This Week on the Breadboard: The Tube Bender Deluxe

[Mictester]

One question i have is would we get even more tonal options if we use Germanium for Q3 and Q4?

What is this obsession people have with Germanium components? They're noisy, thermally unstable, getting expensive, low gain, and really really NOISY! For real!

I've built circuits with both germanium and silicon devices, and with the tiny modifications I sometimes make to the silicon variants, NOBODY who've tried them in A / B tests can hear the difference.

Just for amusement, a couple of years ago, I built a couple of Fuzz Faces in identical diecast boxes, painted the same colour, with the same knobs. One was built with germanium transistors and was temperature compensated (the compensation circuit was ~12 times as complicated as the Fuzz Face circuit!). The other was built with silicon transistors, and had small value capacitors from base to collector of each transistor to tame the "fizziness". Each had a "serial number" on the base, so that I could identify them without opening them up.

I loaned them both to a well-known London Rehearsal / Recording Studio. NOBODY who tried each of them could tell the difference - and we're talking about professional musicians who are always picky about the way they sound!

Properly configured silicon effects can sound just as good as germanium one, without the thermal instability and noise!

 

[Chuck D. Bones]

Ge trannys have a well-deserved reputation for all of the things you mentioned. But, not every Ge transistor is noisy, and not all Si transistors are quiet. I have built pedals with Ge transistors that are dead quiet. Like JFETs, Ge transistors must be hand-selected for the desired properties.

Thermal stability can be an issue, depending on the circuit. I have used a few simple and very effective methods to stabilize the bias of a Ge transistor.

Adding a feedback cap to reduce the bandwidth of a transistor circuit is useful in both Ge & Si circuits. Some Ge transistors, such as the 2N130x series, have a lot of bandwidth and they benefit from a feedback cap the same as an Si transistor.

I have compared Ge & Si transistors on a curve tracer and I can tell you that the transfer function of many Ge transistors has "features" that Si transistors lack. The sonic difference will be apparent in some circuits and not others.

I don't doubt that the musicians who tried your Si vs. Ge FF comparison could not tell the difference because it's a highly subjective test. Did they hook them up to an A/B box and switch back & forth, playing the same passages on the same guitar? What do you think would happen if you provided two electrically identical pedals, except one was painted red and one was painted green and you asked the same musicians to pick which one sounded the best?

I do agree that most of the hype surrounding Ge transistors is just that: Hype. But IMO there are some places where Si just won't do.

So what's your opinion on carbon comp resistors?

 

[steviejr92]

What is this obsession people have with Germanium components? They're noisy, thermally unstable, getting expensive, low gain, and really really NOISY! For real!

I've built circuits with both germanium and silicon devices, and with the tiny modifications I sometimes make to the silicon variants, NOBODY who've tried them in A / B tests can hear the difference.

Just for amusement, a couple of years ago, I built a couple of Fuzz Faces in identical diecast boxes, painted the same colour, with the same knobs. One was built with germanium transistors and was temperature compensated (the compensation circuit was ~12 times as complicated as the Fuzz Face circuit!). The other was built with silicon transistors, and had small value capacitors from base to collector of each transistor to tame the "fizziness". Each had a "serial number" on the base, so that I could identify them without opening them up.

I loaned them both to a well-known London Rehearsal / Recording Studio. NOBODY who tried each of them could tell the difference - and we're talking about professional musicians who are always picky about the way they sound!

Properly configured silicon effects can sound just as good as germanium one, without the thermal instability and noise!

In my humble opinion, I can hear a difference. Simply put a Ge transistor has characteristics that a Si transistor doesn’t have. Sonically I believe Ge have a much smoother tone than Si.

I agree with you that ge transistors are noisy but that’s in generalities. I’ve had plenty ge transistors that were quiet as any Si I’ve tried. I’ve breadboard plenty of circuits where the ge component behaved as it should have with no compromise to the harmonic output and had no background noise.

I only use that cap on the 2nd transistor and have always been successful in taming the circuit (for both si and ge). While those capacitors do help I believe they dull the tone in an unsatisfactory way. I only use them if it’s necessary.

 

[MattG]

Is there anything special about the MPSA13 transistors in this circuit, other than they are Darlington transistors? More specifically, could each be replaced by a pair of commodity BJT NPNs (e.g. 2n3904 or bc549) configured as a Darlington pair?

 

[Chuck D. Bones]

We need the high HFE. You can make a Darlington pair from two transistors. The only caveat is that the two Darlingtons (MPSA13 or a pair of transistors) should be closely matched for Vbe. We want the transistors to operate at nearly the same collector current. Why not avoid the hassle and buy MPSA13s from Mouser? I bought a bunch from AliExpress, but I have not checked them for matching.

 

[aquataur]

Chuck,
May I ask why you chose to use darlingtons for the differential pair? It ist not quite obvious that you did, because you drew them as if they were standard transistors.
In the original circuit they use standard transistors with a hfe of no more than ca. 500 - the darlingtons will be a decade higher.
Thanks.

 

[Chuck D. Bones]

We need the high HFE because we need a high input impedance and low DC base current. We need a high input impedance so we don't load down the BASS filter. We need a low base current so that turning the BASS knob doesn't skew the DC balance of the Q3-Q4 differential pair.

You're right, I drew them as standard transistors. I have a Darlington symbol, but it's not pretty. So I do what most other folks do and draw Darlingtons as standard transistors. It's not enough to look at the schematic symbols, you also need to look at the transistor datasheet. With few exceptions, I'm very deliberate about the parts I choose. BTW, I've measured MPSA13 HFE in the range of 25K to 33K.

 

[aquataur]

Ah that makes sense. I was aware that the parts were chosen with intent. I did not think too much about the bass filter. I had initially thought that maybe there was a sonic reason, comparable to what a higher hfe transistor does in a fuzz face, something that came out of practical experience.

The hfe values I took were the minimum values they quoted in the datasheet according to the natural spread. The ones I can get (easily) are from CDIL, the Indian make-all semiconductor company, and you are at the mercy of them what hfe number you get. However, it will be high.

Thank you for your expertise.

 

[aquataur]

I built a unit. I appreciate the engineering work you have done, some of which was new to me.
I built it because I found the distortion mechanism intrigueing and alluring.

Such as the compound j-fet / bjt Sziklai pair. I did quite a bit of research to understand that. I had to, because it did not work. I tried many j-fets I had, and none would bias regardless of resistors twiddling. I found a small stash of 2SK170 and those worked right out of the box with minor adjustments. I hate j-fets for this habit, as much as I like their virtues, and I tend to avoid them except for buffering.
I measured the resulting input stage voltage gain to be 13.

So this front stage is perfect. At some point it starts to distort in a way that looks benign on the 'scope, and I wonder how much it contributes to tone if the gain of the first stage is maximized with the bypass cap. This Sziklai pair variety might even be an interesting distortion stage by itself.
That said, I did not even try the boost, because even with the drive knob at minimum, the whole unit did not sound to my like. YMMV.

I got some quality KSP13, seemingly the Fairchild equivalent to MPSA13. They were very close in Vbe.
I found that the output voltage was more than enough, despite the attenuation of the BMP tone stage. I came to dislike MOSFET stages due to their noisiness despite some very pleasant dark tone they impart, but I provided a simple transistor boost stage if needed. I probably revert that to a voltage follower. I found p-channel MOSFETs to be far superior in noise figures.
I believe that the ratio of the emitter resistors was chosen carefully to compensate for an alleged loss of gain due to the BMP stack, but on the general virtues of an output buffer I agree.

Tone. The pre-distortion bass cut is absolutely necessary, because this distortion generator suffers from bass overload no less than others do.
Or to put it the other way around, limiting the bass content into it improves the distortion texture. However, I find it slightly misleading to call the control "bass". From a control named "bass" I would expect something different. Friedman's "tight" is not obvious either. I have seen the term "body" for a pre-distortion bass cut, but this is probably just logical for me and is nitpicking on the highest level.

I find the tone stack that was used in the original unit to do very little, although you at least removed the pesky mid-drop. I may try a SWTC which I found very effective in similar situations.

For distortion texture, despite the benign distortion the look at the scope promises, the distortion texture is unimpressive.
I do neither hear any of the claimed tube tone (as heralded in the magazine version), nor any tone bender tone, nor any TS typical mid honk. It is ironical that the unaltered tone stack indeed removes some mids. As mentioned multiple times before me, the unit technically does not remotely even resemble either a tube screamer nor a tone bender of sorts. There is a faint wooly tone that is typical for all circuits that are thumped with bass. So that is a typical marketing hype.

I also don´t see why this unit is called a low-to mid gain OD. According to what people think "gain" means. IMHO it lacks the gradual increase from light distortion to a little more when picking harder. It becomes splatty real quick. I concede that some people like this and that that the perceived "lightness" of distortion is debatable.

My current output booster gain is 3, which already renders my volume pot nearly unusable at the lower range. I probably make it a buffer.
I personally do not see any scenario, where I would need such massive output drive. Letting alone the noise a MOSFET imparts.

Maybe there is more to tweak on the diff stage, but I cannot imagine. On a real LTP the ground leg resistor would be a current source.
Edit: I raised the value of the ground leg resistor to 33k, inspired by the description of the Keeley Fuzz Head. This makes the effect far more palatable.

 

[Chuck D. Bones]

Thanks for the feedback. Not every pedal circuit is going to appeal to every ear. I'm curious to know how your DC voltages compare to the voltages I marked on the schematic. Biasing has a strong effect on tone. With high-output pickups, it's possible to drive the 1st stage to clipping, which might not sound good to you.

Yes, MOSFETs can be noisy. I don't like using them in the front end of a distortion/OD/Fuzz fro that reason. Using them as a boost or buffer on the output side is ok. MOSFET noise is highly variable and I've found that in a bag of MOSFETs, some are quiet noisy while others are dead quiet. My guess is that the noisy ones have been damaged by ESD due to poor handling somewhere along the line.

Biasing JFETs is a pain, no doubt about it. A Sziklai pair is usually constructed using two BJTs and that's one way to solve the biasing problem. DC feedback to the gate is another solution that works well.

A lot of guitar players use the term "gain" when they mean compression. From an engineering perspective, gain in a dirt pedal is the linear gain from the input to the output of the last distortion stage. In that context, "gain" is the maximum amount of linear amplification available. Compression refers to how much the output level changes when the input level changes. Simply lowering the clipping threshold will increase compression. This circuit does not have a large amount of gain or compression. Unless you engage the TURBO switch.

A differential amplifier, aka LTP can have either a resistance or current source at the bottom. As the resistor gets larger, the circuit behaves more and more like there is an ideal current source. Is the difference audible? The vast majority of Fender amplifiers employ a diff amp as the phase splitter. None of them have a current source feeding the cathodes and yet they are commonly referred to as long-tailed pairs.

To quote Robert Keeley, the two most important features of a pedal that affect the tone are the name and the artwork. Putting the words "tube" or "bender" in a pedal name is pure marketing. I used that name here in reference to the origin of the circuit, not its tonal qualities.

 

[aquataur]

Your compound version the Sziklai pair I have not seen anywhere else. Really neat if you get it working.
The larger ground leg resistor made the feeling of the pedal totally different. Even engaging the TURBO switch is now palatable, although too much for me.

I find BMP tonestacks a PITA, I much prefer individual band controls. The pre-distortion bass cut works wonders. I replaced the BMP with a SWTC2, variant#1 but except R1. The role of this is taken over by the series output impedance of the diff stage. This unit does not really suffer from too high frequencies, but needs a little taming without messing with the other frequencies. It's really nice now.

I once made a Z.Vex Box Of Rock, which has several n-channel MOSFETs (BS170) stacked to allegedly mimic a JTM-45. It was noisy like a waterfall.
I reworked the PCB to take BS250 (p-channel) and used a voltage inverter. This unit is totally quiet.

I like the tube bender because of the way the distortion is achieved. The imperfections probably make its charm.
I also like your Keeley quote.

 

[Chuck D. Bones]

Did you measure the DC voltages in your TB?

I too like a lot of EQ knobs, but I don't put them all in every design. I used to hate BMP tone stacks until I figured out how to make them do what I want. I like the SWTC for some applications. It's a variable-freq filter as opposed to a shelving filter like the BMP, James & Baxandall. Works well when you need a LOT of treble cut. There are some interesting variations on the BMP such as the FATNESS control in the Roger Mayer Page-1.

 

[aquataur]

I have made a few designs with TMK (too many knobs) just to determine their value. I found a pre-distortion bass cut is invaluable in units like fuzz-faces, and also in units that get bass-heavier the more your increase the drive. I usually make provision for such a control in the PCB, and decide then if it is justified for it to consume front-panel estate. Which is usually the case for all distortion devices.
I retrofitted a bass cut on most of my guitars for on-the-fly adjustments.

The pre-distortion bass cut does a great job on the tube bender.
I have slung a SWTC onto it provisionally, but as you say, it is a heavy measure. Interestingly, the differential stage does not produce horrendous amounts of treble (after I lowered its gain), but I like to tame it a little. I am probably going back to the BMP you have there.
Good design, if I have not mentioned it before.

The vintage BMP I find pretty useless. On a muff I can see that the tilt between bass and treble has some merit, but the mid-cut is useless.
I find that people like Gilmour use a colorsound overdriver as an EQ after it to restore the missing mids.
Several ideas have popped up to restore that mid-range on site. I adopted the AION method with a 3 position toggle. I have my muffs always in flat mids position.

I have added the RM fatness control to my RM axis fuzz. This vastly improves the latter, but it is still an obnoxious little box.
I very much bedoubt that Hendrix used this. RM tells us a lot of fairytales and does not disclose the whole picture.
Like coily cables that had a horrendous capacity. All such units have to be seen in the context of the time.

All the DC voltages are exactly as per your diagram. 3.4 V on the emitter of the diff stage.

PS do you have an explanation, why the differential stage distortion sounds good? I think I have read about this but I can't find it again. Found it again: click

 

[Chuck D. Bones]

"... a hive full of angry bees."

I had that in my attic once and it is a very intimidating sound.

 

[aquataur]

I can see what Willmott is driving at. You can wind the drive up and it does the bee-hive, BUT only if you don't limit the bass into it. It behaves like an early Marshall amp, which was nothing but a Bassman put upside down. With the advent of the Plexi a 720 Hz highpass was introduced that made it tight.

I find that this unit excels in mimicking the proverbial "tube amp on the verge of break-up" sound. Like most others I encountered, it is not the first choice for other disciplines. But indeed it works exceptionally well as the low man on the totem-pole preceded by other dirt boxes, such as my favorite F-109, which is a Björn Juhl knock-off using BC549's throughout. It masks and compresses obnoxious frequencies very well.

That said, when I engage the "boost" it does a convincing fuzz face tone by itself that can be controlled with the guitar's volume AND the bass cut (on the guitar, if you have one). I begin to understand the "bender" bit in the name. I am starting to believe that the "tube" part was not a hint at a tube screamer, but rather at its tube-like qualities, and of course at the machinery for metal tube bending.

As different to many other units it does not become mid- heavy if you turn up the gain although you may have to readjust the bass going into it to keep the bee's swarm in check. Cutting the bass is akin to treble boosting. So here you go... all the old folks used a treble booster...