[Amps] Oldie but Goodie MLA2500 questions

[Chuck Counselman]

> Charlie carroll <k1xx at k1xx.com> wrote:
> ... the individual that did the conversion [of my amp] replaced the
> single 500pF plate bypass disc cap with a 1000pF doorknob.  I’m
> going to try to shoehorn in a couple of additional disc caps of
> different values.

"Plate bypass”??  Do you mean the cap that bypasses the DC plate supply at the “cold” end of the plate choke?  Or the cap that couples RF and blocks DC from the plate to the tuned “tank circuit” or output impedance-matching network (e.g., a pi- or a pi-L network)?

The former, RF-bypass cap should comprise several ceramic disc caps of various values, connected in parallel, in order to provide a low impedance to ground at every possible operating frequency.  These caps must have sufficient DC working voltages, but their RF characteristics are not critical because they do not carry very much RF current and the RF voltage across them is small .

The latter, RF-coupling cap must also have sufficient DC working voltage, but its RF characteristics ARE critical, because this cap carries high RF current and, if it does not have low loss at RF, it will self-destruct.  Using several caps in parallel here will help by dividing current among them, so that each one dissipates less power.  However, if the parallel-connected caps have a wide range of values, then inevitably one of them will carry more current than the others, because the nominal value of its capacitance is greater and/or because it is series-resonant at a frequency for which the others are not.

Using a ceramic disc cap for plate-coupling can be DANGEROUS because MOST ceramic disc caps have too much loss at RF.  Most ceramic disc caps are intended for bypassing and NOT for coupling high RF current.  The ceramic dielectric materials between their plates are made to have high dielectric permittivity (epsilon), in order to get high capacitance in a small package.  Most of these high-epsilon ceramics have low Q, in other words high loss.  The principal exception to this statement is that the ceramic in a disc cap labeled “NP0” has not only a near-zero temperature coefficient (as the code NP0 is intended to indicate) but also very high Q.  The temperature coefficient of a coupling cap is not important, but the Q is very important.

An NP0 ceramic disc cap, or several such caps in parallel, would be good for plate-coupling if it/they had sufficiently high DCWV and if it/they had sufficiently low reactance(s) at the relevant frequencies.

In most of the high-power HF amps I’ve seen, the plate-coupling cap is one, or several paralleled, ceramic doorknob(s) of the type made for transmitting.  It is important not to use the type made for HVDC power-supplies.  The ceramic dielectrics in the latter type have high epsilon and low Q.  Rather little RF current will quickly overheat and destroy the latter type.

It is important to look at the code painted/stamped on a ceramic doorknob to see whether the cap will handle high RF current.  The code “NP0” is best.  However, the code “N750" is very good.  The plate-coupling cap in my own high-power MF and HF amp, and the cap in my high-power, L-network, MF and HF, antenna tuner comprise parallel-connected ceramic doorknobs labeled “N750.”  I would be wary of ceramic doorknobs having much higher (positive or negative) temperature coefficients, e.g., N3300. 

Having to read temperature-coefficient codes to infer the Q of a cap seems unreasonable because you don’t care about the temperature-coefficient per se.  You care about the Q; but Q is not indicated directly in the labeling (although some transmitting cap’s are labeled with a maximum RF current rating at a stated frequency).  Sometimes one can find a manufacturer’s data sheet on the Web, for a manufacturer’s part or type number that appears on the cap; but, more often than not, you can’t, because the manufacturer has been out of business for decades.

Since you have a VNA, I suggest measuring the complex impedance of a ceramic doorknob cap before using it.  If you perform a careful calibration of your instrument, you will be able to distinguish a cap having Q = 1000 from a cap having Q = 2000 (or Q = 500).  A good plate-coupling cap has Q of at least 1000 at the frequencies of interest.  To test your calibration, you can measure a vacuum cap.  While you’re at it, measure a few garden-variety disc ceramic caps.  You will be shocked to see how low their Q’s are.

Some websites where you may find helpful info on the RF characteristics of caps are, in alphabetical order:

First, a few present-day manufacturers of high-voltage, high-quality, ceramic caps.  New caps are quite expensive.  I buy mostly surplus caps; and I measure them before using them.  I measure complex impedance with a carefully calibrated AIM 4170, and I test DCWV with a high-voltage DC power-supply.

<http://www.avx.com/>
<http://www.calramic.com/>
<http://www.calramic.com/Design/Assets/PDF_files/CRT-0023.pdf>
<http://www.calramic.com/Design/Assets/PDF_files/CRT-0006.pdf>
<http://www.kemet.com/>
<http://www.murata.com/>
<http://www.vishay.com/docs/28542/sseries.pdf>


G3YNH’s website is helpful regarding surplus caps (and inductors etc.).  See especially:

<http://www.g3ynh.info/zdocs/comps/part_4.html>
<http://www.g3ynh.info/zdocs/comps/part_5.html>
<http://www.g3ynh.info/zdocs/comps/part_6.html>
<http://www.g3ynh.info/zdocs/comps/refs.html>
<http://www.g3ynh.info/zdocs/z_matcing/>
<http://www.g3ynh.info/comps/vac_caps.html>


73 de Chuck W1HIS