Development (and failed) projects

Quite a few of my projects don't work out as expected or, occasionally, are superseded by better projects. This page is about these old and failed projects and also experimental work and lash ups that I have tried with along the way.

I've large "work in progress" project it on it's own page. It's my TR40 forty meter SSB transceiver This is very stalled which is a shame. One day... For the TR40 , see the dedicated page here

As shown, a lot of the circuits are not enclosed - this approach should be used with caution for obvious safety reasons. Please read the comments on safety and site disclaimer here if you are tempted to try anything like this.

Forty meter (7.0 MHz) Regenerodyne receiver

40m regenerodyne, front view

My progress on the 40m SSB transceiver project had been stalled for some time when I built this radio; I thought I would build another (relatively) small project to get over the stalled project blues!

A little travel had highlighted the fact that I have no portable Ham equipment and so I decided to build a small valve receiver for the 40m (7 MHz) band. A rather obvious advantage of the higher HF bands is that they require progressively smaller aerials and so are more suited to portable and temporary operation than say the 80m or 160m bands.

I decided that I might have trouble getting a standard regenerative receiver to perform as well as I wanted at 7 MHz and so elected to use the slightly more complex "regenerodyne" design where a conventional tunable regenerative detector is prefixed by a mixer using a fixed high frequency local oscillator. Typically the fixed local oscillator will be crystal controlled and so will exhibit very little, if any, drift.

With this arrangement it is possible to down convert from high frequency bands to a relatively low frequency where the regen detector can work its best - in terms of both stability and selectivity.

By using a switched local oscillator frequency and suitable input filters it is possible to make a simple, stable and effective multi-band receiver with a constant tuning rate on each band. To make sure this particular project got completed and to keep the size to a minimum I decided to make my first regenerodyne just cover one band.

40m regenerodyne, front 3/4 view

The regenerodyne concept was first used in the 1920/30s to provide stable reception of the higher HF bands (perhaps 14 MHz and above) whilst using simple regenerative receivers. The name was more recently coined by Gary, WD4NKA who is probably the main current exponent of this architecture. You can find his interesting pages either from my links page or directly from here.

With space at a premium (I'd already picked a small case :) I decided to try some miniature Toko coils I had a number of. These will adjust between about 4 and 7 μH and have a tap on the main coil as well as a separate link winding. The Toko part number is KANK3334R.

After much deliberating and parts box searching I worked out that a 25 pF variable capacitor would just about tune a 100 kHz range at around 3 MHz (my chosen regen frequency) using one of the Toko coils set to it's maximum inductance and the right fixed capacitors as well. I found a 10.165 MHz crystal and so my regen tunes 3.165-3.065 MHz to cover 7.000-7.100 MHz with this crystal.

The tube line up is: 6C4 xtal oscillator, 12AT7 mixer (Pullen configuration) 6BH6 regen detector and 12AT7 1st and 2nd audio amplifiers. Operation is via headphones and/or an external loudspeaker. The audio output transformer is a tiny mains 240->3-0-3 Volt 200 mA unit, perhaps not ideal but it works at the low power level generated.

40m regenerodyne,top view - it's busy!

The radio is wired for both 6.3 and 12.6 Volt heater operation with an intended HT voltage of between about 150 and 180 volts. Power is connected via a 6 pin DIN connector, the internal connections in the plug selecting the heater voltage.

The four tubes used provide 6 active sections, all are triodes with the exception of the 6BH6 pentode. The regeneration is controlled using an adjustable DC voltage on the 6BH6 screen, it's cathode being connected to the tap on the Toko coil. I tried the 2 main connections on this coil both ways round to get the best regeneration operation as the tap is not in the centre of the winding.

During testing I obtained a sensitivity for the regen (using a signal generator on 3.1 MHz) of about -120 dBm for a 10 dB signal to noise ratio, a very good result. This S/N ratio assessment was carried by eye on an oscilloscope; a 10 dB ratio (in power terms) being (approximately) a factor of 3 in voltage terms.

I use this set with the low power valve PSU I built and detail here . I power the set from the 180 V line which under the modest load of this set falls to about 165 Volts. I use the regulated 12 V DC line to power the filaments rather than the 6.3 V AC available in an attempt to reduce power line hum and also to keep the temperature of the valve heaters independent of variations in the supply line voltage.

40m regenerodyne, rear 3/4 view

Front panel controls are volume, regeneration, input attenuator and tuning. There is no antenna trim control, the input circuit was "peaked" during testing using a trimmer inside the unit. The back panel connections are: antenna, power, speaker and headphones together with an earth terminal.

The idea of putting the valves inside the unit (as opposed to sticking out of the top) was so that it could be transported easily without risk of damage. The springs that can be seen in some of the photos keep the valves well pushed into their sockets. I made a lid for the unit using perforated metal sheet which allows for a flow of cooling air and also allows the faint glow of the filaments to be visible from outside.

One disadvantage of the internal valve design is that it takes around 30 minutes from switch on before the set stops drifting. I think the drift is caused by regenerative circuit tank components changing their values as the set warms up. Once the set is warm it's actually pretty good so if I know I'm going to use it I often turn it on some time before hand so that it will be stable when I get round to using it. In future projects I'm not sure I'd put the valves in the case again but it was probably worth it for this radio as it is pretty small.

The first listening test proved both encouraging and disappointing. SSB stations were heard from Germany, Russia, Scotland and Spain in about 20 minutes of listening. Problems identified were lack of filtering on the front end allowing out of (ham) band signals through to the detector and also an overall lack of sensitivity.

40m Regenerodyne receiver circuit diagram

I replaced the second Toko coil I'd used for the input filter with a home wound toroid and altered the turns ratio to provide a greater impedance transformation and so raise the design Q of the input circuit. A better design would be to have two loosely coupled bandpass filters but there is just not enough space in the unit to allow this.

In my opinion an input attenuator is just about essential in regenerative designs, it allows adjustment of the signal level into the regenerative detector which can be juggled with the regeneration control to give the best signal recovery. In this set it also has the side effect of eliminating the broadcast band break through. Only a little attenuation is required to eliminate all background stations. Once these are gone there is just a little noise from the set together with the signals of interest.

After some testing the mixer bias and mixer to regen connections have been optimised but tests with the signal generator reveal that the set is now still only about as sensitive with the mixer as the original tests showed the regenerative stage to be on its own. It should in fact be more sensitive (because the mixer should provide some gain) but because the set it probably sensitive enough for 40m operation and further development is extremely tricky I have decided to leave it as is!

My first evening's listening with the optimised set (actually about 4 AM in the morning :) produced some good signals from South America (I live in the UK) and other DX and so I was pleased! I left the attenuator set for a small reduction in signal, in the daytime this is not required because the unwanted signals are much weaker.

In summary I think the regenerodyne architecture has much to offer for such a simple arrangement and I'd like to build another, larger set and spend more time optimising the performance. It would be possible to make a really good receiver using this approach but considering the small size of the current design it is still mission accomplished.

back to top

Single valve eighty meter (3.5 MHz) transmitter

This circuit was inspired by a re-print of a German WWII single valve crystal controlled transmitter in the "Technical Topics" section in the RSGB "Radio Communication" magazine.

One evening I decided to knock up a derivation of this circuit using an 807 tetrode. The 807 will dissipate 25 watts and so with only the 5-10 watts input I envisaged for this circuit I hoped it would be robust and un-flustered.

Single valve eighty meter (3.5 MHz) transmitter

You'll see from the picture the unit is no work of art and also as it is not enclosed offers a risk of electric shock. I will re-build this sometime, probably using a 6L6 tetrode and a proper aluminium box.

The 6L6 is a close electrical equivalent of the 807* but because the anode connection is via the base (together with the other connections) it's easier to enclose all the live parts without having to build an enclosure for the entire valve.

The circuit employs a parallel resonant tuned circuit itself in parallel with the 807. Two extra windings on the coil former are used, one provides feedback to the grid of the 807 via the crystal and the other is the output winding, used to extract the RF power from the circuit.

By using 2 separate windings to control the feedback and output coupling the circuit could be optimised fairly easily to ensure reliable starting, low "chirp" with good output.

Small crystals plug into the unit using a section of IC socket! I have rocks for 3.555, 3.560, 3.565, 3.579 and a non CW segment frequency 3.686 (Baud rate crystal). I tested one of my 3.579 crystals using HT up to 500 volts (and obtained outputs up to 20 watts in the process) but the crystal remained OK so far as I could see. I therefore figure that the crystals should be OK at lower powers.

Keying is direct, via the cathode. Not (that) safe but certainly expedient! The transmitter sounds best at about 5-7 watts output which makes for a reasonable signal.

Below is a table of input voltage and current for RF outputs ranging from 1 to 10 watts. I'm not sure how accurate my RF power meter is (it's part of an inexpensive ATU and SWR meter) but I hope it's at least a reasonable guide to the power.

The efficiency values quoted at about 60-65% are what I would expect for a class C oscillator. The output is surprisingly clean with the strongest harmonics being at least 35-40 dB down on the carrier even without the low pass filter I added "just to make sure".

power out (W) voltage (V) current (mA) power in (W) efficiency (%)
1 116 13 1.5 66
2 166 20 3.3 60
3 197 25 4.9 61
4 225 28 6.3 63
5 250 32 8.0 63
6 270 35 9.5 63
7 295 39 11.5 61
8 315 42 13.2 60
9 335 45 15.1 60
10 350 48 16.8 60

*Actually the 6L6 has reduced ratings for power and voltage as a result of the smaller glass envelope and insulation limitations of the base design.

back to top

Eighty meter (3.5 MHz) Double Side Band transmitter

My first attempt at a Double Side Band transmitter was not a success. The reason was simple, I failed to breadboard the design first and allowed so little room in the case I chose for the project that there was no room to work round any problems in the design. Picking a small case is a great idea if you are sure the circuit you are building will work as is or will be simple to improve upon with minimal extra components being required :)

After making a number of CW (Morse) transmitters I decided it was about time I built a speech transmitter. Thirty or more years ago a simple speech transmitter would probably have used Amplitude Modulation; these days it is more customary to construct simple "DSB" speech transmitters.

80m DSB rig, top view

Double Side Band Suppressed Carrier (most often referred to simply as DSB) transmitters have two disadvantages, the bandwidth required for transmission is twice that of an equivalent Single Side Band signal and with conventional SSB receivers only one half of the transmitted DSB signal is actually demodulated resulting in a waste of half the transmitted power on the unheard sideband.

Despite these moderate disadvantages DSB transmitters are a lot simpler to build than SSB transmitters and still allow direct inter-working on the air with SSB stations. SSB is used for the majority of Ham radio speech transmissions at present, at least on the High Frequency bands.

DSB is easily generated, an RF carrier wave signal is fed into a Balanced Modulator and the speech signal used to modulate it. A Balanced Modulator has the particular property that it will let one of the signals pass from the balanced input port to the output port except when a second signal is applied, effectively un-balancing the modulator.

DSB(SC) and SSB(SC) are actually still forms of Amplitude Modulation but specifically don't transmit a carrier wave along with the sidebands and the carrier conveys no information on it's own, it's simply used to assist demodulation in receivers that do not have the means to re-insert the carrier locally. A full treatise of these forms of modulation is beyond the scope of this site but the internet will provide a lot of very erudite text on the subject.

To take advantage of the relative simplicity of DSB generation it's customary to generate the DSB signal on the desired transmission frequency so that no mixers and heterodyne oscillator(s) is/are required.

Making a stable Variable Frequency Oscillator isn't a trivial task although I felt fairly confident I could make something stable enough to operate on the 80m (3.5 MHz) band. The concept sounded reasonable, use an ECC82/12AU7 for the Variable Frequency Oscillator and cathode follower (buffer) stage, an ECC81/12AT7 double triode as a Balanced Modulator and a 6L6 beam tetrode for the Power amplifier. An ECC83/12AX7 was to be used as a two stage microphone amplifier.

In the interests of minimising component count and size I decided that an untuned (or "aperiodic") matching scheme between the BM and the PA would be sufficient. From the pictures you will see I chose a small chassis and laid the three double triodes and octal 6L6 out together with a 150B2 Voltage Regulator Tube. The underside view obviously shows the only part wired circuitry.

80m DSB rig, underside - only partially wired

I managed to get the VFO working quite well but could not get the ECC81 to produce anything like the drive the 6L6 required to produce the target 15-20 watts PEP I hoped for. I substituted the ECC81 BM for a 6ME8, an unusual "beam deflection" tube, and rewired to suit. [The 6ME8 is similar to, but not directly compatible with, the famous (and now rare) 7360 and both make excellent Balanced Modulators and Product Detectors.]

The 6ME8 made a much better BM but even with a tuned output circuit it only just produced enough output to drive the 6L6 to the desired output. At the time I corresponded with Andy G4OEP quite a bit as he was using the 6ME8 as well. Andy had come to the conclusion that a buffer stage was generally needed.

With the addition of a suitable buffer amplifier the unit I was making would have probably worked well but I'd painted myself into a corner by then and had no room for an additional stage (or even a proper way of mounting the extra tuning capacitor) - I gave up at that point.

All the components were recovered, only the aluminium case was thrown away. The failure of the project was a disappointment but I learned not to "hope your way through the design". The BM in a DSB rig is the key part and I'd never built one before!

back to top

One-sixty meter (1.8 MHz) Hartley transmitters

I built three versions of a 160m Hartley, dismantling the old versions as I continued the development.

The first Hartley was a "lash up" on a piece of wood and used a 203A triode. The second was built using an aluminium test chassis, and used to gain experience and evaluate the circuit prior to building a more permanent version. The latest incarnation (Feb. 2002) is a tidier version using a single 812 triode on an oiled pine base.

I've detailed these transmitters below in reverse chronological order. All shared the same tank circuit I made from a wide spaced EF Johnson variable capacitor and a ceramic coil former wrapped with heavy gauge insulated plastic wire. The variable capacitor has a 6:1 slow motion drive on it which makes the frequency a little easier to set!

The tank circuit tunes quite a wide range of frequencies and covers all of the 160m and 80m Amateur bands. My hope was I could use it on 80m as well as 160m. Limited tests to date have indicated it's too drifty for 80m use unless your QSO partner knows what you are using and appreciates the pitfalls and joys of simple LC power oscillators.

Many people have built substantially better performing Hartley transmitters and I still feel I have "unfinished business" with the implementation of these devices. I have decided however that the next LC power oscillator I build will be of "Tuned Plate Tuned Grid" design.

160m/80m Hartley Test rig (813 and others ...)

testing position

This works best generating about 10 watts of RF, obtained from about 700 volts HT. The same circuit produced 100 watts of RF output after optimisation for maximum power, and the use of a 1700 volt supply! The efficiency was estimated at around 60% in both cases with harmonics at 36 db (or greater) below carrier power.

The two valves lying on their sides are OD3 voltage regulators. Used in series, they clamp the screen voltage to 300 volts (key up). When the key goes down, the screen voltage falls and the purple glow was extinguished.

Most, if not all, of the original Hartley oscillator transmitters employed triodes, if only because "screened grid" tubes such as tetrodes and pentodes had yet to be developed, or at least come into common usage. I wanted to try a tetrode in a Hartley because   a) I had some of the required power rating   and   b) the additional grid allows some extra control of the operating conditions of the valve.

testing position

I experimented with the screen grid voltage but found the circuit worked best with relatively low operating screen voltage, this being obtained simply from the anode HT voltage via. a high value dropper resistor (50 or 100k).

I was not happy with the 813 in this circuit. It was not possible to eliminate a noticeable Frequency Modulation imposed by the AC heater current, despite trying several methods to null it out. The FM on the carrier made the received CW note sound "dirty" or "raspy" not like the pure tone that is pleasant to receive.

By using some crude adapters I was able to try an 811A, 203A, 807 and 4-125 in the same set-up. All the directly heated valves (that is all those mentioned except the 807) had FM problems, although the 811A and the 4-125 were better than the 813 and 203A in this respect.

Whilst the 807 was FM free, it was very sensitive to changes in the HT voltage. Noticeable and unpleasant "chirp" resulted from keying, even with a fairly stable PSU and a stabilised screen voltage. The overall winner by a clear margin was the 811A.

FM spectrum from the AC heater powered 4-125

Shown here is the output spectrum obtained from the 4-125 tetrode operating at 300 volts HT. The sweep is centred on the carrier. The multiple sidebands clearly show the classical spectrum of an RF carrier, frequency modulated by a single tone. [The trace wasn't really as bright as the picture implies - it's just a feature of not using the flash.]

The transmitter output was connected to a special dummy load that has a -40 dB output to monitor signals. I generally connect this monitoring o/p via an additional, inline, 20 dB attenuator to the HP141 spectrum analyser input. It is all too easy to damage the sensitive mixer on the input of a spectrum analyser with large signals.

I was not able to get this test chassis running well enough to simply build it into an enclosing case complete with PSU. I decided that at the voltages required to operate this circuit, enclosure was an absolute requirement for regular use. A few months after this testing, the parts were re-assembled to become the 812 Hartley detailed on the projects page.

160m 203A Hartley

The 203A Hartley lash up. It worked well!

The 203A triode used turned out to be rare but many other triodes would work well in this application. It was picked as it was available and could dissipate more power than was required. You'll notice that this rig was a complete lash up but it worked never the less and (perhaps) more importantly it whet my appetite for the "open style" of simple transmitter construction.

This transmitter produced a just acceptable 10 watt output but would have been better with perhaps just 2-3 watts output. The circuit was tested up to 780 volts and at that voltage produced in about 15 watts on 160 M and 22 watts on 80m although at these outputs the circuit was far too drifty to use.

The inductor wound was too small (at only 16 uH) to provide a reasonable loaded Q on 160m. It did resonate on 160m with the appropriate value of capacitance (about 400 or 500 pf) but actually gave a larger output on 80m where the tuned circuit was better matched to the valve. The drift was actually about the same on both bands, the better match on 80m compensating for the increased frequency.

I decided to build a bigger version in a metal case and so - regrettably I dismantled "Hartley 203A" to use the parts for the 813 (& others :) test Hartley detailed above. After unwinding the coil, I ended up winding an almost identical one again in the hopes of covering 2 bands!

back to top

Power Hartley transmitter using a 1 kW GM100 triode

This project was inspired by a sense of wanting to build something a little unusual rather than with the expectation that it would become a useful transmitter. At the time of writing I'm at the breadboarding stage and have yet to decide if I will keep the unit in its current form or dismantle it to re-use some of the parts in other projects.

The GM100 power Hartley with a 12AT7 and 3-500 for company

One of my friends egged me on to build this device and although the rather large size of the valve is more his enthusiasm, I did get to wondering if I could actually use this on air.

The valve is a Russian made "GM100" directly heated triode. This valve will withstand anode voltages of up to 5 kV, pass currents of up to 1.6 amps (!!!) and has an anode dissipation of 1 kW. The filament takes a massive 18 amps at 17 volts although the only slightly suitable transformer I have drives the heater to about 15.5 volts which is still pretty bright.

To give some idea of the scale, the tiny tube on the left (by the multimeters) is a 12AT7 double triode, the larger tube on the right hand is a 3-500Z, a modern 500 watt triode.

The variable capacitor and inductor shown in the photograph are moderately big components, at least by UK ham standards, but are dwarfed by the huge tube. Having "short" wiring is obviously quite difficult on something so large but ferrite beads on the grid and anode leads seem to have prevented parasitic oscillations.

To run this tube at full power much larger components would be required and the wooden frame would probably be a fire risk! The whole rationale behind this transmitter is to under-run the tube in the interests of frequency stability and longevity although the heater power consumption is a little excessive.

Forming the PSU capacitors, each at about 150V in this shot.

The circuit I've used for this is identical to that of the 812 Hartley shown further down the page. At the time of writing, 20 watts of RF has been obtained at 1.9 MHz with a 730 volt supply drawing about 55 mA and a good 80 watts output with 1300 V and a little over 100 mA.

The harmonics are -30 dB WRT to carrier (or better), not a great figure but ok with a low pass filter or at a low power. Like the 812 Hartley, the output seems best taken via just a single turn link winding so as not to overdamp the tank circuit.

I've a fair amount of work to do to make this into a viable radio. Firstly I have to ensure the frequency stability and tone are good enough to use on air and secondly I have to complete the PSU circuit; at present it's a bit of a lash up.

The unit will probably operate best using 1 or even 2 kV HT and so some form of enclosure or restricted access is (obviously) essential. Most of the experimenting has been done with only 100 or 200 volt HT.

back to top

The variac

My 1 amp variac. The terminal posts encourage a 'lab' type approach to the mains supply, not to be recommended!

One of the most useful items in my "shack" is my 1 amp variac. For those that don't know a variac is an adjustable transformer allowing one to vary the AC voltage on it's output. I think the word Variac is a trade name.

Most variacs with be adjustable from 0% to about 110% of the input voltage. In the UK, for example, my variac will provide between 0 and around 260 V RMS. Being auto transformers variacs do not provide isolation from the mains supply and so should be used in conjunction with an isolation transformer if an isolated adjustable supply is required.


back to top

back to main site index