2 x GI-7B 432 MHz amplifier project

This project had three goals. First was to get more power out on 432 for EME and extended tropo work. Second was to learn more about the GI-7B russian triode. Third was to learn more about UHF stripline amplifier construction techniques. It turned out that this was the most difficult tube type power amplifier project I have attempted to date.

The project started when several GI-7B russian triodes were obtained. A study of designs available on the internet led to the conclusion that the SP5CIC design would be the easiest to build. This design was documented on the ND2X QRO web page. In retrospect that conclusion was incorrect. But after the sheet metal parts were fabricated, I was committed.

The chassis was fabricated mostly from scrap aluminum sheet. I used store bought 1/2 x 1/2 x 0.05 aluminum angle to hold the sides together. There are about 300 4-40 holes drilled and tapped in this thing! The anode line was fabricated from copper sheet 0.032 thick obtained from McMaster-Carr as were the hundreds of 4-40 screws.

The anode compartment was pretty straightforward:

A sheet of .032 aluminum was drilled for the 4-40 screws around the outside edges first. Next the tubes were located and holes drilled and reamed to provide a snug fit around the grid flanges. Then the many holes for cooling were drilled. After finishing the plate, the tubes were held down with aluminum clamps.

Next step was to fab the anode resonator. It was pretty complex as it needed to have slots which matched up with the anode cooler fins. Here it is in place before the coolers are installed:

Here are shown the coolers installed:

When I started to fab the cathode line, I found the first of many problems with this design. There was no way I could connect the resonator to the cathodes of the tubes. I played with some finger stock but was not happy with that. I eventually wound up using the YU1AW technique of matching with discreet components.  This is a photo of the completed cathode compartment:

Left center are the filament chokes. Just to the right of the chokes is the cathode matching network comprised of two small piston trimmer caps and a small inductor. The dark strip in the center is just a piece of black tape covering some unused holes.  Next to that is the cathode bias circuit which provides the necessary 30 or so volts bias. Next right is the anode tuning mechanism which pulls down on the flapper cap.

After the cathode compartment was sorted out, I tried tuning up the anode compartment with my network analyzer and found it was way out of band. A slight tweak of the anode resonator length got everything to tune OK but it was later discovered that the line length was correct and the "flappers" (tune cap and load cap) needed to be shortened instead.

When power was finally apply, the power output was much lower than hoped for and there was a lot of thermal drift. All this made it unusable for EME service. To make a very long story somewhat shorter, I discovered that there were a number of problems contributing to the poor performance.

The first problem I found was anode resonator efficiency.
The SM5BSZ  web page (http://www.sm5bsz.com/power.htm) was very helpful for trouble shooting the anode resonator and Leif was very helpful with suggestions for improving efficiency. Unfortunately the mechanical aspects of this design prevented me from implementing the best of Leif's suggestions. I was able to drop the tank loss from almost 20% down to under 14%. Here is the modified tune flapper cap:

Tube mismatch was suspected to be an issue so I built a tube tester from a cardboard box:

While running grid voltage vs. anode current curves at very reduced anode voltages, I quickly found out that I was not running nearly enough air through the anode coolers to keep the tubes at a reasonable temperature. The claims from people running 400 watts out with nothing but a muffin fan for cooling is pure BS! These tubes need around four times the air flow that a 4CX250B needs for the same dissipation due to the poor efficiency of the anode cooler.

OK, with more air I got some curves and discovered that I had inadvertently selected the two most mismatched tubes I had! The two closest were re-installed in the chassis and a painful mechanical re-build to get more air was undertaken. My initial design used a low profile squirrel cage blower mounted on the side of the chassis providing a nice clean low profile. It pressurized the cathode compartment and air was routed up to the anodes through holes drilled in the base plate. See the above photos. Now I had a much larger blower mounted on the top cover and pressurizing the anode compartment. Some air was allowed to route down to keep the filament pins cool.

During re-assembly a slight unbalance in the cathode matching network was discovered which caused a small drive imbalance. This was corrected and the amp was tried again. With all the above mods in place, I was able to run at a reasonable power output with reasonable efficiency, and manageable thermal drift. I can now run EME duty cycle with now re-tuning required during a sequence.

Here are some photos of the final configuration, keep in mind that it looked somewhat better before all the mods:

The shaft  you see has a cam wheel on the far end which pushes against the load flapper cap for adjustment.

Performance is as follows:

Anode voltage = 1950 under load
Anode current = approx. 550 ma.
Power output = 550 watts (more possible but not without thermal drift)
Drive power = approx. 30 watts

Note the power output is key down CW brick on key, not SSB PEP BS power.

I missed some of my goals on this project due to poor information obtained on the web. I had to discover for myself the limitations of these tubes. With the knowledge I have gained on this project, I feel like I can do better next time around. Compared to building HF and low VHF amps, this was very difficult. The dimensional tolerances required are significantly closer than any G-G linear amplifier I had built before. I'm used to doing microwave stuff in a full commercial lab with full machine shop facilities handy. Doing this stuff at home is less fun than I like. I'm really not sure if I'll ever try a UHF project of this sort again.