Big Muff Triangle voltages?

[LucifersTrip]

I've finally ventured outside the germanium realm again to take a break from my 1000 comp build and just breadboarded this triangle Muff:



...and it sounds excellent right off. I made a note that all collector voltages should be around 4V, though, with a 9.6V supply, I'm getting ~ 5.2V - 5.6V on each of the collectors.

A couple questions. If (and I suppose they are) higher voltages than I should be getting, which transistors are best to manipulate so the overall original tone won't suffer. I have a choice of the base to (-), collector to (+) or even the emitter to (-)

The original transistors were notes as 2N5133, which have a minimum hfe of 50...which usually means that a typical one would probably be in the 50-100 range. I chose similar vintage plastic dome style transistors each with gains of ~ 75-80.

I've read threads where people are building this with modern high gain ones like the 2N5088 with gains of 500+. Does that really not matter?

thanx

[R.G.]

The transistor type makes little difference. Some, perhaps, but not much. All four stages are set up with both AC and DC feedback controlling much of what they do.

And on EH products from that period, "original tone" is a little misleading. There is not one "original tone" because they used widely varying parts, even with different nominal values, and some examples of Big Muffs were made with PNP transistors instead of NPN.  It's hard not to get inside the range of "original tone" because that varied widely.

There are people who will argue fine shades of tone (which itself has become largely meaning-free these days) but I keep thinking of the story of the blind men and the elephant.

[PRR]

> it sounds excellent right off.

So who cares about pin voltages?

As R.G. says, E-H just slapped these together with whatever parts they found laying around. (Early E-H production was in a district full of electronics assembly- Mike could walk across the street and borrow a cup of resistors, or even find them in the street (as I did once: whole box of 47 ohm 1W strewn down the pike).

Zero would be bad. 9V would be bad. Anything not slammed to either extreme is good enough.

If you must fiddle, try changing a 390K to 360K or 330K (or 470K the other way).

[LucifersTrip]

again...thanx much for the replies. With the vintage germanium stuff, I'm so used to gain & leakage have such a large effect, it's surprising that such a large range of gains here wouldn't matter.

There has to be some limit. I'm sure I can't throw 4 transistors in there with gains of 25!

I'll just do a little experimenting as I always do, anyway

>> it sounds excellent right off.

>So who cares about pin voltages?

well, you know how it is....it can always be excellent-er
...and I don't have an original to compare it to


thanx

[LucifersTrip]

after the first round of experimentation, here's what I got...

I lowered the collector voltage to 4.5V, (from 5.2V - 5.6V, where they are now) by altering the 22K and 12K's from collectors to (+).

Q1 thinner, trebly tone & wimpy
Q2 small decrease in volume, thinner, trebly tone
Q3 thinner, trebly tone, raspy, lost fat muff tone
Q4 sharper, brighter tone, again ,more trebles, less warmth

...and higher gain transistors, of course, will lower the voltage on the collector without any resistor manipulation. Using 2N5089's with gains of ~600 dropped the collector voltages to ~4.4 - 4.5V for Q1-Q3 and 3.6V for Q4.  The effect was basically the same as lowering the voltages by altering the 22K and 12K's...thinner & trebly with less warmth (I tried each one separately and then swapped all 4). No wonder I hear so many triangle clones that sound harsh to me.

It's a cliche, but the lower gain ones were warmer all the way around.

I'm going to leave the 80 hfe's in Q1-Q3, but I'm have a difficult time deciding on what I want in Q4.
The lower gain ones can sound better with the tone toward the treble and the higher gains ones can sound better vice-versa.

As I noted, a 2N5089 in Q4 drops the C voltage to 3.6V.  A transistor with gain ~ 175-200 dropped it to right around 4.5V

I'll do some more experimenting altering the 390K's as Paul suggested next.

[R.G.]

There has to be some limit. I'm sure I can't throw 4 transistors in there with gains of 25!

There is a limit. You've stumbled into the underlying principle of 70s transistor design. And in fact, more fundamentally underlies a lot of all kinds of engineering.
The fundamental version is this: whatever cannot be reliably controlled must be made as irrelvant as possible.

In particular, EEs realized that the wide variations in transistors could not be controlled well enough by sorting, pre-selection, matching, and so on to be really usable. So a whole host of techniques relying mostly on negative feedback, was developed to make the variations in particularly gain be irrelevant as long as there was enough.

In the Big Muff, there are two basic circuits; one kind is the output amplifier. This uses the 'stabilized bias' circuit, sometimes called the 'four-banger' from the use of four transistors, two to power and two to ground. The DC and AC feedback provided by the emitter resistor stabilizes the DC bias point against both gain differences and thermal drift. Getting enough stability in this circuit usually limits gain to about 10-20, and requires a fairly large emitter resistor, lowering the output swing.

The second king of circuit has a collector and emitter resistor, but with a much smaller emitter resistor. The collector-to-base resistor provides both DC and AC feedback to bolster that from the smaller emitter resistor. This gives a bigger output voltage swing. The overall AC gain is set by both the collector-base resistor and the series resistor to the base. If this reminds you of the inverting-input resistor and feedback resistor of an opamp [wink, wink, nudge, nudge] it should. With infinite open loop gain, a feedback amplifier with this setup gives a gain of -Rf/Ri. With lower gain, it's less than -Rf/Ri, but it comes close. In general you need at least ten times as much open loop gain as closed loop gain to make the assumptions be reasonably true. This second kind of circuit suffers compared to the stabilized bias circuit in having a lower input impedance.

In both circuit types, the actual gain of the transistor makes almost no difference in its DC or AC performance, as long as it's big enough. So the question of how to get the right transistor gain for circuit function became instead "how can I get enough gain to make negative feedback control it all?" When you get to there, transistor gain being larger makes the actual transistor gain irrelevant.

So - is there a limit? Yes. But it's a lower limit.

I lowered the collector voltage to 4.5V, (from 5.2V - 5.6V, where they are now) by altering the 22K and 12K's from collectors to (+).

Lowering these lowers the open loop gain of the stage; raising it raises the open loop gain of the stage. If you lower the open loop gain inside a feedback amp, it can no longer do as good a job of 'hiding' what the transistor's raw performance is.

Another issue is the feedback clipping diodes. These things clip less ideally as open loop gain goes down. And as we all know now, changing how hard a diode clips affects how it sounds.

People pontificate endlessly about this or that transistor making this or that sound, when it's often just what happens when the internal gain or operating point changes.

[PRR]

Note that this (like all transistor circuits) is sensitive to both hFE *and* Vbe. The 390K+82K form a voltage divider. The transistor won't be "happy" unless its base is at Vbe (plus the small drop in 820r), so the top of the 390K "must" be whatever makes the voltage divider give a happy base voltage. Vbe could be 0.45V to 0.7V, though more likely 0.5V-0.6V for "suitable" (small and cheap) devices. Right there is 20% variation.

And of course the "voltage divider" is more or less heavily loaded by base current, which is influnced by hFE.

A plot would be nice but my stimulator is not co-operating.

Table for Q1 with several different assumed hFE values and a "suitable" Vbe:

hFE ___ Vc
1 _____ 8.2
3 _____ 7.77
10 ____ 6.7
30 ____ 5.54
100 ___ 4.65
300 ___ 4.28
1000 __ 4.14
3000 __ 4.1

Note that 3,000 to 1 variation in hFE gives only 2 to 1 variation in Vc. hFE "hardly matters", as long as it is not very-lame.

If a nominal goal is "4.5V", then any hFE from 50 to 3,000 is spot-on within 10%. By the early 1970s, E-H could be sure of getting hFE>50 even buying sorted rejects for a penny apiece.

The "4.5V" is not magic, it's simply a split-difference between zero and 9V. Say you come home, very late at night, a little stoned/tired, on a bicycle, and must wobble into a 9 foot wide garage door. Where do you aim? Not a foot from the left or right, you need lots of wobble room. The ideal answer is 4.5 feet from each side. However 4 feet or 5 feet is probably just as good. If you only wobble 3 feet, it's all the same. If you wobble 10 feet from your intended path, well, you are gonna hit one of these nights. In this case, the artistic goal is to mash your shoulders, both about the same, and you deliberately wobble until you hit. Whether you hit 4 feet one side and 5 feet the other is not real critical (the diodes at Q2 and Q3 will even-out any asymmetry to a neat 0.6V each side).

And hFE=10 will "work". The 33:66 asymmetry isn't too important. Raw voltage gain drops 10%, hardly a big deal. Perhaps the main effect of hFE=10 is a very low input impedance (~~8K) sucking the top treble off the guitar, and that's often acceptable in a fuzz.

IAC, the most direct way to "tweak" your collectors to 4.50000V is to diddle the 390K resistors, because this shifts both the Vbe and the hFE feedbacks.

[LucifersTrip]

In both circuit types, the actual gain of the transistor makes almost no difference in its DC or AC performance, as long as it's big enough. So the question of how to get the right transistor gain for circuit function became instead "how can I get enough gain to make negative feedback control it all?" When you get to there, transistor gain being larger makes the actual transistor gain irrelevant.

yes, as shown by Paul's hFE ___ Vc chart, once you hit hFE 100, there's a much smaller variation (dropping) of Vc. Though, that's for Q1 and Q4 is biased differently. A 600 hFE in Q4 gives me 3.6V on the collector....and to my ears, there's a significant difference in treble/warmth between Q4C at 4.5V and 5.5V

I lowered the collector voltage to 4.5V, (from 5.2V - 5.6V, where they are now) by altering the 22K and 12K's from collectors to (+).

Lowering these lowers the open loop gain of the stage; raising it raises the open loop gain of the stage. If you lower the open loop gain inside a feedback amp, it can no longer do as good a job of 'hiding' what the transistor's raw performance is.

I guess, then my question would be, do you always want to do a good job of 'hiding' what the transistor's raw performance is?
...and if I raise the voltages instead of lowering & raise the open loop gain of the stage, does it do a better job of 'hiding' what the transistor's raw performance is? ...if so, is there a limit to better?

[LucifersTrip]

Table for Q1 with several different assumed hFE values and a "suitable" Vbe:

hFE ___ Vc
1 _____ 8.2
3 _____ 7.77
10 ____ 6.7
30 ____ 5.54
100 ___ 4.65
300 ___ 4.28
1000 __ 4.14
3000 __ 4.1

again, you're right on the money. an hfe of 225 in the Q1 slot gave me Vc ~ 4.5

though, Q4 is biased differently, with a much bigger change in Vc as hfe changes, ie:

hFE ___ Vc
80 _____ 5.8 (corrected from 5.6)
175 _____ 4.9
600 _____ 3.6

...and the change in tone from 5.8 to 3.6 is pretty obvious.

The original transistors were 2N5133, which, in data sheets have a much larger hfe min/max range than I'm used to seeing (60 - 1000!)
http://www.datasheetarchive.com/dataframe.php?file=Document21138.pdf&dir=Databooks-5&part=2N5133#

Though, hardly any of my 1960's plastic dome transistors have any gains over 200

If a nominal goal is "4.5V", then any hFE from 50 to 3,000 is spot-on within 10%. By the early 1970s, E-H could be sure of getting hFE>50 even buying sorted rejects for a penny apiece.

The "4.5V" is not magic, it's simply a split-difference between zero and 9V.

yep, exactly...I pulled out my Green Muff yesterday, which I kept because it sounded so damn good. It turns out Q4C (5.45V w/ 9.25V supply) on that killer Green  is right around what I'm getting on the Triangle (5.8V w/ 9.67V supply). nice!

I'm not 100% sure if those 4 resistors (390K/100K, 12K/2.7K) in that final stage are exactly the same in the Green, but I can definitely see one is 2.7K. The other 3 are unmarked.
http://www.luciferstrip.com/fuzz/green-russsian-guts.jpg

Say you come home, very late at night, a little stoned/tired, on a bicycle

Albert Hofmann

IAC, the most direct way to "tweak" your collectors to 4.50000V is to diddle the 390K resistors, because this shifts both the Vbe and the hFE feedbacks.

would it be better to not alter any resistors and just put in the correct gain transistor that gives you the voltage you're shooting for?

[R.G.]

yes, as shown by Paul's hFE ___ Vc chart, once you hit hFE 100, there's a much smaller variation (dropping) of Vc. Though, that's for Q1 and Q4 is biased differently. A 600 hFE in Q4 gives me 3.6V on the collector....and to my ears, there's a significant difference in treble/warmth between Q4C at 4.5V and 5.5V

Human hearing happens mostly in the brain, not the ear. There are slight differences, especially if the bias shift causes a difference in distortion.

I guess, then my question would be, do you always want to do a good job of 'hiding' what the transistor's raw performance is?

No. Especially in musical effects, the precise amount, color, and style of distortion matters a lot to all those brains doing the listening. Un-hiding a certain bit can and does affect tone. One good example is the sharpness of clipping on overload. An opamp, with a huge internal gain, gives quite sharp 'corners' on clipping, as the thing has so much gain it can preserve semi-ideal signals for a long time into overload. A transistor, with a raw gain of only maybe 50-200, can't do as much hiding as it runs out of gain, so can't make as sharp an overload signal when clipped. The art, if there is one, is deciding how much and what kinds of distortions you want, and then being able to go get them.

...and if I raise the voltages instead of lowering & raise the open loop gain of the stage, does it do a better job of 'hiding' what the transistor's raw performance is? ...if so, is there a limit to better?

The voltages and open loop gain of the stage are related, but not necessarily directly. It's that AC versus DC thing. DC bias point gives you the DC voltages. Resistor *and* cap values give you open and closed loop gain values sometimes separately from the DC conditions. Higher open loop gain always gives more hiding of the underlying nonlinearities, but especially in circuits with multiple feedback loops, local feedback in a feedback loop can change what you get when the outer loop runs out of gain.

And on the question of "better", you have to be careful to state what "better" means. More linear may not be 'better' in a distortion pedal.

[PRR]

> a much larger hfe min/max range than I'm used to seeing (60 - 1000!)

In older production, out of the pot, a batch could give anything 20 to 1000.

You could pay for a 100-200 sort. Or a 500-1000 sort, but that could be expensive since some days they didn't get any parts over 500.

OTOH you could pay less for a "I'll take almost anything with gain" part. Essentially the left-overs after the high-price parts were sorted out. Maybe, like the discount shoe-cellar, they don't have any size 9 or 10, but lots of sizes 2 3, 13 and 14.

> Q4 is biased differently, with a much bigger change in Vc as hfe changes

The others have DC collector-base feedback; Q4 does not.

However I would not expect any real difference from 5.8 to 3.6 for typical guitar-cord levels.

> would it be better to not alter any resistors and just put in the correct gain transistor that gives you the voltage you're shooting for?

What are you doing?

If you are making a million of them, you don't want ANY individual fiddling. You use a few more very-low-spec transistors wrapped in DC feedback so every unit comes out "working good enough". (This is cheaper than using fewer transistors with very tight specs.)

If you are benching one or a few, do what you want. The combinations will be mind-boggling.

[LucifersTrip]

>However I would not expect any real difference from 5.8 to 3.6 for typical guitar-cord levels.

correct...the difference is mainly on single notes on the high E and B strings, when the tone is approx 75%+ towards the treble

>> would it be better to not alter any resistors and just put in the correct gain transistor that gives you the voltage you're shooting for?

>What are you doing?

just making one pedal...it was really just a generic question: if you have the luxury of choosing from a large assortment of transistors, is it better to pick a transistor to get the "recommended" voltage rather than adjusting surrounding resistors...

[R.G.]

correct...the difference is mainly on single notes on the high E and B strings, when the tone is approx 75%+ towards the treble

With today's transistors, assuming that they are not somehow already abused, this suggests not a transistor selection or biasing problem, nor a resistor selection issue, but a close-to-oscillation condition. Subbing transistors in a feedback-controlled series of stages just doesn't do that kind of treble change. What it can do is set up a near-oscillation condition that emphasizes some frequency even if there is not actual squealing.

This is just a guess, but I would be very suspicious of filtering/near oscillation (which are two ways of saying the same thing BTW) for any major frequency response shift when just changing devices if the DC bias voltages don't get near a power supply point.

Interestingly, this sort of thing is one of the big pitfalls to new boutique pedal builders who learned to solder two weeks ago and who are now subbing in a few parts to make their new "design" be the hottest thing to ever hit the internet. Something may be happening when one subs in a part. But it may not be what you think it is, and may not be for the reasons you think it is.

Sorry - mind wandering again.

just making one pedal...it was really just a generic question: if you have the luxury of choosing from a large assortment of transistors, is it better to pick a transistor to get the "recommended" voltage rather than adjusting surrounding resistors...

It depends on the nature of the design. If the design depends heavily on the nature of the transistor, like a Rangemaster clone, Fuzz Face clone, etc., then select transistors. If the design is based on more modern techniques like making feedback control the operation, swapping transistors may do next to nothing.  But on a more modern (and this means late 60s onward) design by a pro will use feedback stabilization, and swapping resistors to get the voltages right may throw off the AC gain and other conditions. Kinda like the C&W song The Gambler - you gotta know when to hold 'em, know when to fold em, ... every hand's a winner and every hand's a loser...

[culturejam]

In both circuit types, the actual gain of the transistor makes almost no difference in its DC or AC performance, as long as it's big enough. So the question of how to get the right transistor gain for circuit function became instead "how can I get enough gain to make negative feedback control it all?" When you get to there, transistor gain being larger makes the actual transistor gain irrelevant.

Don't tell that to the majority of the boutique world!   

[LucifersTrip]

In both circuit types, the actual gain of the transistor makes almost no difference in its DC or AC performance, as long as it's big enough. So the question of how to get the right transistor gain for circuit function became instead "how can I get enough gain to make negative feedback control it all?" When you get to there, transistor gain being larger makes the actual transistor gain irrelevant.

Don't tell that to the majority of the boutique world!   

that's only true for a small percentage of vintage fuzzes that boutiquers build

[Doctor Algernop]

The transistor type makes little difference. Some, perhaps, but not much. All four stages are set up with both AC and DC feedback controlling much of what they do.

And on EH products from that period, "original tone" is a little misleading. There is not one "original tone" because they used widely varying parts, even with different nominal values, and some examples of Big Muffs were made with PNP transistors instead of NPN.  It's hard not to get inside the range of "original tone" because that varied widely.

There are people who will argue fine shades of tone (which itself has become largely meaning-free these days) but I keep thinking of the story of the blind men and the elephant.

YES! I'm the one to get R.G Keen to 666 likes