For details on GI-7B tube socket construction, click HERE.
This project actually started over a year ago. I purchased a Clipperton QRO amplifier that was a basket case. It had the appearance of a typical radio kit. There were bags of components and hardware and a large assortment of additional items, less the four 572B tubes.
The size of the unit made it an excellent candidate for a pair of the 4CX800A tubes currently available on the market. Before I could start on the 4CX800 project, I discovered some triodes designated GI-7B (350 watts dissipation each and hi-mu) that had a possibility of fitting into the available space in the RF compartment of the QRO. Additionally, the regulated grid bias and screen voltage circuits are not required to operate these tubes. The project began on 2/14/01 and was completed on 3/18/01. A total of 85 hours was used to complete the project.
The tubes were mounted on individual plates and attached to the plates at the grid ring with four individual straps. These two plates were mounted under the top chassis plate and spaced down .312-.375" to allow air to pass through to the anode. This also allows air to cross the heater and cathode seals providing more than adequate cooling for them. Silicon rubber chimneys were fabricated and installed to direct airflow through the anodes and out the top of the amplifier. This 5" x 7" plate was mounted on top of a 5" wide x 7" long x 2" high chassis that was mounted upside down in the back of the RF compartment of the QRO. The center bulkhead of the QRO was cut and moved approximately .500" in the direction of the power supply compartment to allow the chassis plate top to be removed easily. An entry hole was made into the back center of the chassis for the blower and a back plate made to enclose the area the original muffin fan was mounted on. The blower is A Dayton #4C761 and is available from Grainger. A resistive pad of 165 ohms @ 25 watts was used in series with one lead of the blower to slow it down for noise reduction purposes. It will still deliver 40cfm of air at .2" to cool the tubes and is extremely quiet.
The power supply compartment was rearranged to accommodate all the additional components needed for the project. The original transformer was moved forward and rotated 90 degrees. Two DPDT 10 amp relays were mounted behind the transformer and installed in series with the second relay using two 20ohm 50watt resistors to slow the inrush current for a soft start. The first relay is activated from the front panel On/Off switch and the second relay pulls in as voltage goes across the 20ohm resistors, giving a slight delay. An additional filament transformer (12.6vac @ 6A) was installed and was regulated with a series 20ohm 50 watt resistor installed in one primary lead to control the secondary voltage at 12.55-12.65 volts @ 4amps when under load. A time delay unit (120 seconds) controls a separate relay, thus limiting key up until the warm up period is complete. A bias circuit made of discrete components (1 amp fuse, 2N3055, 41 v zener, 50 ohm, and 15K ohm resistors) was installed in the B- return from the filament transformer CT. 41 volts of bias was required to obtain 100 ma. of resting plate current. Originally 40K ohms of resistance was installed as cut-off bias. This was insufficient due to the gain of the tubes and was retrofitted with a 100k value. No plate current was evident after the change. Two K2AW 6Kv rectifiers were used in the voltage doubling circuit along with a bank of eight 330ufd. 450v electrolytic capacitors wired in series (41ufd total). The B- is kept 20 ohms above ground for metering purposes. A grid current metering function was added to the metering circuit. In the original circuit only plate current and high voltage was monitored.
The project went well with one exception. On start up there was an AC component modulating the cathodes that was very noticeable on the transmitted signal, both in the SSB and CW modes. I had wired the heater and heater/cathode rings on the base of the tubes with the heater connections and heater/cathode connections in parallel. The RF drive was coupled with a .02 disc capacitor to the heater/cathode side of the network. A .01 disc ceramic capacitor was added across the heater and heater/cathode wiring for equalization. This worked well and the amplifier performed well with regard to amplifying and elevating the amplitude of the signal. There was a constant rushing sound between syllables while operating SSB. This was the AC component modulating the cathodes. An extensive effort to isolate and cure the problem was undertaken with little success. Finally after consultation with friends K4DPK and K4POZ, both of which are more experienced than myself, K4DPK suggested modification to the cathode wiring. The jumpers between the tubes were wired from heater (tube #1) to heater/cathode (tube #2) and from heater (tube #2) to heater/cathode (tube #1). Equalization capacitors were installed and each filament choke wire was attached to the individual cross over wires. The heater/cathode terminals were fed individually from the input coaxial cable with individual .02 coupling capacitors. This cured the AC problem by eliminating the AC component on the cathodes making the circuit symmetrical and balancing the transformer load.
The end result is a convenient sized tabletop RF amplifier that will easily produce 800-1200 watts out. All indications are that the tubes run clean when managed properly. The GI-7B tubes run tremendously cooler than the original 572B tubes and extremely cool for an amplifier of this power level. Initially, stability was a serious concern. Judging form the outward appearance and specifications, the GI-7B tube is designed to run vhf and uhf. Initially, the stability of such a high gain tube was a concern and much precaution was taken to avoid this scenario. Stabilization was done by using two parasitic suppressors made from Nichrome strap (4 ohms/foot) and two 68 ohm at 2w resistors. No instability has been noted on any amateur band. Great care was taken to isolate the grid and cathode compartment form associated anode circuits.
This was a fun project to do, but wasn't without some trying times. Fitting all the components into the power supply compartment was a true adventure in layout. The RF compartment at this time is essentially stock other than the left to right tank coil (40-10) was moved forward about one inch to allow room for the plate choke and blocking capacitors. The load padding capacitors were moved closer together to allow access for a mounting bracket to mount the high voltage diodes in the supply compartment. Please realize this text does not allow for, or cover every scenario and obstacle encountered in a RF generation project. Have fun!
Lawson Summerrow [firstname.lastname@example.org]
Best Viewed: MSIE 4.0 & higher, or Netscape 3.02 & higher, and screen set to 800x600 pixels
Last Updated: 1 August 2001
Feedback: Paul S. Goble, III, ND2X
Copyright © 1998 - 2001 by ND2X & W4EMF, all rights reserved