The completed transmitter runs 30 watts input and 20 watts output. The power is limited by the 350vdc plate supply. For 50 watts input a plate supply of 460vdc at 150 ma would be required.
Just as a test, an external, 500vdc supply was used as a plate supply. The result was an input of 47 watts with an output of 33 watts. I doubt there would be much difference in performance between 20 and 33 watts output. Certainly not enough difference to justify a seperate, external power supply. However, if you are building this thing, you might plan on a larger transformer and perhaps a larger case.
Operation on 80, 40, 30 and 20 is very good. A slight tendancy to chirp on 20 which can be eliminated by using the OPR position of the function switch allowing the oscillator to run continuously while the amplifier is keyed. On 15 meters, the operation is flakey regardless of operating mode. Grid drive is above 1ma on the lower bands but drops to about 0.5ma on 15 meters. A more active crystal might solve this problem but my stock of ham band crystals is limited. Ten meters has even less grid drive if straight through operation is attempted. Ten meter operation is reasonable with the oscillator peaked to 20 meters and the amplifier doubling to ten meters. Under these conditions, output is only about 10 watts.
The only solution I can see to the drive problems on 15 and 10 meters is to add a driver section between the oscillator and the final. This is done in the majority of better quality CW transmitters such as the DX-35, DX-40, DX-60, EICO 720, HW-16, and others. It may be possible to improve the grid drive situation by using a dual purpose tube such as a 6U8. Use the triode section as the oscillator and the pentode section as the driver. Something to try in the future.
I did a little research into low power, tube, CW transmitters. Here is what I found.
EICO 723 - 6CL6 OSC - 6DQ6 AMP - 60 WATTS
HEATHKIT HX11 - 6CL6 OSC - 6DQ6 AMP (WITH NEUTRALIZATION) - 50 WATTS
HEATHKIT HW16 - 6CL6 OSC - 6CL6 DRIVER - 6GE5 AMP
The HX11 looks like a DX-20 with neutralized final.
The DX-20 clone is a two tube all-band CW transmitter running about 30 watts input. I call it a DX-20 clone because it is very similar to the Heathkit DX-20. However, it is built around a project from the 1971 Handbook titled 'A FIVE-BAND FIFTY WATTER'.
The biggest difference between this project, the DX-20, and the project in the handbook, is size. The handbook project measures 12 x 7 x 9 inches, about the size of a DX-20. This project measures 6 x 6 x 8 inches.
Other differences are improvements over some of the more anoying non-features of the Heathkit DX-20. A decent meter was used. A toggle switch is used to select between GRID and PLATE current. All meter wiring is done with shielded cable.
A rotary switch was used to provide a combination on/off and function control. Functions include OFF, TUNE, SPOT, QSK, and OPR. The off function is self explanatory. In the TUNE position, the oscillator and amplifier are keyed but the amplifier screen is grounded to reduce plate current. In the SPOT position, only the oscillator is actived by the keying line. In the QSK position, both the amplifier and oscillator are keyed simultaneously to permit full break-in operation. In the OPR position, the oscillator runs continuously and only the amplifier is keyed.
This is not an excercise in miniaturization. The small size resulted from use of materials available. The entire case and chassis (except for top, bottom, back and partial side plates.) was made from a single 3.5 x 6 x 8 inch Bud minibox.
The meter is a 0-1ma surplus item of unknown origin. The meter switch is the top toggle switch.
The larger knob below the meter switch is the plate tuning. The output circuit bandswitch is just to the right of the plate tuning knob.
The solid black knob below the meter is the grid peaking control.
The knob to the lower left is the function switch. It turns the transmitter on and selects operation type.
The switch located between and just above the crytal socket and key jack is the bandswitch for the amplifier input.
The larger knob in the lower right is the antenna loading control.
Side view showing recessed 6BQ6 socket and antenna loading capacitor
Additional top view with ruler to give an indication of size.
Top view showing transformer, 12BY7, 6DQ6, and meter.
Top view showing transformer, 12BY7, 6DQ6, and meter with bandswitch, tank coil, and shield intalled.
Seperate band switches for input and output are used. The design could be modified to include a single bandswitch control but this would have resulted in a much more complicated switching arrangement and probably have required a larger enclosure. As it stands each bandswitch is already multiwafer. Multiple sections are required on the plate bandswitch to add capacitance to both the antenna loading and plate tuning capacitors when used on 80 meters.
The 'all-band' feature is made possible by tuning both the input and output of the amplifier. This allows the amplifier to work straight through on all bands. Any frequency multiplication takes place at the 12BY7.
Eighty meter crystals can be used on 80 and 40 meters. Forty meter crystals are used on 7, 14, 21, and 28 mHz bands. WARC bands can also be covered with the appropriate crystals. The only WARC band I had any interest in was 30 meters and this unit was built to accomodate that band by using 10mHz crystals.
The crystal socket is a standard FT-243 style connector. I made an adapter plug which will also allow the use of HC-6 style crystals which also work well in this circuit. My 'standard' crystal, used for checkout is a color burst crystal that was mounted into a FT-243 holder by hogging out the insides of the holder with a Dremel tool until the much smaller color burst crystal fit.
The power transformer is unique in that it has a multi-tapped HV secondary. It delivers both 270-0-270 vac and 160-0-160 vac. With a capacitor input filter, the 270-0-270 connection delivers 360vdc which drops to 350vdc under a 150ma load. The lower voltage comes out at 175vdc and drops to 170vdc under load. We need at least 460vdc to push the 6DQ6 to 50 watts input. Hence, this rig is rated at only 30 watts input. The 160-0-160 connection provides power for the oscillator and 6BQ6 screen.
Both supplies are full wave using two 1N4007 diodes per supply. I had considered using a bridge connection and obtaining the lower voltage from the center tap of the transformer, but that increased both the low and high voltages to values that were too high for comfort. It would have required extra filter capacitors, seriesed and equalized to run the higher voltage and I did not want to use dropping resistors because of the heat they create. There is enough room for the added components and the power transformer appears to have a decent amount of reserve. A possible future upgrade might be to go with the higher voltage bridge configuration and use a 6146 in place of the 6DQ6. Doing so should bring the input power up to over 75 watts. However, I am pretty sure that the transformer I am using is not going to be able to deliver that kind of power.
Below are some pictures of the rig in the early design stages. They show how the minibox was utilized. The power transformer available just fit into the U of the box and was the deciding factor. Initially, partial side panel of wood were considered. This idea was abandoned in favor of metal side panels to obtain more room and better shielding.
Since only one WARC band (30 meters) was implemented, a six position band switch was adequate. Both band switches are dual section but both sections are on a single wafer. Implementation of all the WARC bands would require a switch with more than six positions and at least two wafers.
The function switch uses two dual-section, six-position, wafers. Only five of the positions are needed. One of the dual section wafers is a standard wafer, just like the ones used on the other switches. The second dual section wafer has a make-before-break action. This wafer has a wider wiper which makes contact between two adjacent teminals before releasing contact with the first terminal on its way to making contact with the second terminal. This action is required for reasonable implentation of the function switch as an on/off switch. The make-before-break feature ensures that power is not interrupted when fuctions are changed.
One of the make-before-break sections is used for primary power control. The other make-before-break section (on the same wafer) is used to switch the amplifier cathode function. DO NOT use this second make-before-break section to switch screen voltage on the amplifier. Doing so may place a momentray short to ground on the low voltage supply.
The output tank coil can be constructed from a single length of 3018 B&W miniductor stock. The finished coil is 31-1/2 turns of #16 wire wound 8 turns per inch at 1-1/4 inch diameter with taps at 2-3/4, 6-3/4, 10-3/4, and 22-3/4 turns from the C3 end of the coil. The tap for 30 meters is 16-3/4 turns from the C3 end of the coil.
If you can find the miniductor stock at a reasonable price, use it. Otherwise, make your own.
I used a 1/4 inch thick, 1-1/2 inch wide, 5 inch long, length of plexiglas as a coil form. Grooves were cut 1/8th inch deep at 1/8th inch intervals into the sides of the form. Masking tape was used to mark the location of the grooves and the cuts were made on a table saw using a blade thickness close to the wire diameter. With care, a hand saw with proper blade thickness would also work. Thinner plexiglas or polycarbonate would also work.
A six inch length of 3/4 PVC pipe was cut in half lengthwise. The pipe halves were attached to the center of the plexiglas form using masking tape and the coil was hand wound onto the form. The grooves space the turns at 8 tpi and the pipe halves ensure reasonably round turns at 1-1/4 inch diameter. A total of 32 turns were wound. The wire was bent around the ends of the form to keep it in place. Shoe Goo adhesive was used to glue two end turns to the form at each end.
The wire was salvaged from a ten-foot length of romex house wire. After stripping off the insulation, one end of the wire was clamped in a vise. The free end of the wire was wrapped around a screwdriver and the wire was pulled straight by yanking it against the end secured by the vise.
After the glue has set, the PVC pipe sections can be removed by pulling them from the coil. The resulting coil is not as pretty as miniductor stock but it works just as well.
The coil taps are made with #16 bare copper wire just like the coil itself. The ends of the tapping wire are flattened with a hammer, bent to form a U, and pushed between the coil turns requiring the tap. The U is then pulled forward to capture the coil winding at the tap location, and crimped in place with pliers. After crimping, the tap is soldered. Take care not to short turns. Push/bend the coil turns away from the tapped locations if needed.
The ten meter tap may need to be moved closer to the plate tuning capacitor depending on the length of the wire connection it to the band switch. All other taps worked fine to give a dip at the center position of the 140pf plate tuning capacitor.
Due to size limitations, the antenna loading capacitor had to be mounted below the main chassis surface. This requires two feed-throughs in the chassis to make the connections to the coil and bandswitch. Feedthroughs were made from very thick teflon sleeving that fit #16 bare copper wire. A quarter inch length of sleeving was cut and pressed into a hole in the chassis. The chassis hole size was carefully chosen for a press fit with the outside diameter of the teflon sleeving. A similar feedthrough was used to bring HV to the bottom of the plate RF choke.
The output tank circuit is located on top of the chassis. The input tank circuit is located below the chassis. This provides good seperation but there is still a chance of input to output coupling between the 12BY7 and 6DQ6 on top of the chassis. To eliminate any possibility of coupling, a shield was installed on top of the chassis between the oscillator and amplifier tubes.
Likewise, a shield was built and installed beneath the chassis to prevent coupling between the output lead and input tuned circuit. The rotor of the antenna loading capacitor is already at ground and needs no shielding.
I decided to include neutralization of the final as in the original Five-Band 50 Watter article. It is interesting to note that the Heathkit DX-20 circuit does not use neutralization on the 6DQ6.
I have not yet tried the 1 inch wide strip of aluminum for a neutralization capacitor as in the article in the handbook. I took the easy way out and wrapped some aluminum foil around the tube base, trimmed the foil, and made electrical connection to the foil by wrapping and twisting a length of hookup wire around the foil. Electrical tape keeps this mess in place. YES, NEUTRALIZATION IS REQUIRED!!!! I have no idea how the Heathkit design of the DX-20 worked without neutralization. Those guys must be a lot smarter than me.
I had no end of problems until I neutralized the final. I encountered plate currents higher than 200ma that dropped the HV to 250 volts. I could light up a 75 watt light bulb nicely under those conditions with or without a crystal in the socket. To obtain a reference I rigged up a 3.5k power resistor to give me a 100ma load on the HV at 350 vdc. Under those conditions, the 350 vdc dropped to 340 vdc. Two 3.5k power resistors in parallel for a 200ma load dropped the 350 vdc to 315 vdc. With the aluminum foil neutralizing capacitor, the plate current is only 125ma with the amp fully loaded into a 50 ohm load on 40 meters. Now the output goes away when the crystal is pulled.
NEUTRALIZATION
There are two common ways of adjusting the neutralizing capacitor. If the amp draws grid current, neutralization can be accomplished by monitoring the grid current, while the plate tuning capacitor is moved through its dip. During this adjustment plate voltage is applied to the tube but the screen grid is held at ground. This condition is accomplished by moving the function switch to the TUNE position.
If the amplifier does not draw grid current, neutralization is accomplished by using an RF output indicator on the output tank circuit and the cap is adjusted for minimum feedthrough.
Using the grid current method, the objective is to find a setting of the neutralizing capacitor that results in the least amount of grid current change when the plate tuning capacitor is tuned through resonance. If the grid current increases as the plate tank is tuned to the high side of resonance (less capacitance on the plate tuning capacitor), the neutralizing capacitance is too small. If the grid current increases as the plate tank is tuned to the low side of resonance (more capacitance on the plate tuning capacitor), the neutralizing capacitance is too large. Dont forget to attach a dummy load to the output of the transmitter while doing the adjustment.
There will still be a grid current flicker when the plate tuning capacitor is adjusted for a dip in plate current, regardless of how carefully the neutralization is adjusted. The sign of proper neutralization is reduction in the amount of grid current flicker.
For this procedure to work, you must be able to get a dip in plate current when the plate tuning capacitor is adjusted. That means the tank circuit needs to work at the frequency being used for the neutralization adjustment.
The neutralization adjustment becomes more critical at higher frequencies. Good performance is almost guaranteed when the rig shows proper neutralization at ten meters. However, it is easier to 'get into the ballpark' at a lower frequency. Then use ten meters for final tweaking.
The pictures that follow show the project near completion. All that is needed is top, bottom, back, and side panels, and fuse and power connector. The power supply section turned out to be the easiest to build. The low voltage section uses two 22mfd/450 volt caps that were selected for their small size. Typical capacitor input filter with a surplus 85 ohm power resistor connecting the two capacitors.
The high voltage section is filtered by two 420mfd/300volt photoflash capacitors. These were also selected for their small size. Photoflash capacitors are not the best but they work well in this case because they have twice the capacitance rating required and are being run at only half their maximum voltage. The signal produced by the transmitter appears to be clean.
The tuned circuit between oscillator output and amplifier input was a real pain to fit and get working. The basic idea here is to wire a bunch of variable inductors in series, tune them with a capacitor to ground, and bandswitch by shorting out inductors. I used 1/4 inch diameter, slug tuned forms that were half an inch long. The bottom connections on the forms were not insulated but soldered to the mounting stud. Attempts to unsolder would have probably destroyed the form so they were mounted to a section of double sided copper clad material from which enough copper had been removed to provide an insulated mounting for the six coil forms.
The test setup for tuning the input circuit is shown below. As additional forms were added, their frequency was checked with a grid dip meter and the tuning range of the peaking capacitor was verified. Everything went smoothly until the testing of the 80 meter position. The coil forms were not really large enough to wind a coil that would deliver the 58uhy needed. So the extra position on the band switch was used to add 310pfd capacitance across the tuning capacitor. Enough inductance was added to shift the tuning range into the lower end of the 80 meter band. This only took about 7 turns on an additional 1/4 inch form. All coil wire was #26 enamel salvaged from the yoke assembly of dead computer monitor.
All meter leads were run as shielded cable, bundled, and brought up to the meter area through a grometed hole in the chassis.
The plain, black panel was masked and painted almond to highlight the meter and control knob areas. This way all the controls and meter are clearly visible. There are no adjustments hiding in the black of an all black panel.
As the project advanced to completion, standoffs were used as much as possible and attached to existing bolts to eliminate the need to drill additional holes. Although tempting, 'manhattan' style construction was shunned in favor of conventional construction using appropriate hardware and ensuring that each component was properly supported at both ends. Liberal use of lock washers should help ensure this rig stays together while in use.
The key shaping network is distributed across the bottom part of the front panel. The 51 ohm resistor is connected between the hot side of the key jack and a teflon standoff that is mounted to the same bolt holding the crystal socket. The RF filter is wired across the key jack at the key jack location.
The cathode RF choke for the amplifier was mounted to a small terminal strip. Additional teflon standoffs were mounted to the bolts holding the power supply terminal strips. One of these standoffs serves as the screen supply connection the other is the grid circuit connection.
The capacitor coupling the input tuned circuit to pin 5 of the final amplifier is mounted to a small terminal strip which is held in place by one of the bolts of the function switch. A short length of shielded cable is used to complete the connection between the coupling capacitor and pin 5 of the tube. Only the shield at the capacitor end is grounded.
This project could be drastically simplified by building the rig as a single band CW transmitter with less operating flexibility. This is one of the advantages of homebrewing. You can change it to suit without worry about effects of modifications on a 'factory' rig. Most of the features provided by the function switch are nice but not necessary. Monitoring plate current ONLY is another simplification. In fact, the meter could be removed entirely by using a light bulb RF output indicator. The tuned circuit between the oscillator and amplifier could be removed and the rig could still be useful on 80 and 40 meters as long as 80 meter crystals were used on 80, and 40 meter crystals were used on 40. The problem with such simplification is less usefulness from a rig that still requires a sizable investment in possibly expensive parts. With a little more effort, you get more bang for the buck.
The 6DQ6 is a good choice in this design because it delivers a respectable amount of power at a relatively low voltage. If higher voltages were available (700 for plate and 250 for screen ), the 6146 might be a better choice because it would increase the power input up to as much as 90 watts. This is assuming that the power transformer would be capable of supplying that amount of power. I doubt such a transformer would fit into the space that has been alloted in this project.
The schematic can be found here. There were no changes needed from the original 5 band schematic. The last feature added was an antenna changeover relay which is not shown on the schematic but it is a trial addition. The parts list can be found here.
-
HOME
OFF TOPIC BLOG
Amplifiers, audio
Amplifiers, RF
Antennas
Boatanchors
CW Keys
DRAKE radios
Equipment for Sale
Ham Radio Misc
Parts for Sale
| www.flickr.com |