Common collector amplifier frequency analysis

[truff]

Hi all,

I've started to learn stompoboxes. I've some little background on electronics from school and I try to understand some schematics to learn stompboxes. I want to analyse a simplified schematics for the Ibanez SD-9 (sonic distorsion): http://www.generalguitargadgets.com/diagrams/sonic_sc.gif

I'v found lots of documentation dealing with common collector amplifier (buffer) and I know how to calculate input and output impedance and gain of the amplifier. Now I would like to understand how does C2 and C4 are filtering the signal (I suppose that there's some low or high pass filters out there. I've seen other examples examplaining Thévenin theroems. I suppose that I need to add a serial resistor (something like 50 kohm to represent the guitar) and then try to calculate the equivalent resistance on C2 and C4 but I'm a little bit lost.
It would be very nice if someone can point me to a document or book or forum thread that can explain that.
thanks in advance.

[GibsonGM]

Well, looks to me like R4 is just a current-limiting resistor from the biasing network, and probably doesn't play a big role in filtering (altho probably plays SOME role, it may be small enough to ignore here). 

What you've got is a R3 and C2,  1K resistor and .047u cap right there, working into the input impedance of the transistor.  The 1K acts like a 'limiter', controlling how the filtering works.  This is giving a HPF effect; I am picking up a rolloff around 100Hz, but it's not that great.  Just a little pre-shaping before the signal hits the circuitry....

C4, the output cap,  works with R8 and the transistor's output impedance as a HPF, again tailoring circuit response for the next stage....to AC, V+ and Ground are the same thing....in this case, R8 is 100k, low enough that it IS having more effect on the signal than R4.   I get a cutoff on the output side of about 33Hz, enough to block noise but not doing much to limit bass response. 

I'm probably slightly off on these calculations, but this is what I am seeing.  Close enough for horseshoes )   Read up on filter response...use Google, search "high pass filter" on wikipedia...you'll find tons of info on this stuff.    Play with simulating in LT Spice!!  Invaluable.

[PRR]

Ignore Q1 for a moment.

You say you have a 50K guitar, which makes the 2Meg and 1K moot.

Draw the 50K, 0.047u, and 510K.

Well, the 50K is at-most a 10% shift in the action, ignore that too.

0.047uFd feeding 510K is 6.6Hz. Anything over 13Hz is pretty nearly full-strength.

Put back Q1. What is the input impedance of an emitter follower? This can be a tricky question. But roughly it is the emitter load (10K) times the transistor gain. My old prof said "Assume Beta >50". 10K times 51 is 510K input impedance. This in parallel with R4 510K gives half the impedance, so double the cutoff frrequency, say 26Hz. Or if ">50" really means 500 (a few transistors run that high) then input is 10K times 500 or 5,000K. In parallel with 510K is very close to 510K, maybe 450K, and cutoff is close to 13Hz, say 15Hz.

The "emitter load" should include R8. But 100K across 10K is pretty near 10K, skip it. The loose Beta of Q1 causes much more uncertainty of the true cut-off. But the range of possible cut-offs 13Hz or 30Hz lay far below the guitar band. We only need to know "lower than 80Hz", and we got that.

C4 R8 may be grokked the same way. We want the input impedance of IC1, but any decent 17-cent IC opamp has "infinite" input impedance at low audio frequencies. So it is just 0.047uFd against 100K, 34Hz at -3dB or -1dB at 68Hz.

Another designer might not use Q1 at all. The opamp input network could be scaled to 510K impedance, still have the dominant 34Hz pole with a (cheaper) 0.01uFd cap.

The high-frequency response of Q1 is "infinite for audio purposes". Above fT/Beta (about 1MHz) the input impedance declines from that 500K++ value it had for low frequencies. Even with 50K source it should pass nearly 10MHz.

[truff]

thanks a lot to both of you.

I'm understanding now how this works. While browsing the forum I noticed this reply to a post:
http://www.diystompboxes.com/smfforum/index.php?topic=84149.msg700692#msg700692

This sentence and the associated schema made me understand.

Remember electrons only run in loops. Draw the WHOLE loop.

Then I tried to represent the filter without Q1 for simplicity:



I understand that 2M2 is a lot compared to other resistors and that current won't go to this path so much so we can forget about this 2M2. So we are left with 1K, 510 K and Rs (the guitar impedance) in serial.
1K is very low compared to RS and 510 K so we can forget it also.

We are left with an equivalent resistance of about 560 K which gives us a cut off frequency of about 6 Hz.

now if we put back Q1, using the emitter resistance which is near Beta * 10K we get 2000 K (My transistir's beta seems to be 200). This in paralell with 510 K gives us 406 K and a cutoff frequency of 8.3 hz.

Basically what I understand is that these resistors and capacitor are not here to filter. We just need to take care that cutoff frequency don't go up to 80 Hz.

2M2 resistor is here to let C2 discharge to ground when switching off the DPDT (avoiding some unwanted noise)
C2 is here to make sure the biasing DC voltage don't go up to the guitar
510 K is here to bias the transistor
100 K is here to limit the current that goes into the transistor. I d'ont really understand this one, why do we limit the current before the transistor knowing that the transistor will amplify it ?


I understand that the opamp could make the job of adapting impedance so we may be able to remove the transistor buffer and keep the same input impedance but I can see lots of input buiffers (transistors or opamps) in front of others opamps in lots of schematics, so this must have something useful.

This also makes me wonder why on some schematics we have output buffers. We know that each electronic guitar device (amp or effects) has an adaptative input impedance that allow for the direct connetion of a guitar so I don't really understand why we should adapt our effetc's output impedance. In my case there's no output buffer and I find this normal. Maybe this is related to plugin headphones (or mix tables) directly to the output of the effect ? 


Sorry for asking such noob things but I'm noob and I wan't to understand