A SINGLE 813 AMPLIFIER
A SINGLE BAND 20 METER QRP AMPLIFIER
5W in 125 out .... 15 in over 300 watts out!
SCHEMATIC #1 SCHEMATIC #2 SCHEMATIC #3
FRONT
RF SIDE
HV SIDE
TOP
BOTTOM
Note: 1) It has come to my attention that there is an error in the RF schematic. The capacitor going to the screen diode stack should be a 47nf (.047 uf) capacitor, not a 470nf (.470uf).
2) It appears he 1N4007 rectifier across the coil of "K1" is shown in the wrong direction.
Introduction
There has been many 813 amplifier articles written over the years in a half a dozen radio magazines and this one is not much different except for some outback engineering and humor in building it. Recently there was a very nice article written by C. J. Bourke, VK3YE featured in AR not so long ago. It featured 2) 813’s in the G2DAF configuration with Part #1 in April 1999 and Part #2 in March 1999. I highly recommend you read this fine article.
I have decided not to include the theory and concentrate on what someone is up against when trying to build a linear amplifier when living in outback Australia. You can only pester your mates for so long before they get annoyed at all the requests to source parts and post them to remote locations. It turns out that the local rubbish tip can be an excellent choice for parts if you have access to it and you don’t mind being seen there.
The usual warning applies here.This project uses high voltages and high voltage can and could kill you If you don’t feel that you’re competent with this, possibly you should pass on this project or look into building a solid state amplifier which uses 12 or 24 Volts DC.
The
Transformers
Firstly, the flux shunts must be removed. These are the 2 small stacks of steel laminations between the primary and secondary windings. A small length of 6.5mm square key steel and a hammer where used to do this. This leaves plenty of room to wind the extra turns on the primary as described later.
Secondly, one side of the HV secondary is tied to the actual steel laminations of the transformer. This must be removed so that no windings on the hv secondary are touching the core laminations!
Not having any sort of hv probe to test the transformer and for safety reasons, I decided to apply an accurately known low AC voltage of 6.30 VAC onto the primary and accurately read the results on the secondary. After this was completed it was found that the transformer had a turns ratio of 1 to 9.72, or for every volt applied to the primary there would be 9.72 times the primary volt on the secondary. Being 240 volts on the primary, there would be 2333 volts on the secondary. After rectifying and filtering this would raise to about 3300 VDC. Just a bit to high for the 813. Looking at the secondary, which was totally encapsulated in varnish, it was decided to add turns to the primary to try to reduce the turns ratio, hence, reduce the secondary high voltage. Enough turns were added to the primary to bring the secondary voltage down to about 1980 volts AC. Now after rectifying and filtering the voltage is a respectable 2800 VDC.
The filament transformer is of the 50 VA vintage, again found at the local tip. It had a 240 volt primary but the secondary needed to be rewired to accommodate the 10vac, 5 amps and center tapped requirements for the filament of the 813.
The control voltage transformer was removed from a discarded portable stereo at the tip. It had secondary AC voltages of 9 and 18. When rectified and filtered they produced 12.75 VDC and 25.50 VDC. Almost exactly what was needed!
HV Electrolytic Caps
T
Enamel covered wire from a motor starter contactor coil was used to wind the choke. I’m not sure exactly how many turns are on it because I can’t see that good but I think it’s just about 200. A few coats of varnish were brushed onto the windings. Beware no to over do it like I did. I later tried to remove a few turns from the top of the choke but found it difficult now that the varnish brushed on was so thick that it had encapsulated the fine wire.
The Tuning
Capacitor
To fix this problem all the stationary stator and movable rotor capacitor plates were removed, except one stationary plate on the end. A small piece of circuit board material was used as a spacer and placed on top of this last stationary plate. Now one of the movable rotor plates earlier removed was placed on top of the circuit board material and re-soldered back onto the rotor. The circuit board material was now removed from between these two plates and placed on top of the newly soldered in movable rotor plate. A stationary stator plate was now placed on the circuit board material and re-soldered in place. This continued until as many plates could be soldered back on to the capacitor stator and rotor.
Now that the newly modified capacitor is complete, a new problem has now arisen. The capacitor seems to have lost a lot of it’s capacitance now that the plate spacing was almost tripled! So to cure this problem a small 30pf doorknob capacitor was placed in parallel with the newly modified tuning capacitor. This doorknob is not shown in the photographs.
Plate Coil
Panel Meter
Chassis Parts
The front, back, bottom and internal chassis partitions were all made from wrecked aluminum signs recovered from the tip after the clean up from Cyclone “John” which crossed the N.W. coast in Dec. of 1999. The top was the last part of the chassis to be made. It was cut from existing 1mm thick aluminum sheeting that was on hand and bent with a home made sheet metal brake.
It was decided to recess the 813 socket by the length of the metal base of the tube to reduce the height of the chassis and give the amplifier a lower profile look. More along the lines of a lower and wider chassis rather than a taller and narrower look. Recessing the tube dropped the height by the thickness of the metal base of the 813 and the ceramic socket.
The majority of the chassis parts were “finished” by sanding them with water and 120 grit wet and dry sand paper. Sanding should be done all in one direction and done long enough to remove any pencil, pen or scratches in the aluminum. After sanding a light coat of cheap clear lacquer was sprayed on them. The longer you sand them the better they will look, so beware you might get soar fingers!
The Passive
Grid
Testing and
Tune up
Tune up was a cinch! A small amount of drive [single tone] of a few watts on 14.175mhz. was applied to the input of the amplifier with the SWR/Watt meter inserted in between. The two Arco mica trimmers on the input were adjusted until the SWR reading was flat. SWR at the band edges only slightly raised when driving it at the top and bottom of the 20 meter band.
Now the SWR/Watt meter was inserted in between the amplifier and a 50 ohm high power homebrew dummy load. 5 watts was applied to the amplifier. The screen voltage increased rapidly to about 95vdc. Output on the watt meter indicated 135 watts. 10 watts was applied, screen voltage increased to 150vdc and output on the wattmeter indicated 225 watts. 15 watts was applied, screen voltage increased to 175vdc and output on the wattmeter now indicated 300 watts! Beware, with a high value of passive resistance it can be lightly driven and the 813 bottle will take off like a rocket!
If plagued by low output but everything else seems to be working fine, try testing for shorts in the stack of 1N4148 diodes. I unknowingly drove the 813
for one very brief 50+ watt key up and shorted out one half of the diode stack. This plagued me for a few (hours) until I noticed low screen voltage being developed and finally traced the fault back to the diodes.
Summary
During the construction of this project the only test equipment available was a digital multi meter. All other equipment either didn’t work properly or was unavailable. So only on air QSO’s to other station was the only post construction testing available.
Total cost of the amplifier was around 25 dollars. Of course that’s not including my time.
References
1) The original article by G. Thornley G2DAF
2) G2DAF article by C.J.Bourke VK3YE Apr., May 1999 AR magazine
3) ARRL Hand book 1990
4) Construction techniques from books and articles by Drew Diamond VK3XU