Fullscreen HTML5Med Res (??MB)HTML4

Vintage Radio HeathKit GW-21A handheld transceivers By Dr Hugo Holden Screen 1: a frame from Voyage to the Bottom of the Sea. In the early 1960s, manufacturers such as HeathKit started to lift their game in mobile transceiver design. The clear choice was the single-conversion superhet format, keeping it as simple as possible but not too simple. B y the early 1960s, many germanium transistor radios had been produced, with some capable of excellent high-frequency performance. In Europe, the typical transistors used were the OC169, OC170 and OC171. The similar AF114 to AF117 were ultimately replaced by the AF124 to AF127 series, the former parts all being affected by tin whisker disease. In the USA, various 2N prefix types, such as the 2N2084 made by Amperex, had similar performance to the AF124. The RF-capable transistor types were characterised by having very high transition frequencies and very low collector-­ to-base feedback (Miller) capacitances. That also allowed them siliconchip.com.au to be used in IF amplifier chains without neutralisation. As one example, the AF124, in a grounded base circuit, had a useful power gain of 14dB at 100MHz and was used in the front-end of FM broadcast-band radios operating from 87MHz to 101MHz. In the years that followed, into the 1970s, very advanced germanium types appeared that would work in VHF and UHF TV tuners, such as the AF239 and AF240. These worked in mixer and oscillator circuits up to an astonishing 890MHz. Back in the early 1960s, transistor radios of many kinds were coming to dominate the radio world. These Australia's electronics magazine pushed the older valve (vacuum tube) designs into the background, ultimately making them obsolete. This process was accelerated by the development of temperature-stable, lower-­ noise, higher-power-rated silicon transistors, which generally outperformed their germanium ancestors. Germanium-transistor-based handheld compact transistor transceivers, like the HeathKit GW-21A, started to appear in stores and in popular culture, on the TV and in movies too. Screen 1 (shown above) is a frame cut from an early 1960s TV show, Voyage to the Bottom of the Sea, where a HeathKit GW-21 transceiver was used to save the day. June 2024  99 These 2N2804 transistors were used to replace the MM1056 transistors. They have similar performance to the AF124. Simple super-regenerative transceivers or “walky-talky” designs for children had appeared in toy stores in the 1960s, typically powered by a 9V battery. These ‘toy’ units often used a single transistor stage as an oscillator in transmit mode. A small audio amplifier would amplitude-modulate it. The same transistor oscillator stage then behaved as a super-­regenerative receiver, with the audio amplifier redeployed to drive the speaker in receive mode. Therefore, most of the circuitry in the unit is deployed in both transmit and receive modes, hence the term ‘transceiver’, as the circuitry transforms and reconfigures itself for the two modes of operation. These early transistor-based super-­ regenerative units usually operated in the citizen’s band (CB) around 27MHz. The receiver section was typically quite noisy (as super-regenerative receivers are), and the transmission range was limited. Sometimes the results even disappointed the children as well as the adults playing with them. Transistor superhet receivers of the time were already known to have high gain, low noise and good selectivity in the medium-wave and short-wave bands, up to and above 30MHz. Ideally, the transmitter would also have an independent RF output stage, amplitude-modulated by an audio amplifier, and a separate, stable crystal oscillator would drive that output stage. This two-stage design limits any frequency-modulating effects on the transmit oscillator. Again, the audio amplifier in the transceiver would perform two roles: as a modulator in transmit mode and an audio amplifier in receive mode. This type of design appeared in the HeathKit GW-21 and GW-21A transceivers. They are apparently identical units, except for the transistor types used. Recently, I came across a pair of HeathKit GW-21As on eBay. I had seen them on TV during my childhood and 100 Silicon Chip always wanted them. So, for nostalgia’s sake, I decided to buy them and restore them. Then I could put them through their paces and find out how well they worked. AGC voltage, which is filtered and fed back to Q3 and Q1. It is worth noting that, in a set with PNP transistors, the AGC voltage becomes more positive with increasing signal strength. This tends to take General description the transistors to which the AGC is The GW-21 appeared in the time applied out of conduction, shifting window of 1964 to 1969. The price towards a lower gain condition with per unit at that time was $39.95. In increasing received signal strength. today’s dollars, that is about $380.00 Essentially, the AGC system is a long each; it’s no wonder I did not have time constant negative feedback loop. one back then! The AGC’s time constant & circuit They boasted nine transistors, two resistances are set by the value of 10µF diodes and a single-channel crystal-­ electrolytic capacitor C12 and resistor controlled system using two crystals R14. Note that, with very high signal per unit. Separate crystals were used levels, the voltage on a transistor radio’s for the receiving and transmitting AGC capacitor can reverse polarity, so oscillators. They had an on/off/vol- generally, I replace the AGC capacitor ume control, squelch control, push- with a bipolar or film type. to-talk (PTT) button, an earphone jack, The recovered modulation (audio an external antenna jack and an inte- signal) then passes via “squelch diode” gral whip antenna. A single 9V battery D2 to the volume control. D2 is set up powered the whole thing. with a variable DC voltage applied to The circuit of the GW-21A is shown its cathode from the squelch control. in Fig.1. On the receiver side, the This allows the diode to be cut off, prodesign is of a conventional super- gressively uncoupling the audio feed to het with an RF stage designed for the volume control unless the dynamic single-frequency reception. The RF signal peaks are large enough to overinput from the antenna is passed, after come the diode’s forward voltage drop. appropriate impedance matching, Testing shows that the diode to Q1, the RF amplifier. The crystal-­ has a 0.43V forward bias in the controlled local oscillator (Q2), called ‘unsquelched’ condition. That is more an Autodyne Converter or mixer-­ than enough for the germanium diode oscillator, runs above the received to be in full conduction. With the frequency. knob in the full squelched condition, The oscillator stage receives the sig- the applied forward bias is very close nal from the RF amplifier and the mix- to 0V, so the recovered audio signal ing products appear in Q2’s collector from the detector has to overcome the circuit. The sum and difference fre- diode’s entire forward voltage to pass quencies of the incoming carrier wave through to the audio amplifier. and the oscillator wave appear because The audio is then passed via the the non-linear mixing results in prod- press-to-talk switch (in its unpressed ucts of these two waves. or listen condition) to the input of the The first IF transformer, T1, effec- audio amplifier stages. tively filters off the difference freThe audio amplifier design is typquency of 455kHz and feeds this to ical of the era: a Class-A driver stage transistor Q3, the first intermediate driving the bases of two output transisfrequency (IF) transistor. tors in Class-B. The output transistors Typically, in most superhet radios have just enough initial bias to avoid with a 455kHz IF channel, the receiver crossover distortion. oscillator frequency runs 455kHz These simple amplifiers are energy-­ higher than the incoming carrier wave. efficient, have a low quiescent curIn my GW-21A radios, the transmit rent and are generally suited to batcrystal frequency is 27.085MHz (CB tery operation. The only difference channel 11), while the receive oscil- here is that the output transformer lator crystal in the converter stage is has an additional winding to ampli27.540MHz. tude modulate the power supply to From Q3, the IF signal passes via the RF output stage when the unit is T2, Q4, then T3 in the IF amplifier in transmit mode. to the detector diode D1, where the amplitude modulation is recovered. Restoration In addition, the detector generates an Both the units arrived in good Australia's electronics magazine siliconchip.com.au Fig.1: this is the circuit for the GW-21A. The GW-21 (non-A) version used the following transistors. Q1: 2N1726, Q2: 2N1727, Q3 & Q4: 2N1108, Q5-Q7: 2N185, Q8: R425, Q9: R424. Otherwise, they were mostly identical. June 2024  101 Australia's electronics magazine siliconchip.com.au Because one of the 10µF electrolytic capacitors read high at ~38µF, I decided to replace all of them. I also replaced the 100W resistors in the emitter circuits of the oscillator and RF output transistors. condition, and fortunately, there was no evidence of previous repairs or modifications. Having worked on several items of this vintage with germanium transistors, I decided to start with a standard protocol, checking the electrolytic capacitors and replacing them where required. I removed seven electrolytic capacitors in each unit for inspection and detailed testing. There were some abnormalities. All had leakage values over 100 times higher than a new electrolytic of the same value. Interestingly, the ESR of all of them was within normal limits. The capacitance values were reasonable, except for the axial 10µF electrolytics, which interestingly read around 38µF. Due to the high leakage values, I replaced them all. I also quickly determined that the 100W resistors in the emitter circuits of the oscillator and RF output transistors were out of spec at 135W each, so I replaced them too. All the other resistors were in good order and within the expected ranges. One of the units had a cracked section on the lower corner of the phenolic PCB. I strengthened it with a small 2mm-thick brass plate tapped with threaded holes for 1.6mm brass screws to secure it. I cleaned the potentiometres, transistor sockets and PTT switch with CRC’s CO contact cleaner and then lubricated them with Inox’s MX3, which I have found better than using a combined cleaner-lubricant product. Inox MX3 is a very high-­purity oil; I have subjected it to several experiments on various metals, and it is my preferred lubricant for restoration work. Before attempting testing and alignment, I have a standardised approach when transistor sockets are present for checking the transistors for gain and noise. I check the audio transistors in-­ circuit, though. I replaced the speaker with a 10W dummy load (the original speaker is a 10W type). I then connected my oscilloscope across that dummy load and fed a test sinewave signal from a generator to the input of the audio stage (in the driver transistor/volume control area). It is easy to see if the audio transistors are OK in this sort of amplifier. If either output transistor is unwell, it unbalances the output, and the sinewave becomes asymmetric. Also, the Fig.2: the circuit I used to test for defective transistors. 102 Silicon Chip Australia's electronics magazine driver transistor can easily be checked against a known-good germanium PNP audio driver transistor like an AC126. The output transistors can be verified against known-good AC128 types. One final check is to compare the audio amplifier sections between the two units for gain and power output. I was satisfied that both units’ audio stages were normal and that all the original audio transistors, RCA 2N407 types, were perfectly operational. The radio-frequency transistors are a different matter. I check them out of the radio in a test jig with a socket, to examine their gain and frequency response up to 100MHz. Its circuit is shown in Fig.2. This is a way of screening out defective transistors. I use a Philips PM5326 RF generator, which has a 75W output resistance, and a Tektronix 2465B ‘scope, set on its 50W input resistance option. The transistors are placed in the socket of the simple test jig to evaluate their basic performance and compare them to some excellent AF178, 2N2084 and AF124 transistors that I have, as well as comparing the same types from the two units with each other. The test circuit quickly screens out noisy and weak transistors. On testing, the 2N1525 IF transistors all had similar properties, with nearly identical gain to an AF124 reference transistor below 1MHz. Unlike the AF124, where the output amplitude in my test jig drops by 50% at 70MHz, the 2N1525’s output reduces by 50% at about 10MHz. The 2N1525 transistors are just satisfactory enough (low enough collector to base feedback capacitance) to work in an IF amplifier without neutralisation. You will notice from the GW-21A siliconchip.com.au The phenolic PCBs for the HeathKit GW-21A transceivers. An original PCB is shown at left; the adjacent PCB has new electrolytic capacitors and a crack repair in the lower left corner. circuit that it has a non-­neutralised 455kHz IF. The A1384 transistors in the RF, converter, and transmitter oscillator stages were all good in both units. These are not 2SA1384s; they are an Amperex part. In the test jig, their output drops to 50% at around 50MHz. They are higher-frequency capable than the 2N1525 transistors used in the IF amplifier, as they have to be for the role they play operating in the 27MHz stages. Then there were the two RF output transistors to test for each unit, the somewhat mysterious Motorola MM1056. I could not find the original Motorola data sheet for them, so I didn’t know the expected transition frequency. Some basic data I found online suggested they were similar to the AF124. I also posted on the Antique Radio forums but had no luck finding the original Motorola data sheet. The logical place to find it would be in an early 1960s vintage Motorola transistor data book. siliconchip.com.au One of these transistors was defective, and its leads had been cut by someone in the past. The junction was damaged and badly leaking. The transistor from the other GW-21A unit was good. Testing the good one in the test jig, it was clearly capable of very high-frequency performance, being very similar to the AF178, with its output dropping to 50% by about 110MHz. However, during alignment and testing of the transmitter section of the radios, I elected to replace the MM1056 transistor in both units with Amperex 2N2084s, as they gave more stable results with slightly higher output. I also found some capacitive coupling effects on the transistor body. In these HeathKit radios, all of the transistor sockets have three pins; there is no shield connection. The quick solution for the 2N2084 was simply connecting its shield (case) to its emitter wire (which is at RF common). That solved the problems of higher frequency parasitic oscillation I Australia's electronics magazine observed with the original MM1056 transistor and the Amperex 2N2084, when the body of the transistor was floating in both cases. After aligning L5 & L6 in the transmitter section, Scope 1 shows the output of the transmitter with the antenna retracted into the unit and the scope connected to the base of the whip. The measured voltage was about 16V peak-to-peak. With the antenna up, the amplitude drops to about 8V peakto-peak. Of note, if L5 is peaked for maximum Scope 1: the transmitter output with the antenna retracted. June 2024  103 The underside of the GW-21A PCB. Note the modified AAA cell holders; I did that because the battery compartment was too large for a typical 9V battery. This time, the repaired PCB is shown on the left, although both have new capacitors. power output and then the slug is unscrewed further, the oscillator can drop out or fail to start when the pushto-talk button is pushed. So it is best to adjust it just a little on the opposite side of the peak, with the slug a little further into the former. With the speaker replaced by a 10W dummy load, I couldn’t talk into the speaker to test the transmitter, so I applied a 1kHz sinewave modulation signal from a signal generator. I set the generator output to 0.5V peak or about 350mV RMS and used a 3.3kW series resistor to deliver the signal across the 10W dummy speaker resistance. That corresponds to only about 1mV RMS of signal to the input of the audio amplifier. The result is shown in Scope 2, with the carrier at the antenna base now at about 28V peak-to-peak on the modulation crests. Increasing the modulation signal level from the generator, the RF output stage clipped fairly softly, and the 104 Silicon Chip carrier was not modulated to zero, as shown in Scope 3. This occurred before clipping in the audio amplifier. I was pretty impressed by the reasonably soft carrier clipping and residual carrier signal. RF output power I read on the internet that the output power of this radio was 100mW, but I wanted to check it for myself. After working on these radios for some time, I noticed that the 9V batteries I had been using, which had seen some use before, had dropped to 8V. So I repeated the carrier output test with fresh batteries and got the result shown in Scope 4. With a fresh battery, the RF output at the antenna base (with the antenna retracted) comes up to 12V peak or 24V peak-to-peak and about double that at 100% modulation. Raising the whip antenna caused the voltage to fall approximately 50%. That suggests Australia's electronics magazine the antenna impedance has been well matched with its loading coil to the RF output stage. I decided to test with various load resistors at the antenna’s base, with the antenna retracted, to find which resistance also lowered the RF level to 50% to estimate the antenna’s impedance at its full extended length. A 680W resistor resulted in the level dropping by 50%, much as extending the antenna does. With no modulation, the voltage developed across the 680W load was 6V peak or 4.24V RMS, and at full modulation, it was about 8.48V RMS. Therefore, the peak envelope power (PEP) delivered to the 680W dummy load resistor (or the fully extended antenna) is approximately 106mW (8.48V2 ÷ 680W) at full modulation. With zero carrier modulation, the RF output power is ¼ of that, about 26mW. So, the suggestion that these GW-21A radios had a 100mW RF siliconchip.com.au output probably referred to a PEP measurement, not an unmodulated carrier wave power, which is ¼ of the PEP. In another attempt to estimate the RF power output, I tested the signal out of the external antenna jack. The output impedance here appears very low. Unloaded and unmodulated, it delivers a signal of about 4V peak-to-peak. Loaded with a 15W resistor, it drops to 2V peak-to-peak (0.7V RMS), corresponding to around 32mW (unmodulated) into 15W. I made a 1:2 turn ratio (1:4 impedance ratio) ferrite RF impedance matching transformer and found, unmodulated, it could deliver 28mW into a 50W load, or around 112mW PEP at full modulation. Receiver alignment The receiver alignment was pretty straightforward. First, I aligned the IF by connecting the ‘scope across the 10W dummy speaker load resistor and applying a signal to the antenna connection from a Philips PM5326 RF generator. I set the generator for precisely 455kHz at a carrier modulation level of 30% and the volume control to maximum. I unplugged receiver crystal X1 to disable the converter. Enough level was provided so the recovered signal was visible just before significant AGC activation, and I peaked IF transformers T1, T2 and T3. After that, I plugged the receive crystal back in and set the generator for 27.085MHz, then aligned the receiver for maximum gain by adjusting L1 and L2. I then disconnected the generator, attached a small antenna to the generator output and adjusted L1 and L2 again, with the GW-21A’s antenna extended a few metres from the generator. I did that in case the attachment of the generator had caused some Scope 2: the amplitude-modulated output with a 1mV signal injected. siliconchip.com.au detuning effects, but it turned out that the slugs of L1 and L2 were already in the correct positions. The signal was audible above the noise floor when the generator’s variable attenuator was in the region of -70dB to -80dB. With noise and signal about equal to the ear, the attenuator was on -75dB. The PM5326 generator on 0dB applies 50mV RMS into 75W and about double that to a high-Z load. This suggests the receiver can resolve a signal of about 17µV from the noise floor. Once the receivers were aligned, it was time to try them out. In practice, at full volume, there is moderate audible noise; nothing as severe as a super-­regenerative radio, though. The squelch control works well, unlike a typical squelch that suddenly kills the noise; its effect is more gradual. I could hear intermittent transmissions of people speaking at times, with American accents, making me wonder if that was some sporadic short-wave transmission on CH11 from overseas. In any case, the receiver appears very sensitive indeed. So far, I have tried these radios with about 100m separation with very good results. I am going to perform a maximum line-of-sight test on them soon. so the batteries would fit snugly. The final photo shows the two restored units with the batteries fitted. While many HeathKit radios were sold as kits, the quality of the construction makes me think these two were factory assembled. Summary The battery compartment is a little large for a typical 9V battery, so I modified some six-AAA cell holders and fitted them with a 9V battery power clip. That gives a much higher capacity battery at a lower cost. With these holders, it pays to tape the batteries in. I use Scotch 27 fibreglass tape as it can be reused a few times, and it stops the holders from sliding around, too. The photos show the relative size. It was necessary to add some soft packing into the battery compartment The GW-21A is a remarkable early germanium transistor handheld transceiver. While it does not have a spectacular RF output power compared to modern transceivers, only 100mW PEP, it makes up for that by having a very sensitive superhet receiver. The GW-21(A) is far from a toy radio. It would have been a dream to have owned a pair of these as a boy, back in the 1960s, when most transceivers children could get their hands on were poorly performing noisy super-­ regenerative types. These sorts of transceivers make an interesting restoration project, and replacement or equivalent germanium transistors are SC still available if required. Scope 3: the amplitude-modulated output with maximum modulation. Scope 4: the carrier output test signal with new batteries. Australia's electronics magazine June 2024  105 Batteries