Build Report : Colorsound Tone Bender 3 knob - bias troubles

[flektor]

Hi guys,

I´ve had serious trouble (fun) adjusting the bias of a Colorsound Tone Bender 3 knob. I´m repporting my experiences with the circuit, so that you can correct me if I´m wrong or add some observation that can help me improving the project. I´m no technician, so forgive me if I´m saying anything silly.

I´m following the General Guitar Gadgets: 
http://www.generalguitargadgets.com/diagrams/tb3cs_sc.gif
but I designed a brand new layout.

Differently from what I´ve read in several posts, it´s not an easy bias circuit. The easy thing is that with almost any good trannie the fade out of the pedal is relativelly good, giving the impression that things are ok. But I´ve noted some dramatic tone changes relative to the transistors that I used (2SB187, CV7003, metal can Mullard OC76, metal can Phillips OC74D), and at different hfe gains.

I also disagree also with the idea that Q1 and Q2 are not involved in the main tone generation. With different transistors in Q1 and Q2 I´ve had real tone changes. But it´s true that Q3 is crucial to the overall fuzz tone os the Colorsound TB.

Researching a little, I realized that the first stage is a Darlington pair. The problem that follows is that in the Colorsound 3 knob we can´t simply follow the Fuzz Face bias procedures (as exposed in RG Keen´s Technology of the Fuzz Face), as we certainly can do with the Tone Bender MKII.

This went because the Darlington Pair "is two transistors connected together so that the current amplified by the first is amplified further by the second transistor. The overall current gain is equal to the two individual gains multiplied together:

Darlington pair current gain, hfe = hfe1 × hfe2
(hfe1 and hfe2 are the gains of the individual transistors)

This gives the Darlington pair a very high current gain (...), so that only a tiny base current is required to make the pair switch on. A Darlington pair behaves like a single transistor with a very high current gain".
http://www.kpsec.freeuk.com/trancirc.htm#darlington

The author follows explaining that the first transistor of the pair can be a low power, but that the second one MUST be a hi power.In this sense it´s impossible to follow in the Colorsound a la letre the bias advices of RG Keen relative to the Fuzz Face bias: a circa 70hfe trannie in Q1 and a c. 130hfe in Q2, in order to obtain 4.5V in Q2 collector. We must consider the first stage of the Colorsound as a Fuzz Face Q1 and Colorsound Q3 as a Fuzz Face Q2, and we must look for different hfe values and other collector voltages.

Things got complicated when I tryed the mod from the schematic of the Boutique Tone Bender posted in the General Guitar Gadget.
http://www.generalguitargadgets.com/diagrams/tb3bb_sc.gif
This mod  (I´ve heard it´s supposed to be a Mike Fuller´s Soul Bender mod) is to extract the 10k resistor ("R10" in GGG schematics), and add a 10k resistor and a 500 K trimmer in series from the collector to the base of Q3. This thing is supposed to act as a controller of the overall agressivity of the fuzz. 

I believe that this disposition is a variation of the generally called "static" or "fixed potarization" (Am I right?). Tradicionally it was intended to stabilize the operational point of the transistor. That´s ok. The problem is that it also reduces dramatically the gain of the circuit stage in question. Q3 collector´s voltage with this mod went to circa 1.5 to 2.0V as I turned the trimpot, while Q1 and Q2 where giving me 3.0 to 4.5CV, depending on the transistors that I used. With the 500K trimpot the sound went a little sweeter, but the overall gain of the pedal was lower (too low if I turned the trimmer), maybe changing too much the character of the original Colorsound. Good if you like overdrive-fuzz textures, bad if you want beast capable of something more. That´s why I gave up the Boutique mod and returned to the stock Tone Bender schematics.

Without the Boutique mod things got better. I´ve tried many transistors and gain configurations. By the end I discarded the multimeter and by ear I decided for a combination of 1 Mullard OC76 and 2 Phillips OC74.

When I verified it with the multimeter I had the following values:

Q1: OC74 hfe 80 - Collector voltage 2.68v
Q2: OC74 hfe 80 - Collector voltage 2.68v
Q2: OC76 hfe 67 - Collector voltage 3.99v
With a 9.18 v battery.

(The gains where obtained with the help of Keen´s Ge Transistor Tester exposed in The Technology of the Fuzz Face).

:icon_smile:Good! Agains all expectations, Q3 voltage got higher and it sounded fantastic -- strong (but not too much) and well defined. The sound cleans a bit with the guitar volume down, and I´ve noticed a very cool fade out effect, kinda ghost feedback note that sometimes show up behind the main distorted texture fading out. The tone went to the side of Yardbirds and Cream!

Some directions that I´ve abandoned:

1) I attempted to made Q3 collector go to 4.5v with 100hfe CV7003 (military metal OC44 similar). It sounded ok, more buzzlike, but not too much. As I turned Q3 hfe higher (120 over) I went to the neighborhood of Doors Maestro, generating some velcro and buzzyness.

2) I´ve also tried to reduce Q1 and Q2 gain with two OC76 around 30hfe and 40hfe, but it reduced too much the overall gain. Adding a 120 hfe CV7003 with these OC76s did not correctec the problem. Seems like the gain depends too much on the Darlington pair.

3) At first I´ve considered two 2SB187 (AC126 similar) circa 80 hfe in Q1 and Q2, believing that they would not influence in the tone. Sounded ok, but nothing like the OCs. From this I concluded that the Darlington pair is important to the overal character of the sound.

4)I´ve tried to chage values of R18, R9, R4 or R3 (GGG) with trimpots, but it didn´t helped me to increase Q3 collector voltage. The only bias resistor that changes Q3 cv is the 10k (GGG R10) resistor. If chaged to a 20k trimmer it can put the voltage in 4.5, but you´ll have no dramatic tone changes (maybe because it also raises the cv of the Darlington Pair proporcionally with Q3 cv). But it is a good controls the output level.

I´ve concluded that the best combination is to have the 3 real OCs in the circuit, and elected this combination. The CV7003 sounded ok, but with it in Q3, even with 80hfe, the sound losted that incredible ghost feedback note that I discovered and intend to maintain. (the CVs will probably go to a future MKII, let´s see if they can give me what I expect from them).

Cheers, Orlando.

[flektor]

CORRECTION:

Sorry, I posted a wrong voltage and Q number. Above in Q3 is a CV7003 100hfe collector voltage.

The correct is as follows:

Q1: OC74 hfe 80 - Collector voltage 2.68v
Q2: OC74 hfe 80 - Collector voltage 2.68v
Q3: OC76 hfe 67 - Collector voltage 2.62v
With a 9.18 v battery.

[Ge_Whiz]

Oh, er, um... I built the Vox Tonebender MkIII from Fuzz Central, threw in two lowish gain PNP silicons for Q1 and Q2, a decent germanium for Q3, and it sounded great. Sorry, I didn't really approach this in the technical detail that you have... 

[petemoore]

  Q1 and Q2 are looking like a darlington discrete pair.
  For the most part biased with the 3k3/uf at emitter, collectors both at same voltage of course, alot like a single darlington stage except
1 the 10k from Q1e to ground drops gain
2 the B/C 220pf cap rolls off a little bit of gain in HF range.
  Other than that it looks as though Q1/2 are mostly biased as a single high gain stage. I'm not sure I understand that 10k on Q1e.
  Q3...reverse bias diode on it's B/E...I won't venture to speculate...other than mine held what I thought was 'good' bias.

[flektor]

Thanks Pete,

I noted that the 47k from Q1 collector to ground also lowers the gain of the first stage Q1+Q2.

I´ve tryed several diodes and had no tone changes. Seems the  diode is there only to protect Q3 from temperature variations.

[petemoore]

  Now that you point that out, I see Q3 actually has a variable bias resistance between B/C, the 500kpot + 10k stop R.

[flektor]

the 500kpot + 10k stop R are only in the GGG Boutique Bender, not in the stock Colorsound 3 knob schematics.

I´ve tried this trick, but the bias changes with the 500k pot were very low, besides the pot. tended to lower too much Q3 gain.