Colorsound Tremolo

Started by RLBJR65, August 13, 2004, 02:39:27 PM

Previous topic - Next topic

PRR

If you want all speeds slower, increase all *three* caps.

Go back to original values. Now tack three more 220n across all three 220n, so they are all now 440n. The speed will be *half*. You can scale the rate much further either way, to once a minute or the middle of the audio band.

As for why it "weakens".....

The drawing is correct but roundabout. I've re-drawn for clarity of the C-R-C-R-C-R network.


The simplest form makes all three C-R pairs the *same*, so they add phase-shift at the same frequency.

(A variant "tapers" the values, and will work with a little less gain, but is more demanding of the amplifier parameters, so would be more complicated than this.)


We need three C-R phase-shifts, all nearly the same. Two won't do. The three C-Rs do not have to be the same frequency but if they are far apart they don't add-up to enough phase-shift to oscillate.

C1 R1 is 220n and 100K. C2 R2 is 220n and a pot which is about 100K at mid-settings. C3 R3 is funny. Rb is 3m3, but if the thing is oscillating then the transistor gain is near 27. Miller Effect works on Rb so that for *signal*, Rb "looks like" about 120k, and we can use that value in the AC analysis.

(Rb can not be picked arbitrarily because it also sets-up the DC bias in the transistor, and the bias is fine.)

So as-designed at mid-point we have 100k ~~100k and 120k. All three are near the same.

At one pot extreme we have 100k 225k 120k. Not the same but not very different. The 2:1 difference still has enough phase-shift to oscillate. (The other extreme, 4k7, looks dubious but you are not complaining.)

Now you change R2 to 1Meg, for 100k 1,000k 120k. That 10:1 difference is too much, there is not enough total phase-shift to oscillate, it goes weak and then stalls.

The Designer did pick reasonable resistor values *and* ratios. You won't do much better.

The cap ratio is also fine.

However the values of the caps can be selected arbitrarily, and bigger caps is directly lower frequency. Change the caps.
  • SUPPORTER

Rob Strand

#21
A couple of things up front.  I'm not 100% sure the layout for the Colorsound Temollo follows the correct schematic.  Also, I'm *NOT*  sure there is a verified schematic for that one.  I've seen three variations.   I've got some pictures of the PCB and they only show one  3.3M resistor.

Quote1- What is the technical reason of the "tremolo vanishing" ?
When you design phase-shift oscillators the RC network has an attenuation (a signal loss) and the transistor has a *signal* gain.  When the gain of the transistor is higher than the loss at the oscillation frequency the oscillator will oscillate.  When you mess with the values on the RC network it can make the loss go up and when the transistor amplifier doesn't have enough gain it won't oscillate.

When you have a speed pot it messes with one of the resistors and messes with the loss of the RC network.   For given transistor amp there is a range of speed pot resistance values where the oscillator will oscillate.  So there is fixed ratio of maximum frequency to minimum frequency.  Trying to get more adjustment doesn't work as the oscillation will die out.

QuoteOne question, trying to understand the circuit better : what would happen if i select different cap values (instead of 3 times 470nF) ?
If you scale *all* the caps up and down you can change the frequency of the oscillation but you cannot change the ratio of min to max frequency.  So by making all the caps half the value the frequency will double.
By carefully choosing caps you can centre the available frequencies over the most useful range.  In that way you don't need to have such a wide adjustment ratio.

From what I remember the colorsound oscillator adjusts the resistor to ground after the second cap (starting at the collector)  whereas something like the Electronics Australia Tremollo adjusts the resistor after the first cap (starting at the collector).  I believe adjusting the first resistor gives you more range of adjustment before the oscillator dies.

Going back the uncertainty of the true circuit. 
- If the collector resistor is too low you will end-up with less gain and less range of speed adjustment.
  I think the circuit with the 100k collector resistor may actually be better.
- If the DC bias point of the transistor is off the gain might be low.   That will also reduce the range of speed adjustment.

The layout shows a 4.7k resistor on the collector and two 3.3M resistors on to the base.  That configuration was originally published for a darlington transistor  and I wasn't even sure it was the colorsound circuit.

The circuit I saw as "original" used a 100k collector resistor and a single 3.3M from the collector to the base.
You could tune the 3.3M resistor by lifting one of the oscillator caps then adjusting the 3.3M resistor to get say 4.5V DC on the collector.

Having said that I don't know what else is different in the circuit.  Maybe changing the 4.7k to 100k introduces other issues in the circuit. [EDIT:  in this case maybe not as the circuit here is the darlington + 4k7 version]

If anyone has a schem verified against the original PCB I'd be interested to see it!
Even some shots of the track side of the PCB would help.

(Edit: fixed a lot of typos)
Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

Rob Strand

QuoteThe drawing is correct but roundabout. I've re-drawn for clarity of the C-R-C-R-C-R network.
That's looks a lot like the "original" ckt but it doesn't line-up with the layouts I've seen.
Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

snk

Wow, thank you, that's a great in-depth explanation !

Rob Strand

#24
FWIW,

This page speaks of the Original 4-transistor circuit.   You can see a Darlington, 2x3.3M resistors and a 4.7k collector resistor:
http://effectslayouts.blogspot.com/2017/08/colorsound-tremolo.html

So most layouts I've seen (PCB and Vero) follow that "4-transistor" configuration.

Here's the schematic (maybe),




It also mentions "the re-issue is the 3-transistor version".

I don't know if the three transistor version is *only* on the re-issues.

A pic is here,
http://www.effectsdatabase.com/model/colorsound/tremolo

1x3.3M, 2x100k, 4.7k, 47k, 3x220n       ; one of the resistors isn't for the oscillator

Maybe someone more knowledgeable on the finer points can clear things up.
Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

Rob Strand

#25
First version:
Three Transistors. Narrow Box. 1973 to 1975.
C7 seems to be backwards in real units.

Pot tapers are not specified



2nd version:
Four transistors. 
See previous post for schematic.

Wide Box 1976 to 1985 (probably without parts in brackets)
1990's Re-issues (probably with parts in brackets)


Macaris re-issues (from approx 2016):
Seem to use the first version.

Most vero layouts I've seen use the 2nd version.
Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

StephenGiles

I think that I drew my circuit from the unit I had, tested by breadboarding and found it to be correct.
"I want my meat burned, like St Joan. Bring me pickles and vicious mustards to pierce the tongue like Cardigan's Lancers.".

Rob Strand

QuoteI think that I drew my circuit from the unit I had, tested by breadboarding and found it to be correct.
What I've drawn isn't that much different.   What you had before will probably work.

I've got pics for a number of boards.  In some areas the pics all have the same part values but those parts don't agree with your schematic (see effects database link and comments a couple of posts back).   I found an old fsb thread and one of the posters (IIRC Analog Guru) said something along the lines of what I have drawn.

The main point I wanted to make is there's two versions.
Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

snk

Hello,
Two last questions :

- I have replaced the 4.7K resistor with a 5K pot, acting as a volume knob. It is very efficient, but when i turn it it makes noise : is it because it is biasing the transistor ? Can something be done to get rid of it ?

- While the protoype was sounding great (even without enclosure) plugged into my mixer, now that it is enclosed, it is a bit noisy when plugged into my amp (my Danelectro Cool Cat is noiseless on the same amp) : what can be the cause ? It is not really hum or buzzing (so i don' suspect grounding), more of a general white loise underlying...


Rob Strand

#29
QuoteI have replaced the 4.7K resistor with a 5K pot, acting as a volume knob. It is very efficient, but when i turn it it makes noise : is it because it is biasing the transistor ? Can something be done to get rid of it ?
Yes, it's due to bias shift.  Check out the way a fuzz-face does it.  You need to wire pot pin 1 to gnd, pot pin 3 to the emitter and the wiper (pot pin 2) to a large cap, then the cap goes to ground.   The cap needs to be large like 47uF.  You might want to put a 470 ohm resistor in series with the cap so the gain doesn't go nuts at the end.

Having said all that that circuit has a 330k feedback resistor and that means you need to make large changes to the 4k7 to get noticeable changes in gain.   

Another way to adjust gain is to adjust the feedback resistor but you need to do that in a way that doesn't affect the bias point.

Quote- While the protoype was sounding great (even without enclosure) plugged into my mixer, now that it is enclosed, it is a bit noisy when plugged into my amp (my Danelectro Cool Cat is noiseless on the same amp) : what can be the cause ? It is not really hum or buzzing (so i don' suspect grounding), more of a general white loise underlying...
You might find some of that is related to the DC shift and extra gain.    The 100k's on the input of the trem don't help since they add noise.  I haven't built that one so I don't know what the base-line noise is.
Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

snk

Hi, Rob
Thank you for your in-depth reply, that is very useful.
I suspected a bias shifting, but i didn't had a clue to the way to fix it. I will see if i can manage to get rid of it (it is not very annoying, as the volume of the tremolo is quite "sest and forget", but still, i like to do things right :) ).

Quoteyou need to make large changes to the 4k7 to get noticeable changes in gain
What do you mean by that ?
You mean that I need to change the resistor (which is now a pot) value ?
Or you mean that turning the knob a lot will result in small volume changes (I don't need a huge volume change, as it is now it is perfect) ?

QuoteYou might find some of that is related to the DC shift and extra gain.    The 100k's on the input of the trem don't help since they add noise.  I haven't built that one so I don't know what the base-line noise is
Would it improve the signal/noise performance if i change the 100K value ? I was surprised to have a perfectly clean and silent pedal on one system, getting noisy on the amp.

Rob Strand

#31
QuoteWhat do you mean by that ?
You mean that I need to change the resistor (which is now a pot) value ?
Or you mean that turning the knob a lot will result in small volume changes (I don't need a huge volume change, as it is now it is perfect) ?
It's hard to explain without too much theory.  The amplifier gain acts like an inverting opamp stage.   The feedback resistor is the 330k and the input resistor is the 100k (roughly)  so the gain is 330k/100k = 3.3.   The opamp has a very high gain which makes the actual gain equal to the predicted 3.3.   In the transistor case things are far from ideal.   The transistor amplifier gain [when I say "transistor amplifier gain" I means the signal gain without feedback] isn't that high when Re=4.7k the actual gain ends-up being lower than the predicted 3.3, perhaps only half what is predicted by the resistor ratio.   However when Re is reduced (or bypassed with a cap, which preserves the DC biasing) the transistor amplifier gain increases and the actual gain with the 330k + 100k will start to approach the predicted gain of 3.3.  So you might only get 2:1 gain change for larger changes in Re.

If the transistor amp didn't have the 330k present the gain change between no cap and a cap across Re is much more, perhaps a factor of 10.

So basically the presence of feedback around the transistor reduces the variation in the overall gain when the transistor gain is varied.

QuoteWould it improve the signal/noise performance if i change the 100K value ?
If you decrease the 100k it will load down the input more.  So if you have a guitar plugged straight in you might lose some highs.   If you did try to lower it you will have to tweak the 330k to make the gain like it was before, then since you tweaked that you will have to fix-up the biasing.   There's other follow-on issues, in order to get the same performance the depth pot might need to be lowered, Q4 would need to be driven harder.   You end-up getting caught-up in a redesign.  The improvement might be small.

Having said that if you put 47k in there and you are happy all the technical aspects go out the window  ;D.   (I suspect it won't work like that as a 47k will increase the gain.)

Another angle is to optimize the transistor current.    For example I changed 18k to 100k and 4k7 to 33k and it was possible to reduce the noise by 1 to 2dB.  Quite worthwhile.   
[Note:  in order to get the same overall gain as 18k and 4k7 the 33k needs to be tapped with a cap to ground:18k + 15k to ground then cap to ground on 15k.  There's other ways to do this but that's what I did.]

QuoteI was surprised to have a perfectly clean and silent pedal on one system, getting noisy on the amp.
There may be more going on.  For example if the one system has more high-cut in the 3kHz to 10kHz region then that can reduce the perceived noise but only have a small effect on the brightness of the guitar signal.


EDIT:

What transistors did you use in your build? 2N5088?

With 2N5088's the noise improvement is not a much only about 1dB when the gain is cranked.

Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

duck_arse



excellent use of different knobs there, many would not have thought of that.
It won't work, Wayne.

snk

Thank you :)
I like to play with shapes, sizes, colors... and the overall ergonomy.
The RATE knob is the biggest (as it is the first one you reach for), and VOLUME is the smallest (as it's a set and forget kind).

snk

QuoteYes, it's due to bias shift.  [...]You need to wire pot pin 1 to gnd, pot pin 3 to the emitter and the wiper (pot pin 2) to a large cap, then the cap goes to ground.   The cap needs to be large like 47uF.  You might want to put a 470 ohm resistor in series with the cap so the gain doesn't go nuts at the end.
Ok, I tried this, and it doesn't work...

Well, I must say that i am on the verge of putting that project on hold and focus on other projects, because i was hoping it would be a fast and easy build, sounding good and easy to tweak, but at the moment it is much noisier on my amp than my Danelectro Cool Cat, so it kind of defeat the purpose  :-\
I'm glad having build my first tremolo, I like the rate LED indicator, I also like the tweaks I made to the speed pot, as well as the enclosure design...
So, all in all, it was a fun project, but as a musical instrument, with a scratchy pot and noise on my amp (while being dead silent on my mixer, i checked again), the goal is not reached yet :)
I think I will now focus on the 4 other projects I have on my workbench, and come back to it in a couple weeks...

Rob Strand

QuoteOk, I tried this, and it doesn't work...
I suspect there are some issues.  The 4k7 is inside the feedback loop.  As you increase the gain pot the amount of gain levels off (in fact it levels off to the resistor ratio 330k/100k).  Also the tremolo effect is reduced.   You really need to adjust to 330k (You can't just adjust the 330k as it stuffs the DC bias.  You need scheme with a cap and a pot.  The best way depends on how much gain variation up and down you want.)      I think your first method must have been stuffing the bias-up which ended up changing the gain.

QuoteWell, I must say that i am on the verge of putting that project on hold and focus on other projects,
..
is much noisier on my amp than my Danelectro Cool Cat, so it kind of defeat the purpose  :-\
I don't see a simple way to get rid of the noise.   The simplicity of the circuit promotes it being noisy.   A while back I tried to improve the coloursound inductorless wha.   It also has some issues.  There's better options so I stopped playing with it.  So yes, if it's not going to do the job there's  no point continuing with it.  It is what it is!
Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

snk

QuoteThere's better options so I stopped playing with it.  So yes, if it's not going to do the job there's  no point continuing with it.  It is what it is!
Yes, that's it.
All in all, it's a nice tremolo, but it will not replace the one i already have. Sometimes, you need to take things for what they are, and you can not change a horse into a mouse :)
I also think that it will find some use on another system than my regular amp, as it is noisy using the amp, and very silent with my mixer.

In any case, thank you for your help !