>Thanks to all who finally came forward with some data to look at. The bad
>news is that the data is somewhat confusing. I am inherently suspicious of
>data that does not 'compute'. This is the result of a life-long experience
>in the electronics business, in which I have seen an enormous amount of
voodoo
>in mfrs. specs.
Amen to that.
>The reasons are numerous and variable, but suffice it to say
>that in many companies the documentation is prepared by people who have
>little or no connection with the engineers who did the design.
>
>I do not wish to step into yet another quagmire here. I would like to back
>up a little and try to return to the essential point that I was originally
>trying to address in this whole discussion of mica capacitors. I would ask
>all of you to re-consider it from this viewpoint.
>
>First of all, just exactly what does it mean for a capacitor to have a
>current rating?
It means when you ignore the current rating, the capacitor can destruct.
I ignored current ratings with the tetrode with handles amp. I
experienced 6 padding capacitor fires and a meltdown in a 200pF Series-57
doorknob cap (the silver solder melted and flowed out like hot lava).
> I can already see signs of opinions lurking out there that
>if you accidentally run 8.6A through your blocking cap rated at 8.5A, there
>will be a meltdown of your beautiful 4cX00000000 amplifier.
>
>My main focus, and it should be yours also, is on the Q, loss factor, power
>factor, dissipation factor, or whatever term you want to put on it - they
>are all the same thing. Dielectrics have losses due to molecular processes
that
>accompany the polarization that produces capacitance in the first place.
>
>Thisloss shows up, from a circuit analaysis view, as a loss resistance in
series
>or parallel with the 'pure' capacitance. When your HF plate or antenna
>current passes through this resistance, it will produce I-squared-R loss and
>heat up just like a resistor. Our main goal in all of this capacitor voodoo
>is to minimize this loss - plain and simple.
>
Capacitor fires are something to avoid. I've been there.
>The only way that the manufacturers could REALLY help us would be to publish
>some kind of a loss-factor vs. frequency data. From that we could calculate
>the amount of loss that we have to justify and/or dispose of in our thermal
>design.
>
>I have seen a trend among contemporary manufacturers, especially in Europe,
>to rate capacitors in KVAR, which is one step closer to what we want. But it
>still doesn't give us the whole picture, especially any frequency
>dependencies.
>
>This is getting longer than I planned. My overall point in this posting is
>that I would really like to hear from others what they are looking for (with
>regard to mica capacitor ratings.)
Mica caps can work well in 160m and 80m tank padding apps. However, for
DC blockers, ceramic doorknob - and vacuum capacitors - seem to be
preferred -- probably because of the substantial increase in current
through the DC blocking C at higher frequencies. For example, in
tetrodes-with-handles amps, at 29MHz, >30A-rms can flow back through the
DC blocking capacitor and the anode C to chassis-ground. [calculations
on request] 30A-rms at 29MHz is nothing to sneeze at since its heating
effect in conductors is like c. 120A-rms at 60Hz.
>What does it mean to you when some mfr.
>stamps 8.5A on the package?
It means that the capacitor can dissipate the heat caused by passing that
amount of current at that frequency. Typically, current is spec'd at
1MHz, 10MHz, and 30MHz.
>What kind of data would make you feel confident in your component selection
>process?
Current stated at several different frequencies - such as 1MHz, 10MHz,
and 30MHz.
>What would you do if you had an unknown capacitor that you really
>wanted to use but didn't know all of its specs?
>
Parallel the test cap with a roller coil to resonate it at the test freq.
Parallel the L/C combo with a 50-ohm termination. Connect an o'scope/HV
probe in parallel with the three in order to measure RF potential.
Increase P until the capacitor reaches a temp that is uncomfortable to
touch. Record V-peak, multiply by 0.707 to obtain E-rms. Calculate Xc
ohms at the test freq. Use I=E/R to calculate current.
cheers, Eric
- R. L. Measures, a.k.a. Rich..., 805.386.3734,AG6K,
www.vcnet.com/measures.
end
|