Hints and kinks for Linears

Hints and kinks for Linears

Postby v51pj » Wed May 11, 2011 6:37 am

the following was published on qrz.com

i think it is good tips

de V51pj

Hints and Kinks
For building high power

Steve Gross

Over 50 years of building power amplifiers for HF, VHF and UHF I have picked up lots of tricks which make the job easier.
Many of these ideas are a compilation of ideas suggested by others who are as passionate about building as I am.*

I use only Russian tubes. The GS35b triode is my favorite for 6 and 2 meters although I’m contemplating trying one on 432 soon. The GS23b tetrode is used on 432 Mhz
Russian tubes are indestructible, 1/10th the price of American tubes and highly under rated as far as plate dissipation is concerned. A gs35b and an 8877 both are rated at 1500 watts but the GS35B weighs 10 lbs while the 8877 weighs 5 lbs..
I have never lost a tube. I rest my case.

All amplifiers are built in an 8 inch square box 19 inches long with 3 shelves in the box. The middle shelf supports the tube, the top cover provides hot air outlet from the anode and the bottom Rf containment plus air escape for air on both from a blower which straddles the shelf holding the tube. Screening of all holes is a given.
The open side is coveres by an 8 by 19 inch cover.

In recent years my focus has been on 50, 144, 220 and 432 Mhz.
8 have been built for 50 Mhz, 21 for 144, 3 for 222 Mhz and 56 for 432 Mhz.
In the early years I like many used stripline designs because I had a nice box and thought it would be nice to use it.
So, I jammed an amplifier into it. As my experience grew I realized the proper approach was to first build an efficient amplifier and then enclose it in some way.
All of my amplifiers are rack mounted and capable of nearly twice the legal limit.

I am convinced that coaxial design or radial cavities are more efficient and make sense because the tube is round and it stands to reason that a symetrical resonator would be more efficient. I even ran across a 1296 stripline design where the tube occupied a full half of the anode stripline!!!
I find it hard to believe that the RF potential around the tube is symetrical.
All of my amplifiers produce efficiencies of 60 to 70 % when properly designed and built.

My experience with stripline designs produce at best 40-50% efficiency.
This, in my experience leads to thermal drift and the tendency to run the tubes hotter than necessary.

The critical part appears to be the output resonator. I have tried coaxial input resonators as well as simple coils and capacitors on 6 and 2 meters. On 432 I took K5GWs 1/2 wave stripline with the cathode in the center of a half wave stripline. and an input coupling probe on one end and a tuning probe on the other end.
It works very well and is far less complicated than the original complicated coaxial input configuration which was suggested by others.
The W6PO “T” match input circuit commonly used with tubes like the 8877 has caused me so much trouble that I came up with an old idea from the 50s.
My solution for 144 and 50 Mhz it a simple series tuned inductor with the tube cathode on one end and a variable capacitor on the other end tuned against ground.
A series tuned link placed at the tube end (cathode end) produces a perfect 1:1 match and does not change with drive level as the W6PO “ T” match does.

The output cavity on 144 Mhz consists of a ¼ wave line section about 17 inches long and mechanically grounded at the end without a bypass. Actually 17 inches is the total length including the tube and sleeve.
Instead of constructing the bypass at the cold RF end of the ¼ wave line I stop the grounded portion of the line an inch or so before it reaches the tube anode and fabricate a sleeve which slips down over the tube and extends up over the mechanically grounded pipe about 3-4 inches with 15 mills of Teflon or Ultem between the sleeve and the grounded pipe which ends up at the cold end.
The only voltage between the sleeve and insulated from the grounded pipe is high voltage.
No RF appears between the sleeve and the grounded pipe since all RF flows on the surface and doesn’t enter between the sleeve and the grounded pipe. Capacity for an overlap of 3-4 inches is so high that it looks like a dead short at RF frequencies.
This is far simpler than trying to build a bypass at the cold RF end of the ¼ wave coaxial section.
Output probes and tuning discs are placed as close to the tube end as possible in order to be able to couple power out and tune the coaxial line section.
In the case of the 2 meter coaxial output section, HV can simply be run down through the inside of the pipe which is grounded and clipped onto the anode of the tube.
No bypass is needed since there is no RF inside the grounded pipe.

432 Mhz Requires a ½ wave long coaxial line section on the GS23b.
It turns out to be 4 ¼ to 4 ½ inches from the bottom of the GS23b anode to the top of the tubing which then equals a ½ wave long line including that part of the ½ wave which is swallowed up inside the tube..
Output probe and tuning probe are obviously located at the very top end of the 4 ¼ inch tube.
Several different configurations have been used for the screen and grid bypass capacitors.
I have used a bypass 6 inches square of brass ( DO NOT USE PC BOARD) for both with the tube in the center with finger stock. 15 ml Teflon or Ultem can be used as dielectric for both. Capacity runs around 6-700 pf. I have also tried grounding the screen directly, using a bypass only on the control grid with the cathode floating. Grounded control grid with bypass on the screen with the cathode floating has also been used.
All work equally well once you get your head around the idea that the cathode doesn’t go to chassis ground.
For a HV feedthrough bypass on 432 I have resorted to a plate of round 1/8 inch 3 inches in diameter disc of aluminum insulated from the wall of the box with 15 mils of Ultem or Teflon. I HAVE SUSPECTED THAT A HIGH CAPACITY .005 OR SO CONTRIBUTES TO FLASHOVER PROBLEMS.
I make a shoulder washer to hold the disc of 1/8 inch aluminum to the wall with lots of clearance for the 8-32 screw which holds the sandwich together. RF choke on 432 consists on 4-5 turns of #18 wire an inch long about ½ inch in diameter and attached to the very bottom edge of the 4 ¼ inch sleeve which is the ½ wave plate resonator.

Now for a couple of tips.
Instead of going through all the trouble of finding a big chunk of teflon with which to make a chimney, I found that a block of STYROFOAM can be cut on a band saw or other means to build a cheap chimney.
It is totally invisible to RF up to 10 Ghz and I’ve been using it for years.
Should you break it, you can knock out another in a couple of minutes.
Next, In order to clamp the output probe in place after you have located the optimum coupling point do this.

Go to the hardware store and locate the section where they sell compression fittings to couple two straight pieces of ½ inch or larger copper tubing together.
They are double ended. Cut it in half producing two pieces.
Remove the ring inside and cut a slot in the ring so that you won’t permanently lock the probe in place. Now you can lock down the probe and by unscrewing the collar you can loosen it up again.
I solder the thing to a square plate and bolt it to the side of the box into which I have built the amplifier.

Ever had a FLASHOVER in your high power amplifier?
Here’s how to stop it.
Always put a 10 mfd electrolytic between control grid and cathode right at the chassis where your feedthroughs come out of the chassis.
I have seen 1 inch sparks in my power supply at the instant it flashes over.
I suspect it’s an audio parasitic which is triggered by who knows what.
The 10 mfd will kill it. Guaranteed

My thanks to all the people who have helped me over the many years are due.
Here’s a partial list.
N4PZ/W9OJI Steve Gross, 1-815-734-4255, w9oji@yahoo.com, n4pz@live.com
602 W. First St.
Mount Morris, Illinois
User avatar
Posts: 64
Joined: 04.2011
Gender: None specified

Return to "Homebrew Corner"


Who is online

Users browsing this forum: No registered users and 4 guests