[R-390] R390 RF Cap Confessions

[Roger Ruszkowski]

Cecil,
 
You ask
___________________
The trick is what are better values for those since we are now replacing them 
during our restoration.
------------------------------
I do not have a pat answer.
It is an engineering problem.
I say with a straight face lets ask Tisha.
 
Power supply filter caps can never be to big.
But you need to watch the inrush current at power on.
You can suck to much current and after power on the inline
current limiting resistors just waste usable power as heat so there is 
a curve some where and likely a nice mathematical formula that 
establishes a line that should not be crossed as you get bigger in filter caps.
 
Bypass caps, I do not know why but higher
working voltage looks to do a better job of filtering with less random
popping noise. More capacitance does a better job of filtering.
 
Coupling / DC blocking caps can get higher in value but you open up the
band width as noise along the way.
 
Some times you are looking for an impedance match so you can not 
just change the cap value as you upset the filter characteristic or power transfer.
 
There is an overall noise floor. As long as the stage you are in is not the critical
stage that sets the signal to noise ratio you can allow more noise through a coupling cap.
 
But you start changing stage gain and that may not be adjustable.
AGC starts to not track changes in signal level. Strong signals get really reduced
more than needed so signal fade is not well regulated by the AGC as expected.
 
But as those by pass caps off a B+ line after a plate resistor go up in value you
expect more of the signal from the stage that makes it across the resistor to get by passed.
The filter cap stores a bit more charge and you get a bit stiffer B+ for the stage.
Small changes to be sure. Screen grid and cathode bypass caps like wise may help
the stage if there values are increased. 
 
First I see higher voltage rated caps as being less noisy and then larger value caps as being
less noisy.  I see noise in two forms. The high speed low level hiss and the low speed pop spike.
I think the larger value and higher voltage caps have less low speed pop and because we are holding the
cap at a larger charge value there is less noise.
 
Consider pouring a cup of water in a 1/2 full one gallon can and a 1/2 full one quart can.
You can pour the water into either can.
But you just get less splashing as you pour that one cup into the two quarts as when you
pour the same one cup into the pint in the one quart can.
 
I just see less splashing / noise in the analogy.
 
We know there is a whole set of caps that when installed in the Audio deck of the R390A do wonders to 
improve the audio band width and thus fidelity of the audio. OK for AM and lost on the ditty effort. 
 
I think we are stuck with it is a cut and try approach.
You have to work from back end to front end as the back end may mask stuff up front.
And when you get to the back end it may all suddenly blow up as changes get unmasked.
 
I think you have to have a goal in mind to guide you.
What is the expected end results.
I want more RF sensitivity.
I want a better signal to noise ratio,
I want more audio level.
I want more audio fidelity.
I want to damp the perceived ringing from the mechanical filters.
I want to reduce the power hum in the audio output.
I want to alter the AGC time constants.
I want to alter the AGC response.
 
With a goal you can then determine the usual suspects.
Then you can begin to question the  usual suspects.
You can apply change to a part and gauge the before and after behavior.
Keep the good changes and restore that which does not yield improvement in the direction you are trying to go.
 
Cap values are a good three credit graduate level class in electronic engineering.
 
Three pages later I add nothing useful to the conversation.
I have no good sound fast rules to tell you what will work.
 
I think the ARRL values used in their projects are reasonable. These projects are driven by good 
design practice and not limited by logistic procurement or parts counts.
 
Did I get three cents in here?
 
Roger AI4NI