Grübelzwang Overdrive
A Take On The Take On The Fulltone OCD (And Some New Findings)
last update: Nov. 28, 2024

Copyright 2024 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 or trademark 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>GRUEBELZWANG
Index


More Disorder?

The Best From East And West
The Clipper Section And Reference Voltages
A Look At The Distortion Generators
Treble Management
Buffering
18V Supply
Verdict
Reference
Update History

Back To Index

More Disorder?

Writings on OCDs abound, so indeed this writing would have been superfluous if id did not contain anything new.
It is probably aimed towards the experienced builder.

Recently I stopped by Guitartest to look at the Baldringer Drive. Over the years quite a respectable number of pedals have gone through the hands of this site' owner, so I trust his expertise to an extent.
He thinks that Fulltone's OCD sounds quite similar to the Baldringer. We will come back to that claim during the course of this writing.

Years ago I had acquired one of the OCD sound-a-like kits, and I DID NOT LIKE IT. Granted, those kits are a tinkerer's nightmare. They are a production unit in a way and do not lend themselves readily to modifications. I did what I could, but the outcome was not pleasing, so I gave it away.


Re-fueled, I wanted to look at this again, but this time make my own PCB.
I  adopted Fred Briggs' Super-OCD schematic and added some simple buffering, which we will hear more on later. This is the basis upon which we will build.

I had some quite surprising insights during building this unit, which may be of interest to the kind reader.
I called the unit the Grübelzwang Overdrive, which is one of many German words for the OCD disease, which the word game is clearly hinting at. I find this a very funny word by itself, leaving all connotations aside.

Back To Index

The Best From East And West

So I basically just used Fred Briggs' super-OCD.
One of the most useful additions to this circuit (which appears rather dark sounding to people in the original version) is the bass-cut.

I wired it as a conventional bass control, which increases bass clockwise. Although boost and cut is just a way of looking at things, having a cutting control, whose action increases clockwise while decreasing the affected parameter, is confusing, despite the fact that this was technically correct.

The very type of control has appeared on numerous other pedals  like Wampler's Euphoria and the Timmy. It is very useful to tame the low-end.

Briggs' version also employs a more effective tone control (treble control), which is relocated to the output. It is a variant of an actually very old type of treble control, which has been made popular by Jack Orman from  AMZ under the name of "stupidly wonderful tone control (SWTC)". On the FSB thread somebody used a different variation for the super-OCD, which cuts to one side and boosts to the other side, but I found this unnecessary and unbalanced.

Fred changed the second OPA to a variable boost, something I do not need. It also may have other side effects (more on this later).

The S-OCD uses another of Orman's ideas, a saturation control for the MOSFETs.

 
Note that the idea does not strictly follow what Orman calls saturation, although it does not really matter. Adding serial resistance to a semiconductor will change  the latter's transfer curve and make it eventually approach a resistor's transfer curve. If applied delicately, this will smooth out the otherwise rapid change of a diode's transfer curve towards what may resemble a germanium diode (which eventually may make the GE version of the OCD redundant...). Time will tell if this is the case. The other extreme is the case of a series resistor for driving a LED, in which case the LED's transfer curve is essentially replaced by the linear V/I curve of a resistor.
In my eyes a better describing word for what is going on here is compliance, but naming a control like that is confusing, for the reasons mentioned above.

In the following we will look into those sections with a microscope.

Back To Index

The Clipper Section And Reference Voltages

The circuit was working right from the start (erm, kind of...), while a few time constants needed time-consuming fine-tuning.

Unfortunately, from a certain gain setting on, the circuit started motor-boating, which resulted in the hated "put-put-put" sound resembling an  idling motor-boat engine. Its frequency increased with OPA#1 gain. The 'scope revealed huge fluctuations on the Vref rail.

Now I've had this before, but only with valve amplifiers, where the cause usually was some supply filter cap time constants.
I never had this on OPAs in conjunction with PCBs. And believe me, after so many years of PCB design I incorporated all known tricks of stabilizing circuits into my PCB.

After visiting the usual suspects I found that there is quite some current dumped into Vref by the MOSFETs when they are conducting, enough to corrode the Vref voltage node. Some OPAs go into latch-up when they are hitting the rails, which means they are "hanging there" for an indefinitely small period of time. This may be part of the signature "sound" a certain OPA imparts to a circuit when clipping (more on this in a minute).

Briggs also references the input's feedback return paths to Vref (like the lower leg of the gain setting R/Cs), which technically works, but also has the potential to corrode Vref. Several reputable designers strongly advise against that practice.

 
Brian Wampler in his design of the Euphoria uses individual resistive dividers for every node that needs a reference voltage, despite the fact that all dividers are nominally equal. I have seen this for the first time, and while we cannot know why he chose that approach, this circuit certainly does not have any repercussions.

Since the original OCD works with a configuration like that, I attribute the malfunction to an coincidence of unhappy trace layout decisions on my PCB. Stuff like this is a bitch to track down.

The cure for this was to reference the clipper section to GND instead of Vref via a large enough electrolytic capacitor (22µF).
Alongside, all feedback return nodes are referenced to GND.

Back To Index

A Look At The Distortion Generators

In the course of observing the waveforms with a 'scope I found that OPA#2 is contributing quite a lot of hard distortion by constricting the signal swing with its internal high and low boundaries.

In my version I have added the Germanium diodes some units have in series with the MOSFETS. I have made provisions to bypass those with a switch selectively for evaluating the sonic impact of the resulting combinations.


Using a two-pole on-off-on switch allows to switch to MOSFETs only (akin to the "SI" version), MOSFET + GE diodes (akin to the later "GE" re-issue) and a version with one GE diode ("Asymmetric", akin to some intermediate version of the OCD.

The MOSFET body diode consumes 0.7 volts, and together with the circa 0.2 volts of a Germanium diode we end up with a clipping boundary of roughly +/- 1 volt, or 2V peak-to-peak. Using compliance will increase those values even more.


Now the subsequent OPA stage has a gain of four in the standard version, which will invariable clip the input signal hard when slammed with two volts on a nine volt supply. This happens very early on the gain pot travel and is responsible for a certain gnarly tone (you will notice its absence when using a higher supply voltage; see later). Note that the driving OPA stage clips the latest (being first in the row), but will eventually clip too.

OPA clipping has the reputation of sounding bad, but several well respected units like the MXR distortion +  or the Crowther Hot Cake and many more use just that effect. The Hot Cake relies solely on op-amp clipping.


Some people think that the OCD sounds similar to the Hot Cake. This is not surprising given the op-amp distortion mechanism. That said, the latter has very ineffective tone controls and sounds harsh quickly.

Fred Briggs talks about using a "better" OPA for the OCD, but consider that the TL082 was chosen with purpose, much as the µA741 sounds best for the Hot Cake. It may be worth trying a 1458 for that reason.

I am ready to believe that Fulltone has wisely chosen the balance between the rather soft signal bounding by the clipper section (despite the fact that this topology is commonly called "hard clipping") and the merciless constriction at the boundaries of the OPA close to the rails by setting the latter's gain a certain way. Besides seeing no need of even more drive voltage (who needs more than a 9Vp-p drive signal?) the delicate balance that mark the transition from one distortion engine to the next may be impaired by fiddling with OPA#2 gain. Increasing OPA#2's gain will make this stage clip earlier.

 
With an architecture that stacks OPA clipping with diode clipping (of sorts) the OCD's architecture is similar to an MXR distortion + or its sibling, the DOD 250, but due to the subsequent OPA stage its dominant distortion generator is not the anti-parallel diodes, but rather the OPA itself. Again a similarity to the Hot Cake. At lower settings of the driving OPA's gain the signal abides in the region of transition.

Using a different type of OPA, as suggested by Briggs, may or may not be an improvement, particularly since the TL082 is not apparently obtrusive. This circuit appears purposefully designed around a low grade OPA. I would definitely stay away from "modern" high-bred OPAs in this respect. Again, a 1458 or something like that may work well. Incidentally, this is said to be a double µA741, the latter which is the heart of a Hot Cake.

Note also, that different OPAs do vary hugely in
  1. the way they clip at the rails
  2. the way they return from latch-up
  3. the way they behave in a given electric environment (such as supply decoupling and HF behavior)
  4. how fast they are

Let's look at the way the OPAs clip. Clipping a signal symmetrically or asymmetrically makes a huge tonal difference. A short aside:

 
Asymmetrically clipped signals predominantly add even-order harmonics  to the spectrum, whereas symmetrically clipped signals add odd-order harmonics. This is called harmonic distortion. The latter sound dissonant, which can make tone interesting in small amounts. However, this only affects a single frequency.
As soon as two (let alone more) frequencies are involved, frequencies do interact and start producing combinations that are totally unrelated to the original frequencies  and thus sound disharmonic. This effect, which is called intermodulation distortion (hereafter: IMD), unfortunately soon takes over and becomes the dominating effect.

Since those generated combinations can be higher and lower in frequency than their source they also can add artificial bass. Circuits that produce only odd-order distortion (such as symmetrically clipping circuits) do not produce artificial bass.

Intermodulation distortion (hereafter: IMD) is unavoidable in distortion systems and rises with the number of tones played simultaneously. Therefore, the more distortion we want, the more we have to rely on circuits that distort symmetrically, because these produce less overall IMD. (abbreviated and summarized from[1])and[2].


"Asymmetrical clipping (...) is ideal for low- to moderate-gain sounds where touch sensitivity is important.
Symmetrical clipping is ideal for high-gain sound where consistency and note articulation are important."

   - Merlin Blencowe, Designing Tube Preamps for Guitar and Bass (second ed.), p.254          

Oscilloscope trace of
                                asymmetrically clipped signal
Asymmetric Clipping by TL082 ( (click on the picture to load larger image)

The TL082 clips markedly asymmetric towards its supply rails; using a centered reference voltage, it clips the bottom part much earlier, which is typical for a standard OPA.

Note that the display has been rescaled so that the 9V supply is coincident with the top division mark. The displayed number for the y-deflection (1v/div) thus is slightly wrong (it is more than 1V). Coupling = DC, so we see the signal as it rides on top of the reference voltage.

The signal has been amplified enough that the upper end is just barely on the verge of clipping.

The bench supply is set so that the voltage at the OPA supply pins measures exactly 9V. The lower clipping boundary resides at 1.5V and the upper boundary at 8.1V. Those are amplifier specific values and will invariably change with a different OPA.
 

Some older OCD versions try to provoke this asymmetry by adding a single germanium diode to one of the MOSFETs. However, its contribution gets swamped by the vastly dominant asymmetrical distortion of the subsequent OPA, even if its purpose were to try to mitigate  the strong asymmetry.

Indeed, toggling the "hard clipper" section (MOSFETs etc.) between symmetric and asymmetric  (as the Grübelzwang allows for) does not have an effect as big as you may expect, which is why it can be assumed that most incarnations of the OCD sound very similar, as is frequently expressed.

Yet with a very small change we can make the OPAs clip symmetric at will: we raise the reference voltage slightly.

oscilloscope trace of symmetrically
                                clipped signal
Symmetric Clipping by TL082 ( (click on the picture to load larger image)

Elevating the reference voltage slightly, the TL082 clips now clips symmetrical. This does not limit the headroom  noteworthy, certainly much less than increasing the diode's compliance (resp. lowering the  "saturation" control) does.

The resulting tonal difference however is remarkable and comes pretty surprising.

The original Vref measures 4.5V, the amended one 4.8V.

I followed the path Crowther has taken by elevating the bias voltage of the OPAs a few 100 millivolts or so. The exact value has to be determined trimming a pot while watching the scope for symmetric clipping.

I made this switchable, so that I can quickly jump from asymmetric to symmetric. In doing so I found that there is some component that withdraws into the back in symmetric mode. It sounds less distorted overall, although there is the same amount of distortion present, just because of a different spread of harmonics in the spectrum. With this active, turning the MOSFET assembly switch to "asymmetric" really makes a noticeable different. There is the interesting aggressive component but transiting into something smooth. Very pleasant. I was astounded, what huge influence that has, and how soft an OPA in distress can sound.

Note that R.O.G. have deliberately added (symmetric) hard clipping sections to their Thunderbird Marshall emulator. Obviously those are needed and indeed, some places in a tube amp are hard clipping. Note also that an output stage of a tube amp is built symmetric, and as asymmetric the preamp stages may clip, by the time the signal hits the power stage, it is mostly symmetric again (at least for push-pull amps).

So pumping a stock OCD (with asymmetric clipping) into a smoldering amp may sound perfect, while a typical bedroom setup with a clean amp and higher settings of input gain may profit from a symmetric OPA clipping. Activating one of the GE diodes at the MOSFET can bring back the picking articulation and thrill.


There is more tricks that can be played to achieve local asymmetry besides adding a GE diode. They dwell in Orman's abode.

Back To Index

Treble Management

The usual resistor/capacitor series shunt arrangement is not very useful. The SWTC (first drawing) is much more effective. Indeed it incorporates all of the tones the funny OCD tone switch settings provide, but with all in between nuances too.

However, the third circuit shown in 
SWTC , as suggested on the FSB thread, did not work well right off the bat. It was too shrill and had limited adjustment range towards reducing treble.
This could have likely been remedied with some more tinkering, but adding more treble to this circuit seemed superfluous in the vicinity of a bass cut control. Remember, less bass lets a sound appear more trebly due to the yin-yang of tone. Moreover, cutting bass in a distortion mechanism usually yields better results than boosting treble.

From my past Björn Juhl experiences I remembered the extended feedback modes he uses in the Honeybee and its siblings. Unlike gazillions of designers before him, he feeds back higher frequencies from the point of their occurrence, such as the clipper stage. This is not the same as dealing with higher harmonics at a spot after or before. It feels like a dynamic thing, like a dynamic equalizer.

I left the first OPA's HF limiting capacitor with 100pF (original: 220pF), but fed back from right at the junction of the 10k resistor and the clipper(s) with another 100pF cap. In this case I prefer ceramics. For HF limiting purposes they are unsurpassed.

I considered a second feedback node from right after the SWTC, but this seemed unnecessary.

The 1nF cap at the clipper stage is not doing much. Together with the 10k resistor this yields a cross-over frequency of 15kHz. So if we wanted to hear any change we would have to add something like 3.9nF for a total of 5.
Changing the feeding resistor, as Briggs suggests, not only changes the R/C time constant (and not much anyway), but also changes the way the MOSFET section behaves. I recommend to leave this alone.

The tone pot works well with the starting values Orman recommends. At its center position, the unit sounds like the unaltered bypass signal upon soft strums, and this is how I set it. Higher gain settings require some backing off.


Back To Index
Buffering

Input impedance may be acceptable as is, but output impedance is certainly on the high side. Many effects are made this way, and you usually get away with murder. Most time this decision is buck-driven. Briggs conceded in a forum discussion that of course buffering would be the proper way to go about.

I took the buffering scheme from 
AION's Xotic AC Booster. This is noteworthy because of the way they derived the bias voltages for the transistors. On the output, they just used Vref, which is exactly at half supply. The buffer's output is a diode drop lower, which would theoretically limit the output swing. However, the signals are likely small at this place.

On the input they use an elevated bias voltage, which is produced by adding a single resistor into the divider chain. The resulting voltage Vref2 is about a diode drop higher than Vref, so the input can swing symmetrically for a hypothetical +/- 4 Volts. Very likely unnecessary, but cheap and nice.

In my opinion, they should have added some filtering too, like it is applied on the first transistor on the Roger Mayer Axis Fuzz. I used exactly the same values, albeit derived from Vref2 and not VCC
. I came up with this during my endeavors to track down the motorboating issue. This pretty much isolates Vref2 from any variation on Vref.

Some of the myriad of OCD clones have a (transistor) output buffer incorporated, and somebody was on a mission to "replace the horrible sounding transistor buffer". I do not share this opinion, in fact I feel it is ridiculous to complain about the influence of a transistor that buffers a circuit that produces nothing but distortion. It is also ridiculous considering how much the subject of proper impedance matching is neglected in general.

Adding symmetry to the "output stage" (OPA#2) is most easily accomplished by amending above mentioned reference voltage divider. Adding a 10k trim pot temporarily across their 6.8kOhm resistor helps to determine the value required to achieve evenly clipped lobes, while monitoring the output of OPA#2 with a scope. This will work on any supply voltage; the divider is ratiometric.

I you prefer to have this option switchable,  so a simple spdt toggle switch can toggle between  symmetric and asymmetric.


Back To Index

18V Supply

Because of what has been found further up on the way the OCD's "distortion generators" are stacked and interact, the charge-pumped 18V supply changes the distorted tone a lot (note that it is not really 18V - the circuit is lossy. The number is usually chosen for marketing purposes and for convenience).

OPA#2 will be driven into clipping later. It modifies the balance between the rather soft clipping MOSFET stage and the brickwall hard limiting OPA stage. De facto, it takes OPA#2's clipping mostly out of the equation, which will refer the clipping action mostly to the MOSFET stage and also results in a considerable increase in loudness.

Increasing the headroom of OPA#2 thus is the complementary action to increasing the gain of OPA#2, and it reduces the grit added by the op-amp clipping. This can be perceived good or bad.

Not saying that his might not sound useful, but it may not be what constitutes a typical OCD overdrive sound, and it might not sound very tubey, if such a thing exists at all. This mode feels like extending the lower drive settings on normal mode.

Switching the voltage while having the effect activated produces a magnificent "thump", so this should be avoided during a performance. The thump itself is inevitable without resorting to complicated measures, because it switches DC.

If Fred Briggs
likes the "extra headroom to the circuit to stop the op-amp clipping from fogging up the tone", he eliminates what constitutes the OCD.

Back To Index

Verdict

Indeed this unit sounds and reacts different that the numerous other dirt boxes I have. It is largely devoid of the fizz others generate and it is addictive in a way. On a 1:1 comparison the Grübelzwang sounds noticeably unprocessed and natural.

Time will tell which of the added options deserves to consume front panel estate or remain condemned to dwell in the hidden.

I cannot say if the Grübelzwang truly bears resemblance to the Baldringer drive, but it definitely does not change the character of my amp, and it truly sounds like what the Six-Pack-Joe would call tubey.

 
What has not been addressed in the comparison of the OCD with the Baldringer (Dual) drive is the noise performance.

While the OCD uses quite a substantial amount of signal boost to trigger the clippers, the Baldringer does not boost the signal much, but instead boosts the control signal into the clipping elements (BJTs) to achieve the same effect. This yields a much improved S/N ratio, although the OCD is not apparently noisy by itself. For those interested, Dirk Baldringer holds a Patent WO2007138105A1 that explains this distortion mechanism. Note that it contains some mistakes as I found (Honi soit, qui mal y pense).

This design really exposes that gain ≠ distortion. See in this context: The myth of "gain"... and it's relation to "clipping" by Brain Wampler.

The Baldringer is a much evolved design and allows for a lot more of tone shaping, so this is not a fair comparison, letting alone its price tag. The Grübelzwang on the other hand, as different to the stock (clone) I had, leaves nothing much to be desired, so I am content with what I have now.

For a while.

May this writing help others to select wisely.

Back To Index

Reference

[1] Merlin Blencove (a.k.a. Valve Wizard) , Designing Tube Preamps For Guitar And Bass (2nd ed.),
    Chapters 1.11 on Harmonic Distortion and 1.12 Intermodulation Distortion, pp. 15-18
[2] Merlin Blencove (a.k.a. Valve Wizard) , Designing Tube Preamps For Guitar And Bass (2nd ed.),
   
Chapter 14 on Preamp Topology, p. 254

Back To Index

Update History
  • Nov 28, 2024: first release
Back To Index
MAIN PAGE>MUSIC STUFF>GRUEBELZWANG
MAIN PAGE | MUSIC STUFF | IMPRESSUM

(c) 2024 AQUATAUR Musik & Elektronik