Hi Everyone,

I built this little 2m amp a couple of years ago, it works perfectly in every way..... except for the thermal drift.

Tuned up with key down, 300W out with 12W drive, within a few seconds it droops to under 150W.

Keeping the power up requires *constant* re-tuning of the plate.

The drift is so fast I think it must be within the tube or close to it, there just isn't time for the tank circuit to heat up.

I don't know if I made a mistake in the engineering of the tank circuit, or if the tube is faulty

Open to all ideas about how to fix it.

Thanks
Tim G4LOH

开云体育

G'day Tim.

Check (physically and electrically) your plate blocking/coupling capacitor very carefully. John, VK6JX, had the exact same problem with his GS35B model 2m amp, using what looks to be the same Russian capacitor.

Likewise, we thought it was some sort of thermal drift problem and you are right, you can't follow the rapid drop by adjusting the tuning. It is too quick and you never catch up.

His 2200pF/10kV capacitor failed on the very first key down and it was not obvious at all, until he removed it and looked at it with a magnifier lens. There was a hairline crack and a little blackening in the ceramic from the failure. I suspect that those caps are not happy at 144MHz.

Good luck and 73, Alek VK6APK

+1 on the blocking cap. I have a 2xGI7B project on the go for 144MHz with a very similar set up, and the looming possibility of thermal drift is a great concern. I thought about employing one of those russian doorknob caps but chose a design that uses 2 slabs of copper with a PTFE sheet between them instead because I had read about thermal issues with those caps elsewhere. Time will tell if that was a good choice or not! Hopefully if the cap is your main issue, you can find a suitable replacement that doesn't send you back to cutting too much metal.

Further, maybe check the rating on your fan as well. You might want to get one with higher CFM. You'd probably do even better if you replace it with a small centrifugal blower that can pressurize the compartment. Muffin fans aren't so good at moving air against static pressure. The tube itself only needs 15 CFM, which isn't much... but if it's only a 15 CFM fan to begin with, it probably can't attain that through the restrictions in the compartment.

Also, I think the thicker the copper sheet, the more stable it will be (ie; more mass). If the copper is quite thin, you might never get it stable... though that might be more relevant to my chosen design than yours.

Good luck, hope to see an update on the outcome! :)

开云体育

Thanks!

I'm down with COVID right now, the junkbox has some other Ceramic button caps, I switch it out in a few days and let you know how I got on.

vy73
Tim

开云体育

On Wed, Dec 27, 2023 at 04:08 PM, mike repinski wrote:

How can this be done?

Insert a teflon sheet or other suitable insulator between 2 copper slabs, one of which is directly on the anode.

Here is mine to give an idea. It's based on plans published by YU1AW.

(Ignore the paperclips! It was just a mock up to see how much metal work I'll need to do!)

Watch out.? You will introduce a ton of stray C on that hb teflon? plate block cap assy.....to the chassis below it.?
That amounts to a bunch more anode to chassis C.??

?

Knowing the spacing, and the surface area, you can calculate out the C.? ?You can also easily measure it with an LCR meter.

?

That hb block cap assy ( with the paper clips) is massive.

?

Jim? VE7RF

On 12/27/2023 9:51 PM, Jim VE7RF wrote:

Watch out.? You will introduce a ton of stray C on that hb teflon? plate block cap assy.....to the chassis below it.
That amounts to a bunch more anode to chassis C.

Negatory.

Being HFers, we can be forgiven, perhaps, for forgetting that at V/UHF, things are different... and there are such things as "transmission lines". This is a traditional stripline plate tank, which configuration can be calculated to resonate at desired V/UHF/microwave frequencies depending upon the width, length, spacing between the line and the top and bottom ground planes, AND the thickness of the line; all of these determine the line parameters. These parameters were all determined nearly a hundred years ago by a guy named Harold Wheeler, and many articles and some books have been written since then to tell and show how to calculate transmission line parameters. My Elmer, W6GGV (SK), originally wrote an article covering the calculation and use of transmission line equations during the early '60s, then wrote equations into programs for the old HP67/97 calculators and published a series of articles in "Ham Radio" magazine just after Jim Fiske, W1HR, died around 1980 or so. Later, I adapted them to the TI-93 calculator but by then, such calculators were no longer in vogue so I never published my programs. Shortly after Hank's article was published, Brian Beezley codified the equations into a for-profit program for the home computer.

Knowing the spacing, and the surface area, you can calculate out the C.? ?You can also easily measure it with an LCR meter.

No need, since it's a stripline. HOWEVER...

That hb block cap assy ( with the paper clips) is massive.

Yes, it is... the impedance of the original stripline line is no longer maintained because of the additional thickness of that humungously-thick sheet of teflon being used between the two very-thick copper plates, which thickness looks like it has been vastly increased from the original dimensions. In reality, for an anode voltage up to around 5 kV, the teflon insulation only has to be about 0.02" thick. For a pair of 4CX250Rs or 4CX300As operating from about 2.5 kV, 0.010" has proven adequate. The flashover voltage of teflon, if I recall correctly, is about 500 - 700 volts per mil (one-thousandth of an inch). As with a lumped-constant blocking capacitor (such as a doorknob), you want the voltage-holdoff capability, or in this case, teflon thickness, high enough to withstand at least double your plate voltage and preferably 3 or 4 times. But if you make it too thick in efforts to increase the breakdown voltage, then the whole line becomes too thick and the line width or length has to be changed to compensate. This becomes especially critical in the original K2RIW design which used standard 2" or 3" aluminum chassis to enclose the tank, because the shorter chassis height caused the line thickness to become a much more significant portion of the line thickness within the stripline.

And BTW, this is also why these types of amplifiers do not perform properly when the top cover is left off: that top cover is actually PART of the anode tank circuit. And as a consequence, it absolutely MUST be solidly-connected to the bottom chassis walls with el mucho fasteners! A screw every 1-1/2 inches is not unreasonable; one every 2 inches is questionable and further apart is just asking for low efficiency and poor overall performance. This, of course, becomes even more important at 432 MHz.

It is absolutely unnecessary to use quarter-inch-thick copper "heat spreaders" for such plate lines: they just don't get that hot. When they do, the outlet air from the tube tends to keep them cool enough that the teflon insulation is not compromised. This is an excellent argument for having the exhaust outlet located at least part-way down the chassis, rather than directly on top of the tube. Otherwise, you have absolutely NO air flow whatsoever around the plate tank; so of course, it's going to get hot with use, and probably change dimensions and thus resonance.

Years ago, when I was building lots of these amplifiers using -250Rs, -300As, 8122s, and Svetlana 4CX400s, I used to use ordinary double-sided teflon PC board for my plate lines, most often 0.012" thick. I never had one fail, nor did they ever exhibit thermal instability. I also used sheets of 0.012" teflon AND G-10 fiberglass PC board to build plate bypass capacitors for 220 and 432 amplifiers. The copper thickness of that material was 1 or 2 mils; the capacitors typically were just 4 square inches large. To increase the bypass capacitance, I would sometimes use 0.006"-thick teflon sheet but when I did, I had to be extremely careful to make sure there were absolutely no burrs or sharp edges anywhere on the surfaces carrying high voltage. (I had a lot of those capacitors punch through!)

I have seen half a dozen otherwise-beautifully-machined 144 or 220 or 432 amplifiers using this -RIW stripline technique that failed to work properly, most often exhibiting very poor efficiency. The cause is almost always because the builder substituted thicker materials for the original anode stripline circuit components, failing to recognize that the line impedance changed as a consequence.

Another cause of such failures out-of-the-box were due to the builder substituting for the original chassis height enclosing the whole stripline, usually going to higher chassis heights such as 4" or even 6".

What has to be kept in mind, when dealing with V/UHF circuits that no longer use simple lumped-constant L-C circuits, is that now, you are into the realm of the stripline... and it does not resemble our ordinary inductor-capacitor circuits at all.

73, Steve K0XP

On Wed, Dec 27, 2023 at 11:30 PM, Steve wrote:

the impedance of the original stripline line is no longer maintained because of the additional thickness of that humungously-thick sheet of teflon being used between the two very-thick copper plates, which thickness looks like it has been vastly increased from the original dimensions.

I'm not sure if this is in reference to my picture or not, but if so the camera may be playing tricks on you. The teflon I'm using is .4mm thick (in fact it's not even teflon in that mock up, it's a sheet of paper). The copper is only 1.2mm thick (except the bent support bracket, it is slightly over 2mm). Both fall within what was specified in the design I used (which, to be fair, I have no knowledge of its effectiveness... but hey I wanted to try, this is how we learn).

Anyways I didn't mean to hijack the thread. Your notes about the dimensions of the cover are appreciated. I plan on making the enclosure soon!

Happy new year to all!

For the OP, does that plate block cap even look like a cap at? 144 mhz ??? ?Will it handle the calculated rf current flowing through it at 144 mhz ??

John Lyles,? K5PRO helped myself and Scott out on that 10 kw 6m amp.? ?John used his test gear during a 3 way phone conversation, and was testing various HEC / centralab caps, like the 200 pf? and also? 500 pf variety, in both HT-50/58 size, and also HT-57 sizes.? On some of them, the SRF freq (self res freq)? was as low as just 42-45 mhz.? If 3 x caps were in parallel, SRF rose to 62 mhz.? On the HT-57 caps, in 200 pf size, SRF was 60 mhz...and? even higher, when paralleled.?

Above the SRF freq, the cap looks inductive.? ?It's straightforward? test process to find the SRF freq for a given combo of single, or multiple caps in parallel.?

开云体育

4x HT50 or 55 do well on 2m with 4CX1000/3CX1500A7 amps.

GI7b and similar tubes drift when cooling is insufficient.

?

73

Peter

?

?

?

-----Original-Nachricht-----

Betreff: Re: [ham-amplifiers] Gi46b on 2m - Thermal drift problem

Datum: 2024-01-04T17:29:29+0100

Von: "Jim VE7RF" <jim.thom@...>

?

?

?

?

?

?

It seems fine, I suppose, even above SRF, it's only a few nH.... and is still a DC block ;-)

The Gi46b runs low power as a driver for the final triode so has a pretty easy life.