best layout practice: coupling caps

[ric]

The way I look at it is before the cap it will be part of the plate circuit and after the cap it will be part of the grid circuit. I feel the grid circuit is a lot more sensitive to interference so minimizing that possible antenna by shortening the grid path is best approach. But from what others have said it could be splitting hairs too.....it works for me so I stick with it.

Greg

Greg, that's more or less how I was looking at it. Before cap = HT / After cap = signal.
Thanks for the reply.
I know there are differing opinions on everything. Steve C., and I'm sure others, feel a cap is a cap, only the value (and of course the rating) matter. At the risk of seeming self deluded or thinking I have "golden ears" I choose different cap types because of their different sound qualities.
So it's no surprise some feel this is an exercise in futility. I have to put em somewhere, so why not the best place.

Enzo, the voice of reason.
You practical son of a gun you. Ok, I will carefully try your experiment. Hopefully I won't end up hanging my head in shame, that I couldn't tell the difference. But, hey, they gotta go somewhere, so....

[Enzo]

Whenever I hear the old debate about caps start up - inside foil, outside foil, which way is up, etc - someone always says the cap somehow shields itself. And I am always left wondering, what does it shield from what? Not saying they are wrong, but all I ever see are rationalizations. All the signal, noise, and what have you comes from that plate and continues through the cap to the grid. After all, the job of a cap is to pass signal.

SO a real live test might answer the questions.

[octal]

Agree with Enzo on this for coupling cap applications... both ends of the cap are at the same AC signal level except below the "corner frequency" of the cap, which usually falls below the rolloff frequency of a music instrument speaker anyway.

In a bypass cap scenario, having the outside foil go to ground seems to make sense, though it probably doesn't really matter much most of the time.
As a slight tangent, what era did they stop labeling the "outside foil" end of caps anyway?

[Enzo]

SOme still do. It is just a stripe near one end, they don't usually print the words.

[ric]

Thanks for the input octal.
My original thought was should the cap go as close as possible to the following stage like a grid stop resistor. The general consensus seemed to be yes.

Enzo felt Fender's lack of issues with this shows me worried about nothing, and how about an experiment to see rather than talking it to death. Then we sort of skidded off into outside foil.

I tried the outside foil this way and that test some years ago and I couldn't hear any sonic advantages or lower noise floor so I just put all the print the same way so it looks tidy.

I only build for myself so I don't have to worry about neon beer sign hum at the bar etc. So the foil thing for me is decided. But I wondered about this "upstream" vs. "downstream" thing as far as best practice.

[Gingertube]

As a slight tangent, what era did they stop labeling the "outside foil" end of caps anyway?

When caps stopped having an outside foil, those that do are still marked.

The outside foil end should connect to the point with the lowest impedance to 0V. This will vary with the circuit, for a simple RC coupled stage then that will be the anode of the driving stage which is probably the opposite of what people might intuitively choose.

Cheers,
Ian

[soundmasterg]

I think the biggest reason for placing the cap close to the grid on the circuit following it is because anything attached to a grid acts pretty much as an antenna, so if you have a longer wire vs a shorter wire going to the cap, then it could contribute to a higher risk of oscillations.

Greg

[ric]

Ian, I did ask about best practice, so I guess outside foil is fair game. I had come to understand outside foil to previous anode was best. I noticed no difference so I abandoned the practice. You state that the caps that still have outside foil are still marked. It may be that since I build with orange drops they have no outside foil? That would explain not hearing a difference. If you read this and feel like it, would you elaborate on the situations where outside foil should not face the previous stage's plate?

Greg, thanks for the most recent reply. Mick and you both seem to have experienced times where moving the cap downstream helped stability. I take that as strong evidence of the desirability of that location.

[raiken]

Coupling capacitors can have a "good side" and a "bad side", as far as placement goes, because of circuit impedance. If you think about it for a minute, you'll come to the realization that the capacitive reactance, given by Xc = 1/(2*PI*f*C), goes up as the frequency decreases (because of the "f" term in the denominator). This means that the capacitor looks like an ever increasing resistor as the frequency goes down (converging to an infinite resistance, or open-circuit, at DC). Because of this, the circuit node impedance rises as the frequency decreases.

At the plate of the previous stage, the node impedance will be equal to the value of the plate load resistor in parallel with the impedance seen looking into the tube plate, which means it can never be higher than the value of the plate load resistor. Since the typical commonly used 12ax7 has an internal plate resistance of 62.5K, the effective output impedance at the plate node will be 62.5K in parallel with 100K, or around 38K.

Now, since we are starting with a relatively low 38K output impedance, the circuit should be relatively immune to noise pickup (not near as good as a true low impedance, like a few hundred ohms, but not as bad as a typical 1Megohm grid circuit). However, here is where things get bad - if you consider, for example, the 0.0022uF coupling capacitor used in the bright channel of a vintage Marshall, it will look like a 7.2k ohm resistor at 10kHz, a 72k resistor at 1kHz, a 720k resistor at 100Hz, and a whopping 1.2Megohm resistor at 60Hz!

This effective resistance will be in series with the output impedance of the previous stage, so our relatively low 38k impedance is now over 1Meg at 60Hz! This high impedance can easily pick up hum from filaments or fluorescent lights (if the chassis is not shielded). It gets even worse if you tack a long wire to the end of this high impedance, because the entire trace is now a high impedance node at lower frequencies. If, however, you put the coupling capacitor at the grid side, you now have a relatively low, relatively constant, 38k node impedance at all frequencies for the entire trace length up to the capacitor, and the only sensitive node is right at the grid where the coupling capacitor connects. Note that the grid-to-ground resistor of the following stage is effectively in parallel with this series output resistance of the previous stage, so a lower value of grid-to-ground resistor will help lower the node impedance. Also note that a grid stopper is effectively in series with it, so it is even more critical that they be placed right on the pin, to avoid an even higher trace impedance.

As far as the outside foil is concerned, the same thing happens - the outside foil and the inside foil are separated by the reactance of the capacitor, so one side will be much more susceptible to picking up noise and crosstalk from other signals inside the amp. This can easily be seen by holding connecting a capacitor to an oscilloscope (or to a cable going to your amp input jack) and laying an AC cord next to it (or just grabbing it with your fingers). The scope amplitude (or hum) will be much larger in one orientation than the other, provided the capacitor value is not too large to swamp the noise. Whether this is noticeable or not in a circuit depends on the location of the capacitor in relation to other signal or noise sources and it's value in relation to the circuit impedance. Some capacitors, including some types of Orange Drops, are made such that they have no outside foil.

Randall Aiken

[ric]

Thanks, Randall.

When you said: "if you think about it for a minute", and laid out that formula, I thought it'll take more than a minute. I'm pretty challenged in that area, but I get it. And the natural progression of lower and lower frequency to DC, where the cap blocks it. I knew it did, now I have some understanding of the math. That's a new one for me. Thanks for an explanation I could follow.

It's looking like my intuition on this is shown true by the physics: coupling caps as far downstream as possible is best practice. Thanks also on the outside foil primer. My not noticing any difference with orientation with the orange drops was because they had no outside foil. So it is important, or at least best practice to put outside foil to the lowest impedance, if the cap has an outside foil.

[raiken]

Some types of Orange Drops are wound with an outside foil, some are not (716P series), so you may have had either type.