harshness or raw diode signal bounding is to place small capacitances in parallel with the diodes. The capacitor creates an RC roll-off with Rin or the compliance resistance. This roll-off removes some of the higher harmonics in the square wave, resulting in a rounding of the corners of the wave. This method is used in second generation effects  pedals such as the Ibanez TS-9 Tube Screamer.
Capacitive compliance has its limits since it is most effective at high frequencies."

- Kevin OīConnor, 1995, The Ultimate Tone, Power Press Publishing, London, Ontario, p. 5-17f.

[To understand the above quote, it must be known that all bounding circuits are to be preceded by a series current limiting resistor...]

It should be mentioned, that some diodes have inherently high capacitances, such as the 1N400x[2] (15pF). It is interesting in this aspect, that the KoT uses deliberately low-capacitance diodes.

So if we recapitulate, the TS as a protagonist for endless copies uses a hard limiting diode pair with capacitive compliance and does some palliative measures to remove the harshness and achieve a wanna-be tube tone. The KoT in the series feedback diode pair at least, uses resistive compliance to achieve some much more tube-like tone with little compression. This leads further to the conclusion that the type of diode used is less critical, since its behaviour in the turn-on region is vastly swamped by the series resistance.

The KoTīs distortion section (shunt clipper) however, depends entirely on the characteristics of the diode selected. It has no compliance whatsoever and can be expected to clip hard and with heavy compression. Consequently, this has not been generally accepted by musicians as well as the other modes. Also, a volume step goes with it.

Adding a little compliance by means of a series resistor would not only
  • remove the critical influence of the diode type but also
  • make for less compression
  • more tube-like tone and
  • reduce the gain step
Back To Index

More Tube Tone

There also exists some indication that a tube amp limits the signal asymmetrically. This has been discussed in a DIYstompboxes thread. Fulltone lives up to this by adding a ge diode in series with one of the MOSFET diodes in both their OCD and Fulldrive (MOSFET ed.) pedals. This gives of course different bounding for the two devices. One could alternatively use two different bounding resistors, but this would shift the whole "feel" and response away from bold distortion. Too big a compliance would eventually result in no bounding at all, at least not with the headroom available in typical battery-driven stomp boxes like that.

Back To Index

Slew Rate Limiting And Psycho-Acoustics

Now a few thoughts on op-amp choice.

Hi-Fi Guru John Linsley Hood, now defunct, spent a lifetime to study the perception of sound by the combination of the human ear and brain. From him we know that the ear can tolerate several percent of harmonic distortion without noticing, particularly second order distortion. However, the ear/brain seems to be very intolerant against distortion of the shape of a pulse. Now, the Hi-Fi guys did not learn from him too much, they still test their amplifiers running at full throttle on sinusoidal waves. This is not a criterion. Amplifiers that are tamed with heavy negative feedback usually exhibit low THD, but are made slooooow. This means that a fast transiting signal will be noticed as distorted, which is allegedly far more obnoxious than spectral distortion.

One of the most overlooked reasons why valve amplifiers sound good for musical instruments is that they are very fast due to their voltage driven nature and their graceful response to overload.

In the fight for low THD figures there are HF-compensation techniques applied in solid state amplification, that yield unnecessary low THD figures, but slow the amp down. This effect is called slew rate limiting, the amplifier cannot follow (slew) a fast signal on the input and it chokes for a fraction of a second. The nature of such slopes is transient, but particularly musical instrument signals are of pulse nature and a mutilation thereof is obviously audible and obnoxious.

JHL postulates (amongst others) that this leads to a phenomenon called "listening fatigue", in other words, after a short time of exposure to such a system you find it unpleasant and want to turn it off.

The goal must therefore be to prevent signals coming into the amplifier that are too fast for it to follow. This is of course a function of amplitude, gain and frequency.

One of the techniques to prevent such signals ever entering an amplifier is to band-limit the signals coming into its input. Fast transits or even RF frequency will noticeably distort the wave shape. A simple RC filter on the input that limits the bandwidth to the audio range will cure all that. There is of course no such a thing on the KoT, so one will be added.

For a discrete power amp there are of course additional compensation points needed that can be optimized for speed, but since we have no influence on that (unless we use an opa that is externally compensated...) this is not applicable for our situation.

A private discussion with R.G. Keen over e-mail brought up the point, that the diodes inside an opaīs feedback loop present such an unexpected jump in gain that it probably chokes for a fraction of a second. This is probably worse, the faster the diodes are...

Not much investigation has gone into this subject, but I postulate, that what holds true for a discrete power amp also holds to an extent true for an op-amp in this special application. Op-amps are usually very fast, because there is little power involved, but due to their limited gain-bandwidth-product they will invariably slew at some point. Although this measure may appear overdone at a first look, it is not exaggerated to suggest a fast opa for this application due to the system-immanent fast transits.

Also not yet addressed has been the subject, that some opaīs tend to latch up and "hang" for an unpredictable time, when they slew. A 5532 for example, has diodes across its inputs to limit the voltage between them to counteract exactly that case. Now normally, both inputs are on the same virtual level in a standard gain application, but nobody can say what happens when things become non-linear for an infinitely small time and how the ear likes that.

This may explain why some opaīs perform better in those circuits than other. Hard to say, how the original opaīs perform in this respect (I would not even know how to measure such phenomena, I suspect it is not so easy...) A LF353 would, by matters of speed, certainly outperform any of the above. Interestingly, it sounds very trebly in a KoT circuit and it was discarded.

The NE5532 I would avoid in feedback-diode clippers. You may not listen to the clipping after the opa, you may hear the internal diodeīs clip. Apart from that, the 5523 is a current eater par excellence.

There is not much we can do to limit the gain step inside our circuit, because this is part of the tone we want. Interestingly, the King Of Tone deliberately asks for high-speed diodes on both clipping sections, that are even faster than the common variety.

Back To Index
More Possible Sources Of OPA Distortion

The question arose, what load a garden variety opa is able to drive. Although not too many datasheets of commonly used specimens display this figure,  a figure that appeared frequently was 25-40 mA, so a load greater than 1k @ 9V (as the KoT has it...) should be well within the mark.


Back To Index

It is not proven that any of the above applies to stompbox circuits, but it may well be, so it is
worth contemplating. It may make future decisions on component choice easier.


Reference

[1] Fairchild Semiconductor
Datasheet 1N4148/1N914

[2] General Semiconductor Datasheet 1N400x
[3] NXP (Philips) Datasheet BA278

Back To Index


Update History

  • Dec 22, 2010: first release
Back To Index

MAIN PAGE>MUSIC STUFF>DIODE_CLIPPERS
MAIN PAGE | MUSIC STUFF | IMPRESSUM

(c) 2010-2020 AQUATAUR Musik & Elektronik