An experimenter's notes... but not every detail


Copyright © 1995 by John Seboldt. Permission granted for further distribution for individual use of radio hobbyists only.
A few changes 6/10/2001

This project uses the existing tuned circuits, physical structure, and some parts and wiring of the exciter from the old Collins military T-368 transmitter as the basis of a solid-state VFO/multiplier for most of the HF region of interest to amateurs.

[check J105/P105]

As supplied, it has a 1.5-3 MHz PTO, a mechanical digital readout, and doubler stages to cover 1.5-3, 3-6, 6-12, and 12-24 MHz (they say 12-20, but the dial reads to 24; probably reflects the ratings of the companion PA). It measures about 10" wide by 9" high by 11" deep, so it's not small, and as I received it there was no enclosure for the whole assembly -- no problem for less critical applications like HF, but probably best to add shielding (I haven't yet for HF QRP!). It's built like a tank, and the tuned circuits in the multipliers track the PTO in typically elegant Collins fashion.

I got it from Fair Radio Sales, a renowned surplus dealer, about 5 years ago without the output tube (the 6000, a rare beast nowadays). It cost $35 then; one of my correspondents informs me that they now run $45 without the output tube. I first used it as an improvement over a Heathkit HG-10 VFO with a DX-60 transmitter, and it was remarkably stable even in its tube form.

Once transistorized, you get an incredibly stable VFO for the home station that covers a wide range, complete with receiver and transmitter offset pots on the front panel. It's not a unit for backpacking, because of its solid, heavy construction. But that weight and bulk is part of what assures mechanical stability. The sheer size of the sealed PTO assembly, plus the fact that the components were designed for the higher RF currents of the tube circuit, means that once you solid state the thing, you have rock-like stability. There's even a thermostat-controlled heater you can hook up around the PTO tuned circuits. True, the unit tunes pretty fast on the higher bands, but the gearing is so good that there is no backlash. The offset pots can help you zero in if your hand isn't too steady.

My sources for circuit concepts and building blocks:

ARRL Handbook

QRP Classics (ARRL)

W1FB's Design Notebook (ARRL)

W1FB's QRP Notebook (ARRL)

Solid-State Design for the Radio Amateur (ARRL)

Peter Anderson, KC1HR, "Transistorizing Surplus VFOs", QST, Feb. 1989, page 45-46 (in Hints and Kinks)

T-368 manual pages and schematic relating to exciter. On ordering, I asked only for the pages from the T-368 manual related to the exciter, and Fair Radio quoted me $10 at the time.


Removing the PTO takes some care -- you have to loosen not only the mounting screws, but the hex set screws on the PTO shaft gear. I suggest you set things at 3 MHz at the top end of the band, so you know where to reset things. All the electrical connections unplug, and all these wires are re-used, also the feedthrough capacitors attached to an angled bracket under the PTO. All the circuitry outside the tank circuit is on a small sub-chassis under the tube sockets.

The original approach - transistorize existing circuit

I started with the QST article, which suggests using 2 40673 dual-gate MOSFETs. (The NTE222 is a currently-available substitute for the discontinued 40673 -- but they're about $5 each locally!). Read this for some precautions about turning the PTO too far beyond its range, and damaging the internal slugs. I think it will go more than 50 kHz beyond, so you have some slop for recalibration. Also, the author warns not to open the PTO sealed inductor can if you can avoid it -- the hermetic seal is one key to super-stability.

Having used various VFO circuits, I decided to save a few bucks, and used a plastic VHF JFET for the oscillator (NTE312 or other similar device). I took a 100 ohm resistor from regulated V+ (after the existing RF choke) to drain; hooked drain to the original plate pin; hooked the gate to the original tube grid connection; and hooked the source to the original cathode connection. A 1N914 diode goes from gate to ground (cathode grounded). That author had put the diode from gate 1 to source, and this meant no oscillation for me without some kind of "kick", like temporarily grounding the gate! When the diode went to ground like all other oscillator circuits I've seen, all is well.

The tuning diode circuit provides an adjustment range for RIT/offset. These values gave me a tuning range of 7 kHz at the 3 MHz end of the PTO when I tested it before re-installation. However, I didn't discover till I put things together that the tuning range at the bottom end (near the 3.5 and 7 MHz ham bands) was substantially lower! At 80 meters, I have barely enough to get an 800 Hz CW offset. At 40 meters, I even have only about +- 2 kHz.

I later put in an NTE618 tuning diode, designed for AM radio tuning service (440 pF at 1 volt!) Of course, I reduced the size of the .01 uF capacitor iuntil the tuning range was reasonable. I think it was about 100 pF, can't remember :-). Tuning range is about +/- 2 kHz at 1.75 MHz, multiplying to +/- 16 kHz at the highest band.. My 10-turn pots make even the widest range quite manageable. The stability is SOMEWHAT degraded by the tuning diode arrangement, but I still got less than 10 Hz drift overnight relative to WWV at 10 MHz.

I used a regulated 6 volt bus for the oscillator (and RIT bias -- see below), coming from a 3-terminal regulator mounted on the back of the multiplier. My regulator was an out-of-tolerance 7805; I'll bet 5 volts from a good 7805 would be fine. T

The buffer is almost identical to the QST article above, except for output coupling and operation from a separate power bus. Here the original plate pin serves only as a tie point, the 10K plate resistor and 100 pF coupling capacitor being removed. Handle the MOSFET with all due electrostatic precautions -- wrist strap, etc.

4 wires that enter the PTO circuit from the power connector are rewired as shown; the other two are for the VFO oven heater. You should not find this last option necessary, unless you are operating in the Arctic! At the feedthrough capacitors under the PTO, you have convenient points to solder your connections to the rest of the assembly.

Re-installation of the PTO IS a bit tricky because of the split gear that eliminates backlash. You have to turn the one gear so the springs are compressed, hold them in place with a screwdriver until you get them meshed with the gearing from the dial.


I just ripped out all the existing parts (in the tube compartment) and built the oscillator/buffer more cleanly, according to the homebrew books..  The tuned circuit components to the left, plus the coupling capacitor, are the ones in the sealed can, and three pins come into the tube compartment for ground, cathode/source, and grid/gate. Don't know exactly why I did this, other than to have less clutter in there.


I was on my own on the multiplier stages. It was possible to get some results by simply plugging in JFETs in the appropriate places, but little anomalies would creep in from the long wiring, old-style "Vitamin Q" bypass/coupling capacitors, etc. As I went on, I became less and less concerned with historic preservation, and more and more inclined to gut the unnecessary parts.

On top of the chassis, rip out the relay K101, intended to switch between this exciter and an external one for FSK. Use the two BNC connectors on angled brackets -- move both of them to the top, with room for the power splitter /drive level pot.

Obviously JFETs are the best choice, to provide high impedance loads to the tuned circuits that were designed for tube grids. I used the NTE312 again.

FIRST VERSION: I just wired the gate to the grid pin of each stage (with a 47K resistor to ground), the drain to the plate pin, and the source to chassis ground, clipping the existing grid resistors to ground. Actually, the first stage has its gate (and associated resistor) going directly to the BNC connector from the VFO (J101). The plate supply wire was used, with its decoupling networks (1K series resistor and .01 uF capacitor to ground) . 1K, of course, is way too high for the transistor circuit, so I bridged something in the range of 68 to 100 ohms across it. They are on the bottom of the multiplier assembly -- three Phillips screws are accessible with a long screwdriver. The bandswitch and tuning shafts couple to the front panel with slotted plates to allow some free play, so getting them back in takes some coordination. I got lots of practice by taking it in and out many times!

The first stage works as a buffer on the low band, and a doubler above that. The low band switches in a load resistor instead of the tuned circuit -- R106, 10K (on the bottom again!). This obviously needs to be lower. I plugged in different resistors (in the 3300 ohm range, can't remember) across this resistor, until the output at S101, pin 8 was the same on bands 1 and 2.

The second multiplier stage had a tendency to self-oscillate. I cured it with a series RC network from gate to ground, in parallel with the gate resistor: 2200 ohms and 0.1 microfarad. This obviously loads down the tank circuit enough to reduce the Q. A resistor directly across the appropriate tuned circuit might work as well; the series R-C method was simply used because it was easiest to apply without taking the assembly off a million more times. Another such network went on the input of the third multiplier when I discovered another oscillation on the highest band.

SECOND VERSION: Further reflection led me to dislike the sheer length of the unshielded bandswitch wiring in the original circuit. So, the little terminal board on the bottom was completely ripped out, and replaced with a piece of double-sided PC board. Everything was basically re-wired as cleanly as possible near the new PC board ground plane, with excursions to the bandswitch being as direct as possible. The dog-bone ceramic coupling capacitors between tuned transformer stages were saved and re-used for "historical" reasons -- with some wire and insulating tubing to extend them, they were placed directly between the appropriate tuning cans, between the new PC board and the metal chassis. Also saved were the 15 pF capacitors that had been placed near the tube grids, which are part of the overall tuning capacitance, apparently being placed nearer the tube grids to aid in bypassing?

The final tuned circuits were also used in this second rewiring. The lowest band tuned circuit was used as a tuned load for the first buffer/multiplier on the lowest band, and the output attenuated with resistors to match the output of the other bands. The others were incorporated into the circuit by means of 20 pF coupling capacitors. The slugs do not make much difference in the upper 3 bands, but overall waveform purity seems to be better with these coupled in, so you might as well use them.

All of this fed into a broadband step-down transformer, wound on an FT-37-73 core, approximately 15 to 5 turns ratio (have to look again).

This yielded about 1 volt peak to peak output.


FIRST VERSION: In the area of the IPA (intermediate power amplifier, another word for "driver") tube, I built a broadband buffer stage based on W7ZOI's buffer in the VFO section of the "high performance communications receiver" project in the ARRL Handbook. Any broadband buffer circuit with a hi-Z input and a high output into 50 ohms would be suitable. The input comes right off S101, section 2, rear, pin 12.

The splitter transformer requires a 25 ohm input. When built separately, the splitter would be preceded by a 2:1 impedance transformer (approx. 4:3 turns ratio autotransformer). With an output transformer already present, just take some turns off the secondary (originally 5 turns).

I got output of about 3V peak to peak into 50 ohms on the lower 3 bands, and 2V p/p on the highest. Pure sine waves are not a concern for me, since diode ring mixers generate harmonics anyway, and even would rather have a square wave. If it is for you, you can put in a low pass filter for each band, using the S101 section 1 rear (rewired) to pick them.

SECOND VERSION: I used a broadband output amplifier based on a 2N3866 (from DeMaw's "W1FB's Design Notebook"), with adjustable gain (pot in place of the emitter resistor). This gave me about 4 volts p-p on the low end, about 3 volts at the upper limit. This is enough to accomodate any combination of splitters, high-level diode mixers, or what have you.

FOR FUTURE EXPERIMENTING: Consider also some CMOS chips as output amps. A recent QST article featured 74HCxxx chips to put out nice clean 5 volt P-P square waves. There is no better driving signal for diode ring mixers.

FINAL ALIGNMENT: Make the best compromise you can between uniformity of output in a given band, and waveform purity (least feedthrough of sub-harmonic signals). This will take some head-scratching and fiddling, but you can do it.


I have two pots (found matching knobs at a local surplus house!) and a switch added to the front panel, and a transistor switching arrangement (built on the pot terminals) to select the correct pot when the transmitter is keyed. I used 2000 ohm 10-turn pots -- you can use probably use anything from 2 to 10K. You can fiddle with resistors on either ends of the pot(s) to limit tuning to the most linear part of the varactor's range if you wish.

The capacitors hold the VFO at the transmit frequency until the CW waveform has decayed to zero. Your mileage may differ as far as the values needed.

Remember that the diodes affect temperature stability -- as the temperature goes up, the voltage drop goes down, reducing the output voltage and lowering the frequency. If you're lucky, this compensates for the temperature characteristics of the varactor diode! For all I know, the transistors behave similarly. Thus you might best place them near the VFO in a small enclosure. I've gotten by with being sloppier, and the thing is still more stable than you can imagine.

The SPDT switch (front panel) attaches the key lines either to the system keying line, or directly to ground. When grounded, you RX and TX on the same frequency for SSB or for spotting a CW signal to zero beat. When in the other position, you have receive offset for a CW beat note, or for RIT in SSB or CW.

John Seboldt K0JD
Milwaukee, WI (but "forever a zero", having moved out of zero-land for the first time in 1999)
k0jd at seboldt dot net


John Seboldt, K0JD, began hamming as WN0QXG in about 1967. Music, electronics, and ham radio grew side by side in his youth, leading to work in the broadcast industry while studying music at Luther College, Decorah, IA, and The University of Iowa, Iowa City. Church music has been his main field -- he served 15 years in the Twin Cities, and moved to Milwaukee in 1999. (Check out samples of his music work at For now, technology has again claimed his working hours: at Time Warner Cable he's a Broadband Technician, having worked in the cable industry since 2001.