Capacitor question - blocking DC

[lukeferg]

Hi guys,

I've been puzzling over this one for the last couple of days and can't find anything to explain it. I've got a tubescreamer that's giving me the shits and the more I'm looking at the schematic the more things don't add up for me.

Question: if caps block DC then why do we bother having Vref going into the drive part of a tubescreamer when it hits a cap straight away?

Here's a schematic for reference. I'm talking about the Vr before it hits C2
http://1.bp.blogspot.com/_y4AYtND8Hz8/R8k87o9XJ4I/AAAAAAAAAIY/VvY0B9P_g5s/s1600/soatebsqt4.jpg

No DC should pass that point, correct? The only solution I kind of figure is that Vref is substituting as a kind of "ground" point for the IC but I doubt I'm right about that.

[antonis]

You're quite right..!! 

C2 passes AC and also "isolates" any DC coming from IC's output, so with the help of C1 we have a fully AC coupled amp...
(actually, it restricts any DC amplification..) 

For AC signals, any DC power supply should be considered as GND..
(PS internal resistance is practically negligible so + to - are short-circuited for AC signals...)

There are some arguments about straight grounding vs via Vref point but, in most cases, it's a matter of PCB populating convenience..

[GibsonGM]

So, that point is a ground for AC which still blocks DC....

[lukeferg]

Ahh, so that's also why it's half of supply? Because AC needs the voltage swing both ways.

It seems I'm learning stuff.

[antonis]

Ahh, so that's also why it's half of supply? Because AC needs the voltage swing both ways.

Ahh, yes..!!

But that stands only for IC inputs - left leg of C2 can be connected to ANY "pure" DC point..
(in a same manner, some of the above circuit GND's can be replaced by Vr..)

It seems I'm learning stuff.

Is that so..?? 

Take your time and tell us which of Vr & GND points aren't interchangable..
(of course, you'll have to justify your answer 'cause there are some inpatient for slapping hands guys around here..) 

[lukeferg]

Alright, challenge accepted.

I'm sure that the input and output jacks aren't interchangeable. Otherwise you'd be introducing DC voltage externally from your circuit. You also wouldn't put Vref on the ground from the DC input. That wouldn't make any sense at all considering DC ground is the 0v reference point.  But they're the easy ones.

Another one that isn't interchangeable is the volume pot. Once again, there's no cap blocking DC before it goes elsewhere. And I think that would make the volume pot crackle when turned too but I'm not sure about that bit.

I think that the Vr at R1 needs to be Vr because pin 3 on the IC needs DC current. The other half of the IC will get DC on pin 3 from the output of the first half.

I think the tone stack needs to be the proper ground. Mainly because I can't recall a tone stack that grounds to a Vref point, but also because if it's dumping signal to ground then if it's not really 'gone' then I can't see that making sense.

The one that I'm not sure about is the ground near C4. I want to say it's interchangeable but only because I've seen similar setups using Vref in other pedals.

How'd I go?

[antonis]

 

First paragraph: OK.. !! 

Second paragraph: OK..!! 

Third paragraph: OK..!! 
(actually, non-inverting input needs a VOLTAGE to refer on - it could be GND in case of dual symmetrical supply or (V_cc+V_ee)/2 in general..)

Fourth paragraph:   
Tone control keeps some (un)balance between IC1b inputs which IC amplifies the section between right & middle pot lugs and C5, R6..
The signal needs a way to GND through the above components (they create a High Pass filter) to be amplified - or else we'll have a simple follower..
As long as there is a cap in that path (like C2 on the first IC loop) it can go either to GND or Vref..
(actually, it could go to Vref despite of cap presence.. - i.e. DC amp, intergrator, only low pass filter in the closed loop & generally in cases that we don't mind to "roll-off" DC gain to unity..)

Fifth paragraph: Hmmmm....  (OK but w/o documentation..)
We've said a DC power source should be considered as GND for AC signals..
To get a Vref voltage from a Voltage source we need (at least) a resistive voltage divider so our signal has to pass through  upper resistor to "fake" grounded and through lower resistor to be "really" grounded - that only stands in case of Vref regulating capacitor absence..
In case of (usually big) capacitor presence, Vref should be straight considered as GND..
(same for the case of Vref created from an op-amp..)

Now, if C4 is connected to Vref it should be considered in series with C8 (resistors R8 & R9 are of high values compared to C8 impedance at signal frequencies and can be safely omitted..) but it's capacitance is much bigger than C4 so it can be ommited too..
E voila..!! We've an effectivelly grounded C4 of 143nF capacitance, for which we don't mind at all..!!

P.S.
Why we don't mind about the 4.5% reduction of C4 nominal value..??

[lukeferg]

So just to clarify the dual supply thing. If I had a pedal that had a 4.5v and a seperate -4.5v supply then the reference on the IC would need to be 0v. I was thinking that before but wasn't sure.

The tone control cap isn't making a lot of sense to me though. The tone control is a high pass filter and dumps the unwanted signal to ground right. If it's dumped to Vref won't that signal be reintroduced to the circuit. Although maybe that doesn't matter because that signal is going to be filtered out again anyway and it's not like we have a Vref point after the tone control.

C4 is part of a low pass filter. By reducing it to 143n we raise the knee point to 1.1kHz right? So it would retain a slight bit more high end after clipping. Is this the reason tubescreamers have very forward mids? To me, 1.1kHz seems a bit low but I guess it works since people much smarter than me have done it for years.

Thanks for the info by the way. It's quite interesting stuff. A lot of the technical details still go over my head but I'm getting there.

[GibsonGM]

Hi Luke,

Well, you are close..You don't need a +4.5V and separate  -4.5V supply.....you'd want what's called a DUAL POLARITY supply.   Ex: if you connect two 9V batteries in series, so that you have  "  +  -  +  -"....you would take +9v from the left end, and - 9V  from the right end.  Where - and + meet in the middle - that node is 0V, ground.  The potential to either side of this point is equal.    You would then be using 0V, ground, as your reference!!  And all would amplify just fine.    +9V would go to + power pin (Vcc), and -9V to 'gnd' or negative pin (Vss).    The potential between those power leads would be the summed battery voltage, 18V.....

But we want things to be more convenient, and to use just ONE 9V battery! So.....
The 4.5V, in our 9V system, is to push the steady-state condition ("quiescent state") of the opamp UP...so that the '0' point of your incoming signal sits at 4.5V...allowing the negative part of the signal some room to go down around this point.  We set our bias point at 1/2 the supply voltage.   Does that make sense?   If you didn't do this, you could not amplify the negative-going portions of your signal with a SINGLE polarity supply....the reason for 1/2Vcc (1/2 supply voltage) is to give equal room for up and down swing of the signal.  We're adding a DC bias to our AC signal....the caps in the circuit remove this DC (output cap, especially).

Tone control cap:  Ok - All this ground hoo haw....that was so the chip could operate. Could be a transistor, FET or whatever, too.   They run on DC....they turn that DC into useable things, like a bigger AC output voltage than what we put in.

Our SIGNAL, the noise we care about, is AC.  To AC, and AC only...EITHER SIDE of the power supply looks like ground!!    The cap is blocking DC, so all that can pass thru it is AC.   It doesn't matter to the AC signal there if it goes back to +, - or gnd.    It is kind of 'conventional' to return it to ground or Vref.   And I suppose if your cap was a little leaky, you'd like it to return to ground or you'd have a short!   That's my guess why they return it to Vref or ground.  When the frequencies we're "dumping" get to ground, we can think that they stay there, for all practical purposes.

Now, in reality - what is separating plus and minus in the power supply?  A 47u cap, right?   
AC can pass right thru it     It doesn't care.  Think on that a little bit, it is sort of strange at first.

Another reason we run the chips on DC, LOL.....

[lukeferg]

Of course! By using the voltage divider with resistors we are faking a dual supply. I always thought two batteries in series double the voltage but in reality they are doubling the voltage difference. That actually makes a lot of sense.

The tone cap is starting to get there with me, maybe I'm a bit slow. So once it hits DC the signal is 'grounded' and the AC signal disappears so to speak. And the leaky cap thing is probably why I don't see tone caps going to Vref much.

[GibsonGM]

Of course, 2 batteries in series WOULD double the voltage if we did not designate the point between them as ground   
When we DO say that point is gnd, we FORCE them to have opposite potentials...just another way to use something familiar for a different purpose.

Just putting them in series gets you  +18V, as your instinct tells you.  Only - one end is ground (usually minus), and one is the 'power supplying lead', we call "plus".

You are very close, regarding the tone cap and AC ground.  We think of DC for operating devices like opamps.  And our signal is AC.   Once we give the opamp its power, and its ground reference (GND, pin 4...), and have figured out if we're adding a DC bias....we're done with DC.   Like putting a battery in a car, we have done it, mechanically, and it's all set ("configured"), it will operate as we planned...it's not very artful but useful - this is how we tune things up so cool things happen.

Where we send something to AC ground is really a matter of what is practical, and what makes sense.  It's not something much thought over, I think...DC gnd is fine, and so is Vref.   V+ would be cool, but as I mentioned...some caps leak, and that would make noise and eventually a short, perhaps...bad form.  I'm not SURE why they don't do this, but they COULD.  Doesn't matter; you probably should not, either.  I'm sure others could explain the physics better.

The whole deal is, as I see it, a way to tie it all together in a nice, neat way that is consistent and makes sense.   The AC NEEDS a route to get out of the signal path.  That's all we're giving it. 

I think you 'get it'...you are seeing how AC is mixed with DC to accomplish some function, then the DC (most often) is taken out, yielding an altered AC signal.  Sometimes a portion of the AC is bled off to (a) ground....you got it     

[antonis]

Hmmmmm... 

Nobody noticed my faulse calculation about C4 & C8 equivalent capacitance..
(which actually is 149.5nF..)

Sir Mike is notorious for his lazyness but you Luke shouldn't take lightly everything is told by guys who type in a hurry... 

[lukeferg]

Whoops, I must admit that I didn't bother to do that calculation. Considering I have to look up ohms law every time I use it I tend to blindly follow those that came before. A worrying habit I can see now.
So was I right at least about the low pass filter thing with C4? I'm pretty sure that's right but my ego needs confirmation.

[antonis]

So was I right at least about the low pass filter thing with C4?

Of course you were right...!!! 

But it just happens to be there, because R4's main duty is to limit the current of IC1a output, so designer found convenient to place a small cap and create a (simple for production manager - cheap for financial director) low pass filter.
(both of guys in brackets need more than you ego confirmation..) 

[GibsonGM]

Hmmmmm... 

Nobody noticed my faulse calculation about C4 & C8 equivalent capacitance..
(which actually is 149.5nF..)

Sir Mike is notorious for his lazyness but you Luke shouldn't take lightly everything is told by guys who type in a hurry... 

I AM a little lazy, ha ha!  Sometimes I do not even measure a capacitor like C4 - I just put one there and see what it sounds like.  I know by the size of the cap if I am in the right range!!    Even then, it is only good for MY guitar, maybe not YOURS!

I would guess that R4/C4 will take away some harshness produced by the clipping diodes? 

[antonis]

So sizes DOES matter...!!
(I pick my caps by colour which also is a matter of taste..) 


< I would guess that R4/C4 will take away some harshness produced by the clipping diodes? >

Maybe, although a cap in the feedback loop could do it better..
(it would also form a (variable according to Drive pot setting) lowpass filter while conveniently lying in pF range would be proper for 500k pot and for "mellowing" clipping corners..)

IMHO, clippers inside NFB loop don't need rounding of waveform edges - but it's also a matter of taste, of course...

[PRR]

> The tone control cap isn't making a lot of sense to me though. The tone control is a high pass filter and dumps the unwanted signal to ground right. If it's dumped to Vref won't that signal be reintroduced to the circuit.

Tone cap is 220nFd (or 150nFd?).

Vref is bypassed with 47uFd.

Which is 47,000nFd.

As Antonius has said, the "dumped signal" is reduced 214 times smaller.

YES reckless dumping into a Vref WILL make trouble.

We may wonder here, because Vref also feeds the first HIGH-gain stage. But IC1a gain is "only" 157, less than the 214 cut from tome to Vref. Also the sneakage through IC1a's 1Meg will be significantly absorbed by whatever is plugged-in that jack (ah, but what if the plug falls out? Go-quiet or go-squirrely?).

In this case, because the tone control is DC-blocked by its own capacitor, there is no need to return it to Vref and its dubious solidness, return to ground. In fact arguably C2 should also go to hard ground. Then only R1 takes Vref, and Vref clamping could be a whole lot less. R8 R9 100K, C8 10uFd. (Hardly saves a penny.)
__________________

> R4's main duty is

No "current limiting" is needed. R4 and R5, with a cap run to one or the other, form a boost/cut amplifier, working only on treble.
__________________

> Nobody noticed my faulse calculation

No.

[lukeferg]

I think I'm going to need to read your post a few more times before I get all of it 

I can see how the signal is reduced enough to not cause issues except in unusual (or reckless) circumstances.

As far as R4 and R5 go. I get how R4 is making a treble cut circuit but the role of R5 is somewhat hazy. Am I right in thinking this is similar in principle to the stupidly wonderful tone control 2 on the AMZ site? We're keeping resistance constant so volume won't be affected. I thought R5 was more about the feedback loop to reduce gain from the other half of the IC. I'll keep reading stuff and it's bound to make sense eventually.

[PRR]

> R4 is making a treble cut circuit but the role of R5 is somewhat hazy

When looking at a pot, turn the knob one way, the other way, and to center. Work out what happens.

And remember that caps act as high impedance for low frequencies, low impedance for high frequencies.

And in this case, ignore C4 for a moment. It is significant. But it basically means the circuit "always" needs a top-cut so one is built-in. This makes sense because clippers throw a LOT of high frequency has, you don't really want ALL of that.

Here, turn R11 full left. R5 just forces the amp to unity gain. C5 and stopper R6 load-down R4. For low frequencies, high impedance, no load, all bass gets through. For high frequencies, the C5 R6 arm drop toward 330r, which is a heavy load on R4, so treble is cut to around 1/4.

Now turn R11 full right. Signal through R4 is not loaded going into the amp. C5 and stopper R6 load-down R5. For low frequencies, high impedance, no load, all bass gets through unity-gain. For high frequencies, the C5 R6 arm drop toward 330r, which is a heavy load on R5. Do you know how to estimate a "non-inverting opamp" stage? The gain for treble is R5/R6 (plus 1), about 4. Treble is boosted.

The mid-turn can be estimated as some-cut plus some-boost so essentially flat. The half-way resistance of each side of R11 makes the boost and cut about 2.5K/3.5K or 3.5K/2.5K, very small, little internal gain or added hiss.

> the other half of the IC

You mean IC1a and IC1b? The two halves "should" act just like two single opamps. When you and I split a 6-pack, I don't worry about your beer affecting my beer. The crosstalk should be so very low that no designer really has to fret about it. IC1a is a high-gain stage, IC1b is a low-gain buffer (with +/- variable treble gain).

[antonis]

> R4's main duty is
No "current limiting" is needed.

Obviously, you have much more confindence in unshorted caps than me, Paul...