PA Bypass Capacitors 4-65A

[W9ZSL]

  I'm confused as usual and it has nothing to do with Santa Claus.  I'm building a rig around an Eimac 4-65A and am relying on their basic schematic for a high-level modulated PA.  So far so good.  They specify .002 mica caps @ 500V for filament and bias bypass with a .001 @ 2500V mica for the screen bypass.  OK.  I've also seen other schematics for the same tube and nearly identical configurations such as the 75-300 watt transmitter in the '57 ARRL handbook.  It uses .001 disc ceramics for all the bypasses and to add further confusion I've seen .003 bypasses in a variety of 4-65A applications with no capacitor type or voltage listed.  In every case all plate, grid and screen voltages are similar.

The rig in the '57 handbook makes use of the wide variety of plate voltages that make the 4-65A happy and that's why I decided to go with that particular tube.  I have a lot of .001 @ 6KV disc ceramics on hand that I would like to use, so my question is how critical is the value of all the bypass capacitors and should I go with mica rather than ceramic?

I'll be running 1800 VDC on the plate @ 120 ma, 250 on the screen and -125 bias.

Merry Christmas!  

Mike
W9ZSL

[w4bfs]

ok ...its time to put the formula for capacitive reactance to work:   Xc = 1 / (2 pi f c)   and calculate these for the lowest rf frequency you intend for the pa stage to work at ...  so at 3.5 mHz for f, 10-9 power for c, 6.28 for 2 pi  this gives a Xc for a .001 uf cap of 45 ohms which is a low impedance

if you calculate for the .002 and .003 there is not too much change in Xc .... this is why any of these is a fair bypass for the high impedance portions

this 45 ohms is a bit high for the filament bypass .... I think I would something a bit larger say .005 or .01 uf ... why do you think I say this ?

the additional calculation needed is for the highest audio modulating frequency (usually say 10 khz) .... repeat the reactance calculations
the .001 ufd comes in at about 16 kohm .... this is good since it is higher than the 15 kohm plate impedance and since I dont know screen current in this application you will need to calculate it

hope this helps

[W2VW]

Disc ceramics work fine for bypass caps as they don't need to handle large current.

All bets are off if the circuit takes off.

[DMOD]

I asked some of the old timer retired engineers at Collins one time if they could tell me why they chose some of the values they chose.

Here's the theory:

They wanted as low a reactance (10 to 25 ohms, or 1/5 to 1/2 the output impedance of the transmitter circuit) as they could get at 1.8 mHz without exceeding the current and power dissipation capabilities of the then available capacitors, with the RF voltages they were allowed, while using available off-the-shelf component values.

Filament and cathode bypass for RF: The average RF voltage allowed was about 5 volts at 1.8 Mhz. A .005 uF cap has a reactance or RF resistance of 18 ohms@1.8 mHz. I = V/R = 278 mA. Pdiss = 1.4 Watts.

Screen bypass without audio modulation: Again, a reactance target of about 10-25 ohms@1.8 mHz or 18 ohms for a 0.005 uF.

Screen bypass with audio modulation or self modulation: In order not to kill higher audio frequencies, a target of 100 ohms reactance was used. A 0.001 uF  has a reactance of 88.5 ohms@1.8 mHz. This value would be modified depending on the time constant of the RC circuit which comprised the screen resistor and the RF bypass components, including tube capacitance.

For the RF plate choke, they wanted a value such that the RF choke value and the RF cap. value formed a low-pass filter looking back toward the power supply, and had a resonant frequency at least 1/10 of the lowest frequency of operation.

So for a plate choke of 425 uH and a cap of 0.005 uF, the series resonant circuit had a resonant frequency of 100 kHz.

For  RF coupling in interstages I prefer the micas. For RF bypass and higher power coupling purposes, I prefer the ceramics since the ceramic material has better overall heat dissipation capability.

BTW, two 0.002 uF filament bypass caps have a reactance of 22 ohms@1.8 Mhz.

So I prefer the larger physical size 0.005 or 0.01 uF ceramics for filament bypass.

0.001 uF is fine IMO for screen bypass for plate modulated final(s).

Your 6kV ceramics should be just fine for any of the bypass functions.

Phil - AC0OB


http://www.radio-electronics.com/info/data/capacitor/ceramic-capacitor.php

[W9ZSL]

To be safe I think I'll put 2 of the .001 disc ceramics at 6 KV each in parallel for .002 happy medium?  I'm only planning on 75 and 40 meters phone.  According to the Eimac literature, this particular amp with up to 2 KV on the plate needs 26 watts of drive.  I have a DX-20 here that might work.  Too much fun.

[DMOD]

According to the Eimac literature, this particular amp with up to 2 KV on the plate needs 26 watts of drive.  I have a DX-20 here that might work.  Too much fun.

I know. I am driving a Henry 2K-4 with a Hallicrafters HT-40 at 15 Watts out.

Phil - AC0OB

[N4LTA]

Xc (Reactance) cannot be used to calculate real power with V x I or I^2 R. There is no power disappated in a perfect capacitor. Capacitors are not perfect but your power calculation for the filament and cathode bypass capacitors is not correct. It is much higher than the actual value.

The power dissapation is approximately I^2 times Resr  or the current squared times the ESR of the capacitor

[N2DTS]

I always used disks for bypass of everything except the plate in high power rigs.
Values do not seem to be critical, smaller is better for high frequencies in screen and plate bypass caps.

I always go overboard on the voltage ratings though.

[DMOD]

The actual power loss is Power Loss = (2*pi*F*C*V^2) *(D.F.).

AC power is dissipated in the capacitor's dielectric due to the (D.F.) which is the Loss Tangent of the dielectric.

But we're talking rules of thumb here.

[W9ZSL]

Thanks!  That all makes sense.  MK   

[N4LTA]

Well - I^2 x the capacitance reactance is not rule of thumb - it is grossly in error.

Just saying

[DMOD]

Hi Mike ZSL,

In the PDF file is an example of a calculation needed in an actual circuit design.

Page 1 is the schematic, Page 2 is the MatLab code, and Page three are the pertinent results. In MatLab, the "%" sign simply denotes a comment.

In this case the power dissipation of this ceramic capacitor of 820 pF is about 780 milliwatts. I chose a 5 Watt cap. for reliability purposes.

Note that if you include the ESR for any critical circuit build, you would need the ESR at a Specific Frequency AND at a specific voltage.  

But I would not call Ham made Homebrews exactly "critical."

I am sure Hugh LTA would be glad to supply you with the ESR for any of the capacitors for your build.  

Phil - AC0OB