The AM Forum
March 28, 2024, 10:29:46 AM *
Welcome, Guest. Please login or register.

Login with username, password and session length
 
   Home   Help Calendar Links Staff List Gallery Login Register  
Pages: [1]   Go Down
  Print  
Author Topic: told caps were RF, but no?  (Read 2588 times)
0 Members and 1 Guest are viewing this topic.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« on: March 07, 2020, 03:35:24 PM »

Bought four CERA-MITE 30DKD25 capacitors at the ham fest. I said I wanted plate blocking caps for an amplifier. The seller assured me they are RF capacitors.

I thought to put them between two aluminum plates and using that to block DC from the pi network.

Got them home and looked them up and they are high voltage capacitors. OK sometimes those are used for RF but not in high power stuff as far as I know, and there is at least one post here about a big cap of the type exploding.

These are 30KV 2500pF Y5U.

What does dissipation factor mean from 2-30MHz?

DISSIPATION FACTOR tanδ
Y5U:≤ 20 • 10-3 (1kHz)

MATERIAL
Capacitor elements made from Class 1 or Class 3 ceramic in a molded eproxy case. Screw terminals: brass, silver plate

Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
W4AMV
Member

Offline Offline

Posts: 672


« Reply #1 on: March 07, 2020, 09:46:39 PM »

Hi Pat.

1/(loss tangent)= Q and in this case at 1 kHz, 50.

The loss tangent is not necessarily constant so it would be useful to measure Q at the operating frequency. I measured several 500 pF 20 kV units at 1 kHz that I have on hand and the dissipation factor was an order of magnitude smaller, .002.

So at 1 kHz these blocking caps would have HALF of the equivalent series resistance at 1 kHz.

I would consider a loss tangent of no more than .001 (Q of 500 to 1000) desirable to minimize self heating due to high circulating currents.

I might add, a smaller value of block is acceptable with minor alteration in a PI match
impedance transformation. You might want to calculate for your application how small by considering a PI match say for 50 to 3 k ohm. And then permitting the block to shift that 3 k by just 10% increase. Surprise, the C block value needed is much less.

Alan
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #2 on: March 08, 2020, 10:18:28 AM »

Just for giggles and grins, I ran a PI match for a device at 3 kV and 500 mA requiring a 50 ohm to 3 k match at 2 MHz. The PI net is:

Ctune: 219 pF, L: 21.7 uH and Cload: 1896 pF and Qo is 12

Look at C block at 2000 pF and then C block at 100 pF and finally at 100 pF where we tweak Ctune and L in the PI match slightly. You can virtually achieve the desired plate load Z with a smaller block. The blue curve is the 2000 pF while other curves are reduced blocking.



* PI_match_block.jpg (48.58 KB, 588x589 - viewed 286 times.)
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #3 on: March 08, 2020, 10:29:28 AM »

Granted the Qo is increased and the bandwidth reduced so a bit more sensitive to the tune, but this was a 100 pF block. That BW reduction is evident in looking at the plots of the equivalent Rp seen by the tube vs. frequency shown in the plots attached. The larger block in BLUE.


* PI_match_block2.jpg (32.26 KB, 637x408 - viewed 272 times.)
Logged
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #4 on: March 09, 2020, 08:04:22 PM »

I'll have to really study what you've written because it is a bit past my knowledge. I don't have the software to make those charts or re-calculate a pi network taking into account the plate blocking capacitor's reactance.

That is, I have always used a DC block with a much lower reactance so that it would not interfere materially. I had no idea about the higher reactance being OK with other adjustments.

I usually have relied on handbook values as ballpark figures and then used the calculations in those books to come up with pi networks. They don't really consider the blocking capacitor in this way.

The application here is a little odd because I have already a commercially made heavy duty pi coil with taps and switch I really want to use. It's a bit light on inductance and I have used Tonne software PI-EL to explore Q values taking that into consideration.

Where can I learn about using a higher reactance blocking capacitor?
What software did you use to display those results?

Patrick

Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
W4AMV
Member

Offline Offline

Posts: 672


« Reply #5 on: March 09, 2020, 09:43:17 PM »

Hi Pat,

Thanks for the reply, comments and questions.

Actually the SW is not required to view this problem. Although the
graphs and some of the items that are calculated provide a nice
view of what is going on.

Anyway, the problem of addressing the value of the blocking C
is pretty straightforward. The PI match does the heavy lifting,
transforming 50 ohm to 3 k ohm in this example. Then the
blocking C just forms a HIGH PASS with the termination of
3 k. If you want that high pass to be negligible, I agree, choose
a very large C block. However, as a general rule you might select a
C block value that has a reactance that is 1/10th of the termination,
provided by the PI NET, in this case 300 ohms. At 2 MHz that's 265pF, so 500pF would be more than sufficient. If you choose 1/100th, seems a bit much,  as you
are now at the 2000pF+ level.

So it all depends on the PI match Z level and that of course depends
on the tube load and desired power out, efficiency and tube V and I
values. For your specific case where you have a PI NET in hand, what is the
Z value it provides when loaded by 50 ohm? Or what is the range of Z? Then for the smallest Z you might encounter, choose 1/10th of that Z to set you C block reactance at the operating frequency of interest.

If you do choose higher reactance values for the block C, then the
point I was trying to demo is we really have a 4 element match system!
Three from the PI and the now the C block. It is possible to work out a
nice arrangement of values that will permit a proper tube load AND
incorporate a more friendly value of blocking C not prone to the problems
you mentioned in the post.

SW that you might consider to address this is LTSpice or SimSmith.

It might be fun to play with the arithmetic and see what
range of C blocks would be acceptable for various matching Z's
without taking the easy solution, namely a C block with zero reactance.

Alan
Logged
Pages: [1]   Go Up
  Print  
 
Jump to:  

AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands
 AMfone © 2001-2015
Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
Page created in 0.064 seconds with 19 queries.