The Evolution of the "Rock Crusher" I built my first homebrew linear amplifier in 1968 as a new ham who dreamed of power and who attended college on the side. It was a lowly pair of 813's in grounded grid laying horizontally. It looked nice and I prided myself in being able to put those puny SB200's and 30L1's to shame. From then on I was hooked on the power game. I soon got a respectable offer on the 813's (which had grown to three) and sold it. I also developed a respect for high voltage having got my fingers into the 1200vac on the secondary of the plate trans- former and having been shocked so bad I couldn't bend my arm for half an hour. Free parts were pouring in and it was time for the "Rock Crusher." Background It was going to be the world's most powerful 4-1000A ever built for the ham bands. Bandswitched..Vacuum Variables..Table Top.. Sharp Looking..CCS rated...awesome..etc. For the uninitiated a 4-1000A (or a 4X1 as the Big Guns call them) is a glass envelope radial beam tetrode with 1000 watts of juicey plate dissipation. Capable of full power operation to 110 MHz, these behemoths of the vacuum tube set are easy to obtain at flea markets, ruggedly built, and nearly 10 inches tall. The object was to build a grounded grid five band amplifier that would maximize the 4-1000A's ability to provide 2 to 3 kilowatts r.m.s. output. The goal turned into an obsession with finding the Holy Grail of Linears. It was eventually to operate on 160 thru 10 meters and take 20 years to perfect! Examination of the general tube characteristics will make the neophyte linear builder salivate with excitement. High plate voltage is desirable for many reasons. First and foremost is higher output power. In addition, one would expect reduced drive requirements, better linearity, and less possibility of exceeding grid dissipation as long as proper biasing was used. I was to find that all the above was true and that the tube can be operated with up to 7,000 volts with only minor drawbacks. What evolved had an estimated plate resistance of 6000 ohms based on a plate voltage of 7000 (10% regulation at 1 amp) with a plate current of 600 ma. at 4000 watts input. The tube would provide 2200 watts output with 120 watts drive. That's 3500 watts peak output on SSB considering power supply dynamics. Not bad for a Rock Crusher! Since the plate dissipation would be 1700 watts in tune I decided to use a 160cfm blower to provide required 20-35cfm. I designed the plate tank around a PIDUX 195-2 and vacuum variables for tuning and loading. Biasing I wanted the amplifier quiet in standby. Vacuum changeover relays are notoriously leakey so I used a spare set of change- over relay contacts to ground the amplifier output in standby. I also provided 110vdc for application to the cathode circuit to assure that the tube was biased beyond cutoff. The operating "Q" point bias was derived in a unique way. The tube requires 5 to 7 volts applied to the cathode circuit to operate in Class B. Since the grids are GROUNDED (I mean with heavy bus bar) the bias must be applied as a positive potential reference to ground to the center tap of the filament transformer. The best system I've found is to use a string of forward biased silicon diodes. The anode end is grounded and the cathode end is connected to the center tap during transmit. A single pole multi position deck switch is used to adjust the 'Q' point by shunting out unneeded diodes. Each diode drops 1/2 volt so about 12 are used normally to get 6 volts. This voltage drop occurs whenever plate current flows and the voltage developed is independent of the current drawn. Parasitics It's easy to make these brutes work up to 20 meters but strange things happen when that plate tuning capacitor is near minimum capacity! All of a sudden thump and you pin the plate current meter and bend the hand, blow out the current reference resistor, blow the power supply fuse (if you have installed one) and maybe melt the tube plate if you don't shut it down fast enough. You are the victim of VHF parasitics. Those 813s look pretty sweet right now! I've tried a whole box full of parasitic suppressors but the recent articles by Rich Measures AG6K treat the subject fully and I will only relate what works for the '4X1'. The first and most essential step is the use of a series resonant parasitic trap from either filament lug directly to ground. Optimum parts values are .2uh consisting of 4 turns of #16 oxidized nichrome wire 1/2 inch diameter 1/2 inch long in series with a 2.5 to 10 pf mini air variable capacitor. Using a grid dip meter, determine the plate parasitic frequency: Adjust bandswitch, tuning, and loading for 10 meter operation with all power off. Look for the dip between 90 and 110 MHz right under the wire connected to the plate cap. Then unsolder the parasitic trap coil from the filament lug and solder directly to ground. Using the grid dip meter, adjust the parasitic trap to the same frequency. Reconnect to the coil to the filament pin disturbing its shape a little as possible. This low 'Q' trap swamps out the feedback path in the cathode circuit. #16 nichrome wire was salvaged from a burned out element in my electric clothes dryer. The final path to total stability requires an examination of the anode parasitic suppressor. The 4-1000A at 6000 ohm plate resistance uses only 14 pf for the plate tuning capacitor in the plate tank. The 4-1000A output capacitance is about 8pf leaving only 6pf for the sum of the tuning capacitor and all stray capacitance. By golly my tuning capacitor is at minimum and I still can't get it resonant on 30 MHz! So I reduced the 10 meter tank coil value just enough to make the circuit resonant on 30 MHz. This hurt the efficiency somewhat but was a better alternative than reducing the plate voltage for just one band! The final value was 2.0 uh- 5 turns of flattened 3/8 inch copper tubing, 2 1/4 inches in diameter and 5 inches long, which replaced the PIDUX supplied 10 meter coil made of strap. The actual anode suppressor was next. Using the formulas from Rich Measures' article, I discovered that 4.5 amps of current was flowing through the anode suppressor and the tube output capacitance, half of my total on 10 meters! No wonder my plate efficiency was poor on 30 MHz, the anode suppressor was a significant part of the tank circuit! Because of the cathode circuit swamping trap I was able to reduce the anode parasitic inductance to .04uh and still maintain total stability. Most of my box full of scrap high 'Q' shiney silver plated parasitic trap inductors were .2uh. How many 200 watt 100 ohm non-inductive resistors have you seen in the anode circuits of 4X1 amps? Plenty. 0.2uh (38 ohms at 30 MHz) developes a voltage drop of 170 volts at 30 MHz, forcing 292 watts of 10 meter energy to be soaked up by that poor parasitic resistor. No wonder they start smelling bad in tune on 10 M! The final solution was a 1/2 inch wide 1 inch long hairpin inductor made of #16 nichrome wire clamped to a 50 watt 200 ohm (cold) Globar resistor. The hairpin inductor is only 0.04uh (7.5 ohms at 30 MHz.) resulting in a 34 volt drop and a resulting 6 watts dissipated in the Globar on tune. As the Globar heats, its value falls to 40 ohms thereby increasing its dissipation to 29 watts. The oxidized nichrome is lossy at VHF and its low 'Q' make it effective over a broader frequency range; You will find that the parasitic frequency varies depending on the bandswitch and tuning capacitor settings, making lossy broadbandedness desirable. The nichrome doesn't help amplifier efficiency on 10 M so some loss in output is experienced. In fact, power output is reduced from 2200 watts to only 1700 on 10 meters at 44% efficiency. I finally found a use for that old gray ugly dryer element wire. It was hard to solder to but judicious scraping and wrapping some #22 copper onto the connection points "wire wrap" style seems to have worked. The amplifier is stable under all conditions. WARC band and 160 Meters WARC band and 160 meter operation provided me with two more years of fun and nightmares. The most difficult problem was finding or building a plate RF choke that would not blow up, roast, melt, or in some other way embarrass its owner on some band or another. In the beginning I used a B&W 800. Then there was 12 meters. The B&W 800 has a self resonance at 24.5 MHz and self destructs on 12. The B&W 800 used #28 guage wire and had an inductance of 100uh. On 160 M too much current flows resulting in overheating and those beautiful windings sag from elongation and loosen. The coup de grace happens the next time you tune up on 10 or 15 meters and the coil resonances have shifted. I built a delrin core choke of #22 copper. It worked well but still showed signs of overheating in the 4X1 on 160 M. It was 1 inch away from the chimney but the failure mode was caused by the wire overheating in 160 M contest operation; too much current for its 110 uh. The wire stretched and moved the coil resonance too close to 20 M. It never exploded but its reactance was so low on 20 M that some associated plate supply components were ruined. I even had a small exciting fire one evening. Delrin will burn if it's hot enough! It was indeed becoming challanging to get this thing to work all bands. I tried chokes at right angles; unsuccessful. I tried segmentation; windings arced to each other. I tried spaced out segmentation schemes; they each ended up with incurable series resonances in the wrong place. I needed 240uh and couldn't get it! Finally, I gave up! I built a two section choke. Each section is soleniodal; the top section is a modified Ohmite Z7 with 34 turns removed making it 80 uh with a lowest self resonance at 34 MHz. It consists of #30 wire close wound on a 1/2 inch diameter ceramic form which is epoxy coated. The modified Z7 is stacked end-to-end on top of a 1 inch diameter 4 inch long delrin rod with 140 turns of closewound #26 which is varnished. The delrin rod bottom section is 160uh with self resonances at 16.2 and 25.5 Mhz. On 160 and 80M the whole combined choke is used for 240uh with a combined lowest series resonance at 11.5 MHz. On 40 meters a high voltage relay shorts out the bottom RFC section. Presto- no more trouble. An extra wafer on the bandswitch shaft is used to provide ground to operate the relay on 40 M and higher. On 160 and 80 meters the same wafer grounds a 1500 or a 500pf capacitor on the loading side of the tank, since the 1000 pf of the loading vacuum variable is insufficient on these bands. Adding extra tank coil inductance on the end of the PIDUX 195-2 proved to have a few novel surprises also. 20uh (21 t) of B&W 3033 coil stock was used to provide the extra inductance for 160 meter operation. The 160 meter conversion also required the bifilar filament choke core be changed to an Amidon R33-050-750, doubling its reactance, as well as construction of appropriate input circuits. The plate tuning vacuum variable was sufficient for 160, a blessing of high plate resistance! The amplifier loaded up fine the first time tried on 160 meters. Success! Well, almost. No output on 20! The newly formed gigantic plate tank coil had a titanic flaw, it was self resonant near 15 MHZ. The tank was so busy soaking up all the r.f. that nothing could make it to the antenna. Now what? I like 20 meters. A brilliant idea struck me. Why not open up the cold end of the tank instead of shunting it out. That should change the series resonances. It worked! The 15 MHz self resonance moved to 18 MHz. The amp worked on 160, 80, 40, 20, 15, 12, and 10. Panacea. I was estatic. Then someone came up with 17 meters. The first time I loaded up on 17 M the open end of the plate tank coil arced like one of those gadgets in a frankenstien movie. 1 inch flames. Not Good. Another brilliant idea hit me. Lets short the open end of the tank to ground and try again. 2200 watts on 17 meters happened, but 20 meters didn't operate. Brute force solution: I built a high voltage grounding finger that shorts the tank to ground only on 17 meters! All h.f. band operation was a reality. Lets hope they don't invent any more bands or I'll have to start over with a roller inductor! QSK The next mod was made to make the amp fully QSK compatable and to reduce the current flow in the keying jack circuit. You must remember, this amp was contructed when my station featured a Drake TR3 tube tranceiver! Fortunately I had already upgraded to a Jennings vacuum changeover relay, so the mod only required the addition of a small perf board TTL logic NOT gate and a PNP transistor switch to operate the relays. The Jennings allows full break-in CW operation; the bias relay is a little slower but keeps the cutoff bias on for <1ms after the vacuum relay makes. (The 4x1 puts out nearly 8 RF amps into a 50 ohm load). The CW waveform looks great on the scope as the relays clatter away. The amp can now be keyed up via grounding a keying jack or by +12vdc applied to a companion jack. The grounding jack has 35vdc when opened limited to 5 ma. The 12vdc jack needs no more than 7 ma making it compatable with the lastest solid state radios. Conclusion I've been using the 4X1 now for nearly 40 years and have gained invaluable experience with it. Every year I learn more about power amplifiers. As you can tell I like to experiment. It sure is fun to rip a hole in the ionosphere and tromp all over those puny 3-500Z's in pileups. It's also fun to hear the horror stories about how someone melted down his 4CX1200A7 on 12 meters let alone 17 meters.... The operating manual is available for my 4X1 which includes all discriptions, specifications, schematics, pictures, and bibliography for $10 to handle postage and repro costs. Or email me at jaybo@pacbell.net for the .ZIP file, about 1MB. Jay Bennett AA6GK ??