The Homebrew Rigs Page.

Back to the Radio Page.


The first QRP rig - for 80m CW using direct conversion.


After a start on commercial gear, I realised that power wasn't everything and, as I had a preference for CW (Morse) operation, I could build simple transmitters and receivers from the bits in the 'junk box'. My first rig was built in 1988 - an 80m (3.5MHz) CW rig consisting of a FET Vackar VFO and buffer driving a BFY52 which in turn drove a pair of BFY52's in push-pull to about 1.5W. Using the VFO and buffer to drive a pair of signal diodes followed by an audio low-pass filter and a bipolar amplifier to drive headphones gave me a transceiver using direct conversion. This worked well with good sensitivity and many stations were worked around Europe using a G5RV antenna at about 5m above ground. The whole circuit consisted of component circuits taken from the GQRP handbook with a bit of my own circuitry. I added and independent receiver tune facility - essential for netting correctly. Transmit/receive control was tagged onto the keying circuit.

I learned a lot from this rig and followed it with loads of trials of VFO's, buffers, drivers, PA's and simple receiver circuits, most of which came from the GQRP circuit book.

Kanga products were advertising a kit for a 20m/80m receiver with the option to add a transmitter later. The 'Cheriton' arrived and after unpacking I studied the parts and the circuits. This was quite a big step upwards as far as I was concerned. The basis of the receiver was a 9MHz crystal filter and a 5MHz VFO. Adding or subtracting the 9MHz and 5.0 to 5.5MHz combination gives 3.5MHz to 4MHz and 14MHz to 14.5MHz - both amateur bands. This technique is known as band imaging and allows two bands to be received using the same infrastructure.

Cheriton - Front View

The only parts that need to be switched are the front end filters. This rig was built into an old KW Electronics chassis so that I could work on it easily whilst having something to fix the extraneous bits to. After a few weeks work, the rig was powered up and I set about getting it to work. The KW2000A was used as a signal source to set up the front ends and after much trial and error, I borrowed a frequency counter to set up the VFO and Carrier Insertion Oscillators. The rig worked beautifully and it was interesting to be able to hear only one sideband after the direct conversion receiver. Thoughts soon turned to the transmitter. The problem was the PA. Speaking to Ian Keyser (the designer) I heard that he'd been setting his up and using only the 22mW generated on the board had worked an American station using SSB! Encouraged by this, I connected the output to the antenna and fired it up. As might be expected, the first few tries on 80m resulted in nothing but I persevered and 'tail ended' a CW QSO. The station came straight back with a 4/5 report. He was only 40 miles away but he was amazed when I told him what power I was running. Later, I bought the Cirkit PA kit and got the transmitter going and the station tally increased. There were various problems with RF feedback due to the poor layout I was using but again I persevered and sorted most of them.

Top View

Then came the big test - SSB. I bought the transmit board kit and again spent a lot of time getting the various circuits to work and checking that they were not being overdriven. I listened to the output on the KW2000A's receiver and all appeared well. The antenna was connected and stations tentatively called. I asked for critical reports and stations were only too happy to give them. Mostly, they were favourable and I gained confidence in the rig. After much thought, I bought a metal case and built the board in along with the transmit board, the PA and control circuits. The result can be seen in the pictures. I had a lot of fun with that rig and learned a lot from its construction. At a later stage, I designed a mixer-VFO circuit to put the local oscillators out of the way on the high side of the IF. This involved adding the 5.0 to 5.5MHz VFO with an 18MHz crystal oscillator to give 23 to 23.5MHz for 20m and with 7.5MHz to give 12.5 to 13MHz for 80m. This worked well and allowed me to fit a frequency counter to the mixer output to give a direct readout (if the MHz figure was ignored). I used the Cheriton both as a base station rig and also portable and mobile with a home made quarter wave whip antenna. Click here to go to the Cheriton and Kitten II Page with larger photos.

30m 'Voxner' - Controls from left: Phones, AF gain, IRT and Tuning
The Voxner.
It was then back to direct conversion and a rig for the comparatively recently released 30m band (well I'd not been on any of the 'WARC' bands at that time). This one started as a home design using Plessey SL600 series chips bought unmarked from Birketts of Lincoln. However, after spending a good while on it with little success, I gave up and started again with a 10MHz VFO. This is getting a bit high for a free running oscillator but with careful construction and selection of components, the oscillator was quite stable. This was followed by a VN10KM VMOS FET power amp and half wave filter. The receiver was based on the tried and tested 'Oner' mixer and a small AF amp to drive headphones. This rig worked well and ran about half a Watt out. Stations all over Europe were worked on CW with a 30m dipole with best results after dark. The direct conversion receiver worked well considering the large amount of broadcast stations in the band.
The 160m 'LCK' in a 2nd hand Die Cast Box.

Next came a rig from Sprat - the GQRP magazine.

This one attracted me for it's simplicity, it's superhetrodyne design and use of ICs. The LCK was designed for 160m or 80m and by all accounts could be coaxed onto 40m and even 20m. This rig is based around the NE602 - a chip designed for the mobile phone market containing an oscillator and a mixer. This chip has found it's way into many simple designs (and some complex arrangements). This rig gave me my first taste of homebrew crystal filters. There were many surplus crystals from TV scan frequencies of 4.333MHz (Europe) and 3.579MHz (USA) on the surplus market and they were well utilised in homebrew filters. However, I found some high Q 4782.72KHz rocks in the junk box and decided they were close enough. Initial thoughts of 'dual banding' this rig were shelved and I decided on 160m - 'Top Band'. A printed circuit board was etched and built up and after a fair amount of fault finding, the rig came to life. The power amp is a VN66AF VMOS FET followed by a half wave filter which gave about 2 Watts out.

The tidier 80m 'LCK' transceiver.

A second version of the LCK was made for 80m. A holiday in Devon looming, I needed a rig to use portable. The 4782.72kHz rocks again provided the IF filter but the VFO at 8MHz proved to be a bit high for the NE602 with stray capacitance and, to some extent, the physical construction. However, apart from a bit of warm up drift, the rig worked well, producing 2W on 80m. I included a variable bandwidth audio filter and an RF gain control.

The 40m 'Oner'

Another portable rig accompanied me on a diving trip to Orkney. This one was for 40m and needed to be compact in order to fit in with all the diving gear. This rig is direct conversion and based upon the 'Oner' series of 1 inch (25.4mm) square PCB's from Sprat. The PA has a VN10KM FET driven to about 1 Watt and transistorised TX/RX switching, VFO control and IRT. The rig worked from a 9v battery and I worked many European stations using a long wire antenna from various portable locations. Orkney was particularly beautiful and the diving in and around Scapa Flow spectacular.

I also tried a very simple crystal controlled 40m rig using three transistors - one for the crystal oscillator, one for the PA (BF139) and the third for the keying switch. I worked a few stations with this arrangement and an Eddystone 888A receiver which dwarfed the transmitter in terms of size and power consumption.

The CSP with the simple Frequency counter above.

Next on the scene was another rig from Sprat.

The 'CSP' is another rig from the G-QRP Club stable (Sprat 67 and 68), this time coming from George Dobbs and Ian Keyser. This CW and SSB rig for 20m is based around the NE602 IC. The circuit utilises a 5 - 5.5MHz VFO and a 9MHz crystal filter. IRT is included and I used a 3W MOSFET power amp from Cirkit. I worked a few stations on both CW and SSB with this compact rig.
The box above is a simple frequency counter (Sprat 73) reading the VFO frequency. As there are CIO offsets and an audio note to consider, the frequency is about 1.5kHz out but certainly close enough to be very useful. The rig runs from a 12v supply.

The 40m 'Oner' MKII with adjustable AF filter and IRT.

The 40m Oner - MKII (July and August 1992).
This was my attempt to put together a 'Oner' on a single PCB and incorporate as many bells and whistles as possible - a fruitless endeavour - Keep it simple stupid!!
This rig actually worked better than I give it credit for and the only real problem was that of the stability of the VFO. I should have paid more attention to detail. Once 'warmed up', the drift became manageable (and the original simple receiver suffered from broadcast breakthrough). A Colpitts VFO running at 7.0 to 7.05 MHz driving the twin diode receive mixer and also the PA. Eventually I replaced the diode mixer with an NE602 and the BC interference disappeared. A bit of a cop-out really considering it's predecessor was comparatively immune to BC interference.

The 30m 'Bitsa'. Superhet CW transceiver for 10.1 to 10.15MHz
30m 'Bitsa' (October 1993).

I wanted to have a go at building a superhet transceiver for 30m and drawing in the experience gained with previous projects, I started to assemble modules with the required functions from various sources. Having built a few crystal filters, I felt confident I could design and build one from scratch and using the crystal characteristic meter I'd built (from W1FB's designs) I set about the task. Scanning through the box of crystals I'd amassed from various PMR rigs and the like, I selected crystals of 6950kHz requiring a local oscillator at 10MHz + 6950kHz = 16.960MHz. I picked out 5 crystals with a resonant frequency within a 50Hz span. The exact freq. of 6941.66kHz emerged so with an RF input range of 10.1 to 10.15MHz, the LO needed to be 17.04166MHz to 17.09166MHz. I intended to use a mixer VFO to achieve this and looked around for a suitable crystal and found one at 13266.66kHz requiring a VFO of 3.775 to 3.825MHz - conveniently within the range of an 80m receiver to check it's operation. The mixer VFO was built up using an NN1G circuit from Sprat. The front end is the trusted 2N3866 run at high level and the Plessey SL6440 high level mixer feeding the homebrew crystal filter and an MC1350 IF amplifier. An NE602 product detector with 6914.66kHz crystal pulled to 700Hz LF for USB feeds an LM386 audio amp to drive headphones or a speaker comfortably. The TX side uses a keyed oscillator at 6941.66kHz using an NE602 to mix with the VFO to give 10.1 to 10.15MHz. This signal is filtered by a top coupled bandpass and feeds the PA. I had a couple of goes at the PA, settling eventually for a MOSFET rather than the original MPSU31 CB transistor. An AGC circuit derived from the audio signal provides good control of the IF gain and a transistorised TX/RX control circuit with switched IF gain provides good 'break-in' operation allowing the transmitted signal to be heard against the background of the received signal. This method is very 'ergonomic' allowing the operator to hear what's going on whilst transmitting. Therefore, should you be called once you've gone back to transmit you can hear the call. The rig is sensitive, selective and easy to operate. I fitted an audio filter to reduce the frequency response of the AF system, effectively reducing the hiss. I'm quite pleased with the way this rig has turned out considering it's really just a pile of 'bits'.

Malta 40. (July 1994)
This rig was designed by Steve Hunt G3TXQ and appeared in Sprat 78. I built the PCB using bits from the junk box and a few new parts. A re-cycled die-cast box provided a stable environment. I deviated from the design a bit as I used some surplus 10.7MHz ex-PMR filter crystals rather than the 10.240MHz units Steve suggested. This required minor changes to the oscillator set-up but nothing significant - in fact, it put the VFO neatly into the 80m band so I could calibrate it using the Cheriton's receiver (10.7MHz - 7 MHz = 3.7MHz and 10.7MHz - 7.1MHz = 3.6MHz). I used the Crystal Characteristic Analyser I'd used for the 30m Bitsa to generate the filter (which had used the Malta 40 format). On air trials gave very poor results - the filter was far from sharp and I seemed to be unable to get it operating correctly. The crystal analyser was brought out again and closer measurements made. I found the 'Q' of the crystals to be low. Calculating values for these crystals gave odd results and after studying the equations, I realised they were completely unsuitable. I split many other filters open but found no better crystals. I had nothing else in the range and new rocks would have been about £1.50 each so I looked through the crystal box and came across the 4782.72kHz rocks I'd used in the LCK. These had high 'Q' and some experiments showed them to be quite suitable. The VFO frequency would need to be changed to (7.1 and 7.0MHz - 4782.72kHz =) 2217.28 to 2317.28kHz. Open heart surgery followed. The results were amazing. The filter was narrow and had steep sides with deep attenuation. I've used this rig extensively on holidays and expeditions to various parts of the country (for climbing and caving) and it's performed faultlessly. It is a pleasure to use. A good friend Hugh, (and no doubt many others) also built the Malta 40. His is far neater and is fitted into a smaller die cast box. Hugh's uses the prescribed 10.24MHz crystals and works well. Output is about 3 Watts.

The Malta 40 - a superhet transceiver for 7.0 to 7.04MHz. Mine (top) and Hugh's (below).

The K2 Project.

The Kitten II HF transceiver for SSB and CW.

Ian Keyser has produced many designs but more importantly encouragement to experiment and this rig certainly did that for me. It's a huge project to set out to build a multi-band transceiver for HF SSB and CW. Ian approached the project in a way that made it feasible and allowed for as much input from the constructor as was felt possible or appropriate. The rig revolves around the 9MHz IF heart with filters for SSB (2.2kHz) and CW (500Hz) if desired.
There are many challenges in a project of this type but some of the greatest are the physical construction, taking into account the distribution of signals and power to the various component boards and their position with respect to one another. Looking back from the other end of the project, I can see that time spent pondering these was time very well spent. Band switching, for example, is always a problem. In older multibang rigs, there was usually a multi-way rotary switch with many elements to control the power and signal switching and more taxingly the different oscillator and mixer components to get the components to operate at the required frequency. Another technique might be to build oscillators and mixers for every band and to switch them in and out as required. This would be better from the operational point of view but very space hungry and one would really be better to build a dedicated rig for each band. The bottom line is that any multiband arrangement compromises performance against cost and convenience and it's a juggling act to achieve a suitable balance.

Consider the front end: this needs to have bandpass filtering for all the bands of interest and for an HF rig, this means operating over 1.8 to 30 MHz a range of 4 octaves. The problems here are those of stability, overload handling (dynamic range) and bandwidth in each section.

Rear view of the Kitten II. All oscillators are brought out to sockets and the counter input is also available making the rig useable as a piece of test gear.

The local oscillator should be HF of the filter to avoid mixer products in band and needs a range of 9MHz+1.8MHz to 9MHz+30MHz (10.8 to 39MHz) with constant output to drive the front end mixer.
The vfo also needs to change frequency to scan across the band and this is typically chopped into 500kHz segments. It goes without saying that the VFO should be stable!
The IF amplifier should be stable with correctly distributed gain and include a suitable AGC system capable of responding to the needs of both CW and SSB signals. It might also need to handle AM and FM signals.
The audio circuit should be stable, bandwidth limited to avoid hiss and have sufficient reserve to easily drive a loudspeaker. An audio filter might also be included.
The RF power amplifier should be stable with correctly distributed gain and have enough power for the application without overdriving. This would be very much down to the user in what level of power was desired. For this rig, with QRP in mind, the 20W level was considered appropriate. Some form of ALC is also desirable.
The modulator/SSB generator must produce clean, undistorted drive to the transmit mixer and have a means of leveling out the drive on each band. It must also generate both upper and lower sideband signals. The mixer must have suitable bandpass filtering before amplification to levels suitable to drive the power amplifier.
CW generation should be click free and level controlled.
There should be an accurate system for indicating the operating frequency. This could be related back to the VFO but would be more accurate, considering the various signals mixed to obtain the final frequency, if the RF frequency were measured. This may require some 'processing'.
There should also be bandpass filtering for the output signal

So, this might be seen as a 'system' wish list.

The bandswitching, mentioned earlier, is done mainly by diode switching, a technique whereby a bias current is passed through a signal diode into which the switched signal is a.c. coupled. Using small chokes and capacitors, small signals can be routed to a common point simply by switching the bias to the appropriate diode on. Where larger signals are required to be switched, miniature relays are used. Click here to open a page with more detailed pictures of the innards.

Back to the Radio Page.