Rob's web

Home - Techniek - Electronica - Radiotechniek - Radio amateur bladen - QST - More stations per megacycle at two meters

An all-push-pu11 100 watt output crystal-controlled transmitter.

Observant amateurs in the crowded areas have noticed that since the shift from 2A to 2 meters Old Man QRM has been more bothersome than ever. Amateurs using superregenerative receivers and modulated oscillators find that fewer signals can be accommodated by the 4 Mc. allotted to their use in the new band, because frequency modulation and receiver selectivity are matters of percentage of the operating frequency. Consequently, increasing the operating frequency correspondingly increases the frequency modulation of the modulated oscillator and the bandwidth of the receiver.

Hams are prone to think of occupancy of a band in terms of the amount of QRM heard. If a dozen R9 + signals are present at once on the 2-meter band, the band appears to be crowded and the impression is that there are lots of stations operating. By way of comparison, try listening on the 6-meter band with a sharp receiver when a similar-dozen R9+ signals are present. One gets the impression that the band is practically unoccupied, because signals are far apart with lots of empty space around them. A comparison of this sort will drive home the fact that the 2-meter hams are really making very poor use of the 4 Mc. allotted to them.

The obvious step in combatting this QRM situation is to use a sharper receiver. The superheterodyne-superregenerative combinations recently described in QST(1),(2) are not beyond the scope of the average amateur and represent a definite improvement. However, the modulated oscillator has difficulty getting through even these relatively-broad receivers, and if that type of transmitter is about the only one in wide use a ham will hesitate to build the sharper receiver. Since a sharp transmitter will come through and even show up to advantage on a broad superregenerative receiver, it appears that the first move could be to sharpen up the transmitters. Further, as pointed out in April QST(3), a power amplifier is much more efficient than an oscillator so an increase in signal strength can be expected.

Transmitter stabilization involves two considerations: first, the variation of carrier frequency with modulation (frequency modulation); and second, frequency drift during and between transmissions. Of the two, the second is the most difficult problem in the simple v.f.o. type of 2 meter transmitter. Several two-stage m.o.p.a.s have been heard around Boston recently, and compared to the modulated oscillators they represent a very definite improvement when received on the superhet-superregen receiver, but the frequency drift is such that constant retuning of the receiver is required while listening to them. During long transmissions this drift ranges from 250 to 600 kc. If a transmitter drifts 500 kc., for example, it will cause QRM at one time or another to all stations located in the ½ Mc. strip across which it drifts. In this respect, excessive frequency drift is almost as objectionable as excessive frequency modulation.

Photo 1
The 144 Mc. tanks and mountings are shown in this rear view of the transmitter. An outstanding feature of the unit is the simple, straightforward layout.

Although frequency modulation can be overcome by using several stages, and frequency drift reduced by swamping and compensating methods, we decided to do it the easy way and use crystal control. As soon as a crystal is slipped into its socket at the front end of the transmitter, you can forget about f.m. and drift; the problem is simply to get the required drive for the final with maximum economy in parts and power. Of course, a low-drift crystal should be used; drift is still a consideration with the third-harmonic or high-frequency type of crystal.

Another feature of crystal control is the opportunity it affords to operate very close to the edge of the band. Two crystals in use at W1CTW give frequencies of 144.04 Mc. and 147.98 Mc., and operating this close to the band limits gives the broad receiver a break because only about half of the carrier "blank" extends inside the band.

The circuit

In planning the circuit, it was decided to use simple, conventional procedure, with trick circuits and methods ruled out. It was desired to multiply as rapidly as possible while still having drive to spare for each stage and operating all tubes well within their maximum limits. If the final stage is to be push-pull, doubling or quadrupling in the exciter requires changing from a single- to a double-ended stage, introducing problems of unbalance. We decided to start out with and use push-pull circuits all the way; this proved to be a very happy choice and produced an exciter that handles as easily as any lower. frequency job we have ever seen. Also, a push-pull tripler from 6 to 2 meters is by far the best way to end up a 2-meter exciter; the efficiency is surprisingly high, running about 30 to 35 per cent if properly driven. By using three push-pull stages, we multiplied the crystal frequency by 27. This called for a crystal frequency between approximately 5.33 and 5.48 Mc., a range in which low-drift cuts can be obtained and which, incidentally, are quite plentiful as war-surplus stock.

Preliminary layout

The question of a tube - or tubes - to be used in the final stage was given first consideration. A carrier output of at least 40 or 50 watts was considered desirable and 100 watts about the maximum necessary, since it was felt that at the present time much higher power output than this would not give sufficiently-improved results to justify the increased cost of the power-supply and modulator equipment. Tetrodes were considered preferable to triodes because they require less driving power and we hoped that it would not be necessary to neutralize them. Preliminary checks had shown that most of the common tubes, such as the 6V6 or 6L6, were worthless at 144 Mc., so the field narrowed down to three transmitting types - the 815, 832, and 829.

The 815 has the advantage of low cost and is capable of giving a 40 watt carrier. The 832, while it operates beautifully at these frequencies, is limited as to output and is quite expensive. The 829 in both the A and B versions is capable of considerably higher output than either of the other two types. While it is quite expensive if bought new, it seems to be quite plentiful since the end of the war. The 829-B is especially desirable as it will take relatively high inputs and is the sweetest little tube at 150 megacycles that the writers have ever seen. With this tube we have obtained a 100 watt carrier with about 135 watts input.

Two final-amplifier stages were built, and either can be used, depending on whether an 829 or 815 is available. The 829 is more stable than the 815, is relatively easier to drive and will give greater output, but the tubes available are apt to vary in characteristics and be more erratic in operation. The 815 gives a 40 watt carrier with 60-watts input.

Photo 2
A plan view of the 144. Mc. crystal-controlled transmitter. The set is push-pull throughout, with all frequency multipliers operating as triplera. The crystal oscillator, using 6AG7s, is at the right, the following stages (progressing to the left) being the 6L6Gs as first tripler, 815 as second tripler, and 829 final amplifier.

Fig 1
Fig. 1. Circuit diagram of the 144 Mc. crystal-controlled transmitter.

Partlist fig. 1
C1,C25 pF ceramic, 500 volts.
C3,C4100 pF mica, 500 volts.
C5,C9,C13,C15.001 µF mica, 500 volts.
C6,C1050 pF per section variable (National STD-50).
C7,C8,C11,C12250 pF mica, 500 volts.
C14,C1610 pF variable (National UMA-10; See text).
R1,R2250 kΩ, 35 watt.
R315 kΩ, 2 watts.
R4,R8,R12250 Ω, 2 watts.
R5,R6,R9,R10100 kΩ, ½ watt.
R730 kΩ, 2 watts.
R1110 to 20 kΩ, 4 watts.
R135 kΩ, 1 watt.
R145 kΩ, 5 watts.
R15100 kΩ, 2 watts.
R165 to 10 kΩ, 5 or 10 watts.
R17Resistance equal to grid-meter resistance, if 0-10 milliammeter is used.
L1,L2750 µH (See text).
L320 turns No. 22 on ½ inch dia. form (National XR-50), closewound, center-tapped.
L47 turns No. 16 on ½ inch dia. form, length 5/8 inch, centertapped.
L5,L7See Fig. 2.
L6See text.
L8See text.
M10-10 milliammeter (0-20 mA. may be used, in which case R17 is not required).
M20-300 milliammeter.
RFC40 turns No. 26 closewound on ¼ inch dia. form.
S1D.p.d.t. toggle switch.
S2S.p.s.t. toggle switch.

The exciter

In choosing the 815 for the driver stage, a, push-pull tripler, we decided not to be fooled by the low driving-power ratings given in the tube handbooks. We believed that we should figure on about five or six watts output from the driver, especially in view of the type of circuit to be used, for the purpose of avoiding neutralizing the final. In the end, we found that our conservatism was well founded because, although we have ample drive for the final stage, there is not too much leeway.

It is very important that the 815 tripler be driven hard. A tripler stage requires more excitation than the same tube will require operating as a straight-through amplifier, and a high value of grid bias is essential. As shown in the circuit diagram, Fig. 1, we use 100,000 ohm resistors on each grid. About 150 volts of bias should be developed which calls for 1½ mA per grid.

About two months of experimental work was done in our spare time before we were thoroughly satisfied with the exciter stages. Various tube combinations were used, involving 6AC7, 6AG7, 6V6, 6F6, 6L6, and other types. The experi mental chassis now has so many holes in it that it looks like a sieve!

The crystal-oscillator circuit is similar to the one used in the 2-meter converter described in May QST.(2) The 6AC7s are excellent oscillators and multipliers in this circuit, but it was found that enough third-harmonic output could not be obtained to drive the following stage without exceeding the maximum ratings of the tubes, so 6AG7s, which have similar characteristics and which perform equally well, were substituted. The 6AG7s will take about three times as much input as the 6AC7s and are capable of providing full excitation to the second multiplier stage.

This crystal-oscillator circuit is a push-pull grid-plate arrangement with electron-coupled output. The screen grids are grounded for r.f. while a fixed-frequency coil-and-condenser combination is connected between each cathode and ground. The coil and condenser values are not particularly critical, but should be adjusted roughly for optimum oscillation. A 2.5 mH r.f. choke could be used for the coils, although it might be necessary to remove one or two pies. 5 pF ceramic condensers are connected directly from grid to cathode to increase the strength of oscillation, particularly with sluggish crystals.

The two tubes are made to oscillate in push-pull by the simple expedient of connecting the crystal from grid to grid. The plate circuit will then triple in push-pull by tuning the plate tank to the third harmonic. One advantage of this circuit is that the crystal will always oscillate regardless of the tuning of any of the controls of the transmitter. Notice that a high value of grid resistor is used on the control grids of the 6AG7s. Quarter-megohm resistors afford strong oscillation and particularly enhance the strength of the harmonics. In operation, the oscillator grid current should develop about 35 to 50 volts per grid. The oscillator should drive the following 6L6 grids to about 125-150 volts of bias.

The second stage uses a pair of 6L6s as a push-pull tripler. Smaller tubes were tried but were found to be inadequate. The rotors of the split-stator tuning condensers used in the first- and second-tripler plate circuits are grounded, and decoupling resistors are used for connecting the B+ to the center of the coils to avoid a second ground connection to the tank. The drive to the following stage should be adjusted by varying the screen voltage rather than the plate voltage, but the screen dissipation should not be allowed to exceed the maximum value specified by the manufacturer. The power supply for the exciter should deliver between 300 and 325 volts at 125 to 150 mA.

The tank circuits

The tanks in the plate circuits of the first and second stages use split-stator receiving-type condensers and coils wound on ungrooved National XR-50 coil forms. These coil forms were tried originally with the iron plugs furnished with them for tuning purposes, using fixed ceramic capacitors in place of the variable condensers. The plugs unbalanced the excitation to the following grids when it was necessary to move them away from the center of the coil in tuning, an unbalance which would not occur with slug tuning of single-ended tanks. The variable condensers are mounted above and the coils below the chassis, the two tank leads passing through holes in the chassis.

The 2-meter tanks for the last tripler and final stage are identical and represent the end result of quite a process of evolution, starting from ordinary condenser-coil combinations and linear tanks. The condenser-coil tanks were not as efficient as the linear circuits, but the latter were too bulky and were not easy to tune from the front panel. Credit for the final form should be given to Marty Oxman, W1NYU, who worked it out using material which was readily available. It is made of 1/16-inch copper strip, h-inch wide, as shown in Fig. 2, and uses a National UMA-10 condenser for tuning. This condenser is easily removed from its Isolantite mounting plate and remounted on the ends of the copper strip, after the required holes have been drilled in the latter. A National GS-1 Isolantite insulator is fastened directly below the rotor to make the unit rigid. The inside of the bend at the bottom should be dimensioned to correspond with the length of this insulator. A flexible coupling is mounted on the end of the rotor shaft, which is turned by a knob on the front panel through a 3.-inch polystyrene rod. The plates of the tubes are connected by thin copper strip to the upper ends of the tank. This arrangement has proved to be very satisfactory both mechanically and electrically, covering the 2-meter band with plenty of leeway.

The grid tank is made of 1/16 inch copper strip, ¼ inch wide. The bend is made at the bottom to give exactly the same width as the plate tank to which it is coupled. The top ends of this strip are cut back to produce two leads which can be bent around so that they can be rigidly mounted directly on the two grid contacts on the tube socket. The tube socket is mounted with the large hole (cathode prong) toward the top. The loop is placed parallel to the tripler tank to give close coupling. This coil should be made as short as possible while still delivering adequate excitation to the final amplifier; the coil shown in the transmitter is slightly less than two inches long, exclusive of the lead length to the grids. It may be necessary to make two or three of these coils before the best size is obtained, a procedure which is to be preferred to using a larger coil and looser coupling.

Photo 3
There is relatively little below-chassis wiring, as this photograph shows. Note the oscillator and first-tripler tank coils mounted parallel to the chassis in the section at the left.

Photo 4
The 144-Mc. tank, using a copper-strip inductance on which a standard midget condenser is mounted for tuning. The cylindrical ceramic stand-off provides mechanical support for the assembly.

The reason for using the small grid coil in place of a resonant one is to avoid neutralizing the final amplifier. It is commonly known that if the grid circuit of an amplifier is tuned to a higher frequency than the plate circuit, oscillation will not occur; this is demonstrated by the fact that a triode crystal oscillator will only oscillate if the plate tank is tuned to a frequency higher than the crystal frequency. By making the grid coupling loop as small as possible the grid circuit is detuned to a higher frequency than the plate circuit, thereby preventing self-oscillation. The result is that the 829 amplifier shows no appreciable need for neutralization. The 815 final stage, which required a larger grid coil, showed a slight degree of reaction on the grid meter when the plate tank was tuned through resonance without plate or screen voltage. However, it would not oscillate, and was very stable under modulation when coupled to an antenna.

The layout

The entire transmitter is mounted on an aluminum chassis which measures 17 × 8 × 2 inches. It is intended for rack mounting, using a standard panel height of 8¾ inches. The plate tank (2 meters) of the 815 tripler stage is placed in the exact middle of the chassis. The components of the three tripler stages are then laid out on one side of the chassis, placing the parts so as to permit the shortest possible leads. The tuning knobs for the two tripler-tuning condensers are balanced on the opposite side of the panel by two switches. The final-amplifier tube is mounted horizontally on an aluminum angle 6-inches high by 7%-inches wide. On the transmitter shown, the bottom part of this piece is extended horizontally far enough to enable it to contain the entire amplifier stage with its connections. This was done so that more than one amplifier unit could be mounted in place for trying out various tubes and set-ups. If only one final stage is to be used, this expedient is not necessary and the bottom flange need be only about an inch wide. In that case, the best position for the final plate tank should be determined after the tube is in place. The mounting holes in the bottom flange of the aluminum angle can be slotted to allow some movement of the angle for varying the coupling between the last,-tripler plate tank and the final-amplifier grid coil. A small antenna-coupling loop is mounted at the end of the chassis next to the amplifier tank.

Fig 2
Fig. 2. Dimensional drawing of the 144 Mc. tank inductance before bending. The material is ½ inch copper strip.

The meter switch, S1, is so wired that in one position it measures the combined grid currents of one section of each of the second and third multiplier stages, and in the other position it measures the grid current to the final amplifier. The other switch, S2, is arranged to add 100,000 ohms to the screen-voltage dropping resistance for the final amplifier. The additional resistance prevents excessive screen and plate current (or dissipation) while tuning up.

On the back of the chassis is a four-prong socket to which are supplied (1) the hot 6.3 volt filament connection, (2) the grounded-filament and B - connection, (3) 300 to 325 volt B + for the exciter stages, and (4) 450 to 600 volt B+ for the final amplifier. Two-terminal strips also located on the back provide for external switching and for connecting to the secondary of a modulation transformer. Wiring is relatively straightforward, and no comment should be necessary except for the usual recommendation to use short, direct leads.

Photo 5
An alternative amplifier unit using an 815. The circuit and general construction are the same as for the 829 unit shown in the transmitter assembly.

Tuning up

If the coils have been wound correctly, no trouble should be encountered in tuning up the exciter. The B+ lead to the final amplifier should be disconnected, or else the screen-supply switch S2 should be thrown to the low-voltage position When voltage is applied to the exciter, tuning the first stage should produce 1.25 to 1.5 mA with the grid meter measuring the current in the exciter stages. Tuning the second stage should develop another 1.5 mA for a total reading of about 3 mA. The meter should then be switched to the final-amplifier grid, in which position it is shunted to read 20 mA, full scale. Tuning the third-tripler tank to resonance should produce about 12 mA of grid current under load when the grid-coil coupling is correct. At this point the B+ to the final can be applied, if it was previously disconnected, and some antenna loading applied to the final stage, making sure that the screen switch is in the low-voltage position. After tuning to resonance, the switch can be thrown to apply full screen voltage and then the antenna loading can be adjusted to make the plate current rise to 200 or 250 ma. The transmitter is then ready for modulation, which requires no special comment since this problem is not peculiar to two meters.

With a transmitter of this type you can go on the air with a signal which is beyond reproach, and with efficiency and general ease of handling equalling that of transmitters of similar power on the lower-frequency bands. We believe you will find the results well worth the work involved in putting such a rig together.


  1. Goodman, "A four-tube superheterodyne for 144 Mc.," QST, November, 1945.
  2. Hadlock, "A two-meter crystal-controlled converter," QST, May, 1946.
  3. Grammer, " Stabilizing the 144-Mc. transmitter," QST April, 1946.

Calvin F. Hadlock, W1CTW
Ralph S. Hawkins, W1OEX.