Signal and bypass caps

[dennyg]

A fewquestions regarding caps in the signal chain (coupling, bypass):
* From a layout perspective, are there any considerations for caps in proximity to other components - i.e. are they 'generators' or 'recipients' of detrimental waves/parasitics?  e.g. coupling cap next to signal bypass cap which are carrying out of phase signals.
* Should smaller (both voltage rating and size) caps be used for bypass caps where there is little to no DC (e.g. cathode or voltage divider feeding a grid?).  Or does it make any difference (e.g. 650v cap vs 50v cap)?
* I've seen many builds, particularly high-gain, where the tone stack caps are mounted directly to the pots.  I would infer that the builder is trying to move those caps as far from other components as possible to minimize 'transmittal' of the highest gain pre-amp signals back upstream?

Any materials/threads on the topic? 
thanks!

[silverfox]

i.e. are they 'generators' or 'recipients' of detrimental waves/parasitics?

For the most part, no. There used to be a heavy black line marking the end of a capacitor connected to the outer foil so the cap connection could be oriented towards the least positive side of the connection for the purpose of acting as a shield. However, this makes no difference. I used an oscilloscope to determine if the noise level was greater on either side of the capacitor and there was virtually no difference, using a common brand of cap. I always refer back the the days gone by when, (ships were made of wood and men were made of steel), ah, ooops, wrong analogy- Days of old when circuits were hand wired point to point and the noise levels were acceptably low. See the post on point to point wiring for some excellent images. I believe, and I could be wrong on this: Caps lack the inductive capacity, (at the low to shortwave band) to develop sufficient signal voltage for the purpose of retransmitting an incidental signal.

* Should smaller (both voltage rating and size) caps be used for bypass caps where there is little to no DC (e.g. cathode or voltage divider feeding a grid?).  Or does it make any difference (e.g. 650v cap vs 50v cap)?

You want to make sure the voltage rating of the capacitor is matched to the DC voltage that will be present . However at times it may be convenient to use a larger voltage rating cap as you may not have the smaller and more appropriate size capacitor on hand. Just don't use a 50V cap where the anode voltage is 150 VDC. At times the AC rating of the capacitor will also be a factor but for the most part, standard brand caps at rated voltages suited to the application will work fine.

 I would infer that the builder is trying to move those caps as far from other components as possible to minimize 'transmittal' of the highest gain pre-amp signals back upstream?

More likely they are moving the caps to the tone pots to reduce the number of components on the PC or TAG board. The caps don't have to be moved far from other parts of the signal path.

Keep in mind, crossing the paths of input and output circuits can very easily cause a problem with feedback. However, those areas of the amp are usually naturally separated to opposite ends of the build.

Regards,

silverfox.

[Willabe]

I would infer that the builder is trying to move those caps as far from other components as possible to minimize 'transmittal' of the highest gain pre-amp signals back upstream?

More likely they are moving the caps to the tone pots to reduce the number of components on the PC or TAG board. The caps don't have to be moved far from other parts of the signal path.

I agree with silverfox but it's also to shorten the length of the grid wire.

Grid wires act like antennas and pick up unwanted signals/noise/hash from the air. That's why we use shielded wire on longer grid wire runs.

After the plate signal goes through the coupling cap it is now a grid wire. So given a choice keep the plate wire long and the grid wire short.

Same with the caps in a tone stack (TS), after those caps, it's all part of the grid wire. With the TS caps mounted on the eyelet/turret board you have wires going back and forth to the caps and TS pots. Mounting those caps (and any R's in the TS) on the pots gets rid of those wire runs. Depending on the amp and it's layout you can eliminate a foot to 2 feet, maybe a little more of grid wire. Since those wire runs are not just 1 length of wire but several shorter wires, it would be tiresome and/or impractical to us shielded wire for them.   

I also try to mount the coupling cap right on the tone pot (and volume pot) lug with as short a cap lead as reasonable. (Just like with a grid stopper R.) Saves me from using a piece of shielded wire.   

[dennyg]

After the plate signal goes through the coupling cap it is now a grid wire. So given a choice keep the plate wire long and the grid wire short.

That right there is a gold nugget of wisdom.  Been at this hobby for two years now and seems I would've stumbled on that fundamental somewhere but no time like the present!!

[HotBluePlates]

* From a layout perspective, are there any considerations for caps in proximity to other components - i.e. are they 'generators' or 'recipients' of detrimental waves/parasitics?  ...

An electrolytic cap usually has the negative lead connected to the case.  A big electrolytic on a board might be seen as a shield, or grounded point, between otherwise adjacent circuits.

Non-polar caps (mostly coupling caps) will have to be checked to determine which leg is the outer foil (if you want to use orientation of the outer foil to reduce noise pickup).

... * Should smaller (both voltage rating and size) caps be used for bypass caps where there is little to no DC (e.g. cathode or voltage divider feeding a grid?).  Or does it make any difference (e.g. 650v cap vs 50v cap)? ...

You can use a 20µF 600v to bypass a cathode.  It might "only" cost you $4-5 instead of the $0.81 you could have paid to get exactly the same performance while taking up less board space.

As for caps in other locations, the required voltage range depends on the specific application and the worst-case possible voltage which could be present (usually when B+ is applied but no tubes are drawing current).