Paraset - a WWII spy radio

Paraset with power supply & headphones

Introduction

I've been interested in "spy radios" for quite a while. Last fall I finally decided to try my hand at building a version of the "Paraset" spy radio used by the Resistance during World War II. The Paraset consists of a one-tube crystal-controlled CW transmitter putting out something less than ten watts, plus a two-tube regenerative receiver. It covers about 3.3 MHz to 7.8 MHz, including the 80-meter and 40-meter amateur bands. There's a good amount of information available on the Web from people who have studied some of the few remaining original sets, and from many people who have built replicas, some just to get an interesting "work-alike" on the air, and some trying to build sa copy that would pass pretty strict scrutiny. My version is meant to be usable on the ham bands, using mostly modern parts, while still looking rather like it hails from 1939. Not only is this the first vacuum-tube radio I've built, it's the first one I've used, so I feel like I'm getting the ham-baptism I would have gotten in the 60's if I'd managed to get my license then!

Front Panel

The Paraset should probably be considered an early "trail-friendly" design: all the controls and connectors are on the front panel! There were two versions of the original Paraset. Early sets were housed in a wooden box with no top, and had no labeling on the panel. They had a funky slide-in connector for the headphones, and a power cord coming out the side of the box. Later sets were housed in a metal "cash box" with a hinged top, had each control and socket labeled, and had a standard 1/4" jack for the headphones. My copy splits the difference here: it's in a wooden box (with a hinged top waiting to be attached), but the panel is labeled and the headphones plug in to a 1/4" jack.
Front panel with parts & jacks mounted

I made some other departures from the original. Like many modern builders, I changed the power connector on the set from a socket to a plug, so the cable carrying high voltage needn't have exposed pins. Along with the socket for the large FT-171B transmitter crystal, I paralleled a socket for FT-243 crystals that are a bit easier to find. I added a 1/4" socket for an external straight key, bug, or keyer. (Some original Parasets were modified to allow an external key, but usually with a pair of pin jacks on the right-hand side near the key knob.)

The most glaring difference from the original is the band switch. The original Paraset's receiver tunes from 3.3 MHz to 7.8 MHz with just turning the dial 180 degrees, so the tuning is notoriously touchy. I added a 4-position switch to allow selecting either the original range or one that spreads a single ham band across most of the dial. (Circuit details are described in a section below.) Even that would be hard to deal with, but for the ingenious mechanical fine-tuning mechanism used on the Paraset. Perhaps this was common in the period, but it was new to me. A round brass plate under the main tuning dial is captured between a pair of small round plates attached to the fine-tuning knob, so that turning the small knob drives the main dial with a geared-down effect. Here is the mechanism disassembled:
Fine tuning parts

Circuitry

I essentially followed this schematic from SM7UCZ:
Schematic from SM7UCZ

All the big parts inder the front panel are placed pretty much the same as in the original Paraset -- only the T/R switch and one indicator lamp moved to make room for the added band switch.
Oblique view of circuit from tune knob Oblique view of circuit from power Circuitry head-on

The big horizontal coil form has the transmitter tank coil on the right, which is coupled to the antenna coil on the left. Each of these is paralleled with one of the 100pF variable capacitors at the bottom. The toggle switch adds an additional 100pF in parallel to the transmitter tank to cover the lower frequencies. I made one circuit change in this area to remove high voltage from the tank tuning capacitor, by putting a 0.022μF capacitor in series between the plate and the variable cap, and connecting the other end of the variable cap to ground. Here's the original circuit -- notice that the left end of the tank coil is at RF ground, due to C4 and C5:
Original transmitter tuning
and here's the modified version:
Transmitter tuning at DC ground

The telegraph key is simply a piece of thin printed-circuit-board material, analogous to what was used in the early Parasets, rather than the elaborate construction used in the later model. It contacts a brass screw in a piece of brass suspended from the front panel. It's not too bad to use, though the short arm means it has a noticeable curve to its travel. I haven't operated with an external key yet. The transmitter uses cathode keying, and the voltage on the key is about 36 volts, with key-down current about 50mA, so my solid-state habits will be tested!

At the far right is the heaviest part, and one of the hardest to find. It's a 30H choke used as the plate load of the receiver's audio amplifier tube. Next to that is the receiver tuning capacitor, and the vertical coil form holds the tapped coil for the receiver's Hartley oscillator, plus the antenna coupling winding. These pictures were taken before I added any of the band-spreading components to the band switch.

The T/R switch just below the center switches high voltage between the plate of the transmitter tube and the plates of the two receiver tubes, and also switches the antenna between the transmitter output and the receiver input. A 1-Meg resistor allows a bit of current to the transmitter even during receive, which allows spotting the transmitter crystal on the receiver dial. Each press of the key gives a brief "bip" or "boop" in the headphones if you are tuned close enough.

Most of the "small" parts -- resistors, axial capacitors, mica capacitors, RF chokes -- are modern parts straight out of the Mouser catalog. Many of them are more compact than their 1939 equivalents, so the wiring wasn't quite such a tight squeeze as in the original Parasets, though a few spots took some planning, and it is definitely a 3-D build.

Bandspread Tuning

As mentioned above, I added a switch to reduce the sensitivity of the receiver tuning. For example, one switch position spreads the 80-meter ham band across about half of the dial. Here's how this is implemented. We start with the original tuned circuit in the receiver:
Original Rx tuning

The frequency tuned is given by
F = 1 / ( 2*pi * sqrt( L*Ceff ) )
Ceff = Cvar + Cstray

with F in Hertz, L in Henries, and Ceff in Farads. Ceff, the effective capacitance, consists of the variable capacitance we dial in, plus any stray capacitance in the circuit. This includes about 6 pF of input capacitance from the tube, plus stray capacitance in the wiring. Stray capacitance also includes capacitance within the band-switch itself, between unselected terminals.

To limit the tuning range, we can insert capacitors at the spots marked "A" (top of the tuning capacitor) and "B" (ground) in the circuit. Generally speaking, series capacitance raises the minimum frequency, while parallel capacitance lowers the maximum frequency. There are two basic configurations possible. Here's the one I used to tune the 80-meter band:
80-meter band tuning

Here C3 is in series with the variable capacitor, and C4 is in parallel with the combination.
Ceff = (Cvar*Cseries)/(Cvar+Cseries) + Cparallel + Cstray
Ceff = 1/ ( 1/Cvar + 1/Cseries ) + Cparallel + Cstray

(The above equations use two equivalent forms of the formula for capacitors in series.) You can try to calculate appropriate values for these capacitors, but I ended up tweaking the values by experiment. The 52pf was made up of parallel 22p+22p+8p capacitors, values I happened to have on hand.

The other configuration works better on the higher frequencies. Here's what I'm currently using for the 40-meter band:
40-meter band tuning

Here C8 is in parallel with the variable capacitor, and C7 is in series with that combination.
Ceff = ( (Cvar+Cparallel) * Cseries ) / ( (Cvar+Cparallel) + Cseries ) + Cstray
Ceff = 1/ ( 1/(Cvar+Cparallel) + 1/Cseries ) + Cstray

I found that adding components to tune one band affected the tuning of the other band. I think this is due to capacitance through the band switch. It's probably best to get the tuning right for the lowest bands first, as the larger capacitor values there are less affected by changes in the stray capacitance as you adjust another band.

I found I could switch between the original circuit (labeled "FULL" on the front panel) and a number of band-limited circuits in either configuration, with two poles of a rotary switch connected to the points marked "A" and "B". Here's a schematic showing the switch, with tested values for 80 and 40 meters, and two candidate configurations for 60 meters, as yet untested:
Bandswitch tuning

One end of each unselected capacitor network connects to the tuning coil, but the other end is, ideally, an open circuit. In fact it is connected to an unselected switch terminal, and has some influence on the circuit through stray capacitance, but this doesn't really present a problem.

What's Next?

I have a few crystals that are usefrul in the 80-meter and 40-meter bands, and I've made a few contacts. I also have some crystals that are a bit below a useful spot, and I plan to try raising their frequency, either by grinding or by etching.

I also hope to add to this account with more information for other enthusiasts, including where I managed to find my parts, more details on the build, pictures of my power supply, and a list of web sites I've found useful. But for now, I want to get on the air!