Valve projects page

I'd really like to be able to list two dozen, or perhaps at least 10 completed projects on this page. Unfortunately I rarely seem to complete, or even start, most of what I envisage.

The items detailed on this page are those projects which did get completed to a level where they work well enough to keep them intact. There are features on all I'd like to improve but at least these projects did get to completion, rather than never getting past the thinking or building stages.

I have several more thermionic projects on the shelves (and even more in my head) waiting for some time to develop them. Hopefully this page will be added to over time; that's partly what it's about. The projects detailed here are described further down the page and linked from the list below.

These are roughly in reverse chronological order, i.e. the newest at the top.

The 812 Hartley transmitter is not enclosed - this approach should be used with caution for obvious safety reasons. Please read the comments on safety and the site disclaimer here if you are tempted to try anything like this.

I have made a number 'lash ups', test set-ups and also projects that did not work well enough to keep. I recycle the parts used in all of these and sometimes write them up for this site. You can see these experiments detailed here.

As a background task I've been building quite an ambitious project from scratch, a 'forty meter' (7.0MHz) SSB transceiver which I've called the TR40. This project has its own dedicated page here.

The TR6C4 single triode 3.5 MHz transceiver

TR6C4 one tube transceiver, front view

This design, which I called the "TR6C4", switches a single 6C4 triode between transmit and receive duty to make a complete transceiver. Three poles of the four pole change-over switch are used for the triode's grid, cathode and anode/plate connections. The fourth pole switches the hot side of the tank circuit as required.

The receiver configuration is a DC controlled "Hartley" regenerative detector (using a cathode tap on the inductor) with Hi-Z phones for the anode load. The transmitter is crystal controlled "grid-plate" (I'd call it a Colpitts) configuration with the tank circuit in parallel with the valve (shunt fed) and an RFC in the triode's anode circuit.

The absolute MDS is about -105 dBm with Hi-Z 'phones, usable is more like -80 which is a nominal "S9". These are pretty much the same figures I got whilst breadboarding. The tuning range is [into a 50 Ohm attenuator and sig gen] is 3480-3715 which is ideal.

On my antenna it is affected by the resonance (or lack of it) of the antenna but seems to be able to cover a similar range, perhaps offset by some amount depending on where the antenna is tuned. The unit requires 6.3V @ 150 mA and 170 V @ about 20 ma (TX) and 2 mA (RX). The transmit output is approx 300 mW.

TR6C4 one tube transceiver, back view

There is no slow motion drive but the tuning rate seems ok. A trimmer capacitor is switched in during receive and has been set so the "dip" in anode current at resonance when transmitting corresponds to the corresponding receive frequency (more or less).

After calling the tuning knob is moved a little either side of the best dip position during reception to check for callers. I can hear people on and around 3650, the QRP calling frequency here. I don't doubt this rig is capable of some contacts but it's going to need some patience :)

I have not yet produced a combined schematic showing the switching but the transmitter and receiver circuits used are shown individually: receiver and transmitter

back to top

A single ended 807 audio amplifier

This amplifier uses a single 807 beam tetrode in the "Single Ended" Class A audio amplifier configuration. Up to 4 watts of audio can be produced before distortion becomes evident. The output stage uses a switch adjustable self bias arrangement where a cathode decoupling capacitor can be switched in or out of the circuit together with a choice of two resistors giving either 35 or 45 mA standing current.

807 AF amp, front view

I was hoping for an undistorted 5 watt RMS output with a 15 or 20 watt power dissipation by the 807 but I initially struggled to get more than a couple of watts without distortion from my chosen HT voltage of around 325 volts. I suspect this voltage is too low compared with the 250V screen voltage for the tube to operate linearly. A specifically designed "kinkless" audio tube such as the KT66 or KT88 would be a better choice in this application of course as would a greater HT voltage.

After much experimentation with bias and connections on the 807 I opted for a slightly odd looking arrangement where the screen grid is decoupled to the cathode but the cathode is connected to ground via an unbypassed resistor. This provides a lot of negative feedback to the stage which helped remove the distortion but increased the voltage that the driver tube must produce to realise the full output. It also gives away approximately 25 volts of the precious HT voltage.

With the cathode decoupling capacitor switched out the output stage has considerable degenerative negative feedback, with the capacitor switched in no NFB is applied to the stage. The resulting sound is quite different. I use the circuit with NFB included when I'm using this amp to drive slightly unpredictable loads, for example when it is being used as a cathode modulator.

Two ECC83/12AX7 double triodes are also included on the chassis. Three of the 4 triodes provided by these tubes are used as pre-amplifier stages. One of the three outputs is selected by a 3 way switch and drives the driver triode which in turn drives the grid of the 807. Twelve volts DC is used on the heaters of the triodes, the 807 uses 6.3V AC.

807 AF amp, underside

A variety of connectors are present on the back panel so that a stereo CD player, Electric Guitar or Hi-impedance microphone can be used with the unit. A simple volume control is provided but there is no provision for tone controls.

This unit works quite well and although by no means powerful will drive the loudspeaker in my shack to an uncomfortably loud level if pushed. No hum or noise is evident at regular volume settings in the absence of an input signal - better than I expected.

I may alter the design to used a fixed bias configuration (using a negative supply) and directly ground the cathode of the 807. Negative feedback could still be applied most probably by taking the a portion of the output voltage (from the transformer secondary) to the cathode of the driver stage. Correct phasing is essential to ensure that feedback is negative as positive feedback might well result in oscillation.

back to top

A multi-rail power PSU for valve circuits

The small PSU I built some years ago is ideal for powering small receivers and other very low power projects but isn't powerful enough for larger projects such as transmitters and audio amplifiers. I decided I needed to make a more powerful general purpose PSU rather than have to build an internal PSU in each new project as it came along.

Multi-rail PSU, front view

After some thought it became obvious that I should simply aim to make a higher power version of the original small PSU as it had (mainly by chance :) provided a good selection of voltages in the first place. Sometimes you never see the obvious staring you in the face...

I had already used the 8 pin "International Octal" valve socket for the output connector on the low power PSU and so decided to try to maintain compatibility with this. By using a suitable multipole connectors on the PSU's output, each project's power lead can be terminated in an appropriately wired plug and so "tap off" whatever voltage(s) it requires.

Fairly obviously different projects require different HT voltages. A 10 or 15 watt transmitter might require 300-350 volts where as a receiver will often work well using a supply around 200V and a small regenerative receiver driving headphones will require less than 100v.

I have used a 100VA 240V to 0-60-120V transformer in this PSU. Using this transformer it's easy to generate 85 and 170 volt DC rails by grounding the 0v tapping and using single semiconductor diodes to half wave rectify the 60 and 120 AC respectively. The output DC voltage will be the peak AC voltage with a light load and low impedance rectification.

Using a voltage doubler fed from the 120v tapping I was also able to generate a nominal 340 rail (I.e. 2 x 170). I also decided that a negative bias supply might be useful and elected to make this a switchable -85/-170 volt rail.

Multi-rail PSU, rear view

So in all the unit produces four HT voltages, -170/-85 switchable, +85, +170 and +340. In all, just 5 diodes, 5 electrolytic capacitors and 5 bleeder resistors are used to generate these four HT rails. Each HT rail is fused.

In addition two more transformers are used to provide the heater and filament supply. A 6V @ 4A x 2 windings provides two nominal 6.3V AC @ 4A supplies, one for each pair of Octal sockets. In addition a 60VA 18V transformer feeds a bridge rectifier, smoothing capacitor and pair of 12V @ 2A regulators, each one feeding a pair of Octal sockets.

With the exception of the -170v DC option these voltages are the same as the small home made PSU and so the Octal sockets on the back of this PSU are wired identically to the smaller unit. This allows low power projects (like receivers) to be plugged in and powered from either PSU. Larger projects like the 3 band MOPA transmitter and the 5 watt 807 audio amplifier can only be powered from this larger PSU.

I decided not to try to provide voltages above 340 volts in this PSU as I have half finished high power multi-rail PSU that will supply up to 1kV - it seemed silly to replicate this effort. In addition those projects (at least for me) are moderately likely to include inbuilt power supplies.

Like the low power PSU, two switches on the PSU control the output voltage. The mains or line power switch applies power to the three transformers in the unit and illuminates and orange neon.

Multi-rail PSU, top view

A four pole switch is used control the four HT voltages and when these are "on" a red neon is illuminated. By having this additional HT switching the dangerous HT voltages can be disconnected from the PSU whilst the heaters are still kept power. Although I do not absolutely rely on this it is a useful additional safety stage.

The voltage metering works on the output of the HT switch and so if any bulk decoupling in present in the project(s) in use this residual voltage

Again, with one meter and a 4 way switch this is not something to be relied upon but is a further help. The meter can also be used to monitor current on each of the HT rails with two ranges being provided on the +170 and +340V rails.

Looking at the top view with the lid off you can see the 18V, 0-60-120 and 6V x 2 transformers left to right with the metering circuit on the 3 small tag boards at the top of the picture. I built the case for this from sheet metal and although not perfect looks tidy enough in the shack.

back to top

Three band (80/40/30m) crystal controlled MOPA transmitter using a 6AG7 and 6L6

This transmitter is a version of the classical 2 section "Master Oscillator Power Amplifier" design using a 6AG7 pentode in a crystal controlled Colpitts oscillator with an aperiodic (un-tuned) anode load to drive the 6L6 beam tetrode Power Amplifier with a band switched "pi" network output.

3 band 6AG7/6L6 MOPA, front view

The unit covers the 3.5, 7 and 10 MHz bands with approx 12, 10 and 9 watts Continuous Wave output from a nominal 325V HT rail. This isn't much from the 6L6, but the unit is stable, under run and it seems to be plenty enough for solid CW contacts.

I quite fancied a slightly 'retro' look and so decided to use octal based valves. A fair period of breadboarding followed in which various oscillator tubes and configurations were tried.

I started with a 6J5 triode in the often used 'Pierce' configuration but this did not provide particularly good performance for me suffering from 'chirp', I think this was caused by the crystal warming up due to the current passing through it.

I decided I need to minimise crystal current (and so crystal heating) while maximizing the output and tried a number of different oscillator configurations and tubes. The clear winner in terms of output level and keying tone was the 6AG7 pentode and the 'Colpitts' oscillator, configuration sometimes referred to as a 'grid-plate' oscillator.

The Colpitts design allows the feedback to controlled by the ratio of the cathode to ground and cathode to grid capacitors. Use of a screened grid valve (a tetrode or pentode) additionally allows the implementation of an 'Electron Coupled Oscillator'.

The ECO configuration takes the output from the anode of the oscillator tube whilst the screen grid is fully decoupled to RF helping to isolate the output circuit from the action frequency determining components.

3 band 6AG7/6L6 MOPA, top view

A very good article about ECO oscillator design and the choice of oscillator valve can be found here and here. The 6CL6 valve referred to in the above article is the miniature 9 pin based glass version of metal octal 6AG7 I used; I picked the 6AG7 for the reasons the article cites.

Because the unit is always used with a 50 Ohm antenna load (via an ATU) the "load" capacitor and coil tap points can conveniently be switched for each band as required leaving only a single "tune" capacitor to be adjusted for resonance in the pi network.

In theory the PA stage can be used as a frequency multiplier (by suitable output tuned circuit adjustment) allowing a single crystal to cover more than one frequency band. For example 3.510 could be used to transmit on 3.510 or 7.020. In practice the output spectrum is not sufficiently clean to with this rig to allow this, a tuned load in the anode of the crystal oscillator would be essential.

A four pole three position mode switch performs T/R switching for the transmitter, antenna and a receiver and also drives a "ground to transmit" control line for a external amplifier. There is also a "net" position and this runs the oscillator (but not the PA) allowing the transmit frequency to be located on the station receiver.

Although it's possible to key only the PA and leave the MO running I quite like keying the MO as well as it means you can use the station receiver as side tone. To maximise frequency stability during keying both the screen and anode voltages of the 6AG7 are stabilised by use of two "VR" tubes in series, 150V for the screen and 240V for the anode.

3 band 6AG7/6L6 MOPA, rear view

Originally designed for CW only the unit has been used to send speech using Amplitude Modulation by cathode modulating the PA. To allow this, the oscillator can be switched (back panel) not to be keyed with the PA but run all the time when in transmit mode.

The unit requires 6.3V @ 1.5A for the heaters and 300-350V @ 100mA for the HT. Connection to the PSU is via a 4 conductor cable which terminates in an Octal plug suitable for the larger of the two power supplies shown on this page.

Although cathode modulation only permits a resting carrier level of 25% (or less) than the full CW level it is very easy to retrospectively apply to CW transmitters that key the PA cathode. My external cathode modulator simply plugs in where the Morse key normally does.

By following the transmitter with one of my commercial linear amplifiers a 3 watt carrier can be boosted up to a respectable 30-50 watt carrier level (depending on amplifier gain) and with 100% modulation will reach approx 120-200 w PEP.

back to top

One-sixty meter (1.8 MHz) Hartley transmitter using and 812A triode

I like simple circuits and a "Hartley oscillator" transmitter is probably about as simple as you can get, at least in part count terms.

From what I've gathered, Hartley transmitters were in vogue in the 1920s and perhaps the early 1930s. Generally they employed a single active device although two devices could be used in "push-pull". All Hartley oscillators feature a tapped coil, either in the cathode or anode circuit, the transmitting Hartleys employing a single valve have the tuned circuit connected to the anode.

812 Hartley, front view

I chose to build a Hartley as it can be configured not to use an RF Choke, (which I'm short of here) and it also seems to have been the most popular early power oscillator circuit.

Hartleys using an RFC in the anode circuit are referred to as "shunt fed" and this configuration does have the merit of allowing the tank circuit components to be a 0 volts DC, rather than at the HT voltage.

The received wisdom for simple Ham band Hartleys like this is to aim for a much smaller output power than the same device would be able to provide when used as an amplifier. The approach may limit frequency drift due to heating effects and should also make the device harder to damage.

In total I built three versions of a 160m Hartley, dismantling the old versions as I continued the development. The final version is detailed here, the previous versions are detailed in reverse chronological order here.

The final incarnation (Feb. 2002) is a tidier version using a single 812 triode on an oiled pine base. All of these Hartleys shared the same tank circuit that 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!

812 Hartley, top view

The tank circuit tunes quite a wide range of frequencies and covers all of the 160m and 80m Amateur bands. My initial hope was that I could use it on 80m as well as 160m. Limited tests indicated it's too drifty for 80m use. On 160m this transmitter produces a just acceptable 12 watts output from 750 volts HT but is quite nice at 4-5 watts output which is obtained from around 400 volts HT.

With a 1300 volt supply, 50 watts was produced with 90 mA being drawn but slow chirp was quite noticeable. I think this was caused by the valve heating up when the key was down and it's inter-electrode capacitances changing, shifting the frequency slightly. Key down the frequency was pretty stable but if the transmitter does not sound nice when keyed the extra power is of limited utility when establishing contacts.

The schematic of this transmitter is shown below. The basic Hartley oscillator is composed of the coil, 500 pF variable & 470 pF fixed capacitor, 10 k grid leak resistor and the 812A triode. The coil former is about 40 mm (1.5 inches) in diameter.

The 1 nF and 10 nF capacitors are for decoupling and the 33Ω resistor and ferrite bead (FB) are to ensure no parasitic oscillation takes place.

A centre tapped filament transformer was not available and so the two 27Ω resistors were use to provide the common point for keying. A two pole three position switch is used to select transmit, receive or net operation. In the net position a high value resistor causes the circuit to oscillate at low power so the frequency can be adjusted whilst listening on the station receiver.

812 Hartley transmitter circuit diagram

The signal marked "T/R control" controls an external transmit/receive relay which is used to switch the antenna between the transmitter and whatever receiver is being used. All my standalone transmitter projects include this "ground to transmit" control signal so they can be used with the switch over box. In the net position the transmitter is not connected to the aerial and so no signal is transmitted on the air.

I used Microsoft Paint to draw this diagram (as a .bmp prior to saving it as a .gif). The symbols are from the "TubePad" library produced by Gary, WD4NKA (linked above and from the links page).

There are only two wires that connect the top of chassis to the underside. The red wire connects to the tapping in the coil and feeds the HT into this circuit. The black wire connects the grid end of the tank circuit to the grid (via the under chassis coupling capacitor). Both cables use plastic insulators made from "rawl" plugs.

Also visible the top view picture is a black wire looped around the coil. This is the single turn output link winding. A piece of coaxial cable comes up through the centre of the wooden dowel and connects to an SO239 coaxial socket on the back of the unit.

I need to brush up on my Morse skills to make reasonable use of this transmitter. Being a very un-ambitions DX operator even short distance contacts are welcome, especially when using a transmitter with just a handful of components in it!

Even on 160m unless your QSO partner knows what you are using and appreciates the pitfalls and joys of simple LC power oscillators they may not be particularly impressed! 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.

back to top

Twin double-triode regenerative 80m receiver

Twin double-triode 80m regenerative receiver

I built a regenerative receiver as a kid using a single MPF102 JFET and the range of Denco "green" plug in coils. It didn't work well as a Ham radio receiver although I used it for years on the Medium Wave and 49/41 meter broadcast bands.

Since getting interested in "glowbug" transmitters and receivers I read much about the potency of the regenerative receiver and I hate to admit that I didn't believe that these receivers would be particularly good. Boy, was I was wrong!

After some deliberation I decided to build a regenerative receiver using a pair of the ubiquitous 6SN7GT double triode tubes. I decided to err on the side of caution as I believed that having 4 sections to play with would mean if my radio was not as optimised as it might be, it would still provide reasonable performance.

I wanted the luxury of a front end amplifier (to isolate the detector from the antenna a little and vice versa) and also the ability to drive a loudspeaker, not just headphones. I decided to concentrate the coverage on just one frequency band, 3.5-3.8 MHz, the 80m band.

regenerative 80m receiver, back view

The slow motion drive I used only offers about a 8:1 reduction ratio giving the unit a tuning rate of about 75 kHz a turn, much too high by accepted standards. The radio is usable none the less, providing the operator has not consumed too much beer at the time of operation!

The pictures show a slightly unconventional construction approach, the receiver is constructed in a "chassis" made of double sided copper clad board. Although this may look a little strange, it is easy to work and is ideal for circuits undergoing development

The RF amplifier stage is in the grounded grid configuration - this approach together with the low mu (gain) 6SN7 provides a stable amplifier with only a little gain.

I did try the higher gain 6SL7 valve in place of the second 6SN7 (which is used for the audio pre-amp and output stages). I had to alter the cathode bias resistors for both of the 6SL7 stages to get the correct standing currents. The unit did give a more audio output but was prone to oscillation. Although I could probably have cured this with a capacitor in the right place I decided that it was nicer to use 2 identical valves in the circuit and replaced the 6SL7 with the original 6SN7 and restored the previous bias circuitry.

regenerative 80m receiver, underside

I run the audio stages of this circuit from around 220 volts. A lot of regenerative circuits often use only a few tens of volts on the detector although my detector is fed from about 80-90 volts. In the next version I make will follow the conventional wisdom a little closer - perhaps also accepting headphone only operation with 1 valve.

I've heard a lot of contacts on this radio. Generally, it can hear anything that my Yaesu FT200 transceiver (my usual radio for 80m) can hear*. When I heard VK on the 'genny the European stations in QSO with him were really struggling to work the Australian. I'd have been struggling a little too, but no more I think. The thrill of looking at 50 year old valves glowing and bringing in a signal from the other side of the world is quite something.

* I've since started using a Drake R4C on 80m. This radio can hear things that neither the 'genny or the FT200 are able to receive. This is a dynamic range and filtering issue, rather than one of absolute sensitivity.


I borrowed an RF signal generator and using this I was able to measure the MDS (or Minimum Discernable Signal) level for this radio. The signal generator's carrier could be heard clearly at a level of -110 dBm. It was even (very) faintly audible at -120 dBm, that's 1 x 10-15 watts or a quarter of a micro volt signal into 50Ω!

On the 80m band this sensitivity is more than acceptable (at least at this location) because of the high level of noise present on the band. The FT200 has an MDS of -140 dBm at 3.8 MHz and -135 dBm at 28.5 MHz.

back to top

Low power PSU for valve circuits

Low power valve PSU

Having built the regenerative receiver I decided that custom low power valve Power Supply Unit (or PSU) would be of great value when developing such circuits.

I do have a trusty 6.3V 4A AC and adjustable 250-325V @ 100mA DC PSU but it is fairly large and doesn't provide lower HT voltages, 12V DC or any negative bias voltage.

A 12V DC rail is useful for powering relays and also the filaments of high gain audio stages where hum induced from an AC heater line can be a problem. Negative voltage rails are useful for both fixed bias configurations and also switching off individual stages, a technique often used for T/R switching in valve radio transceivers.

I decided that 6.3V AC, 12V, 150V, 300V and -130V DC would cover most applications and could be accommodated in a small case using the parts I had. The unit has metering for current and voltage on the -130, +150 and +300 volt outputs, these outputs are also individually fused.

The PSU contains 3 transformers: 240->6V 4A, 240->15V 1A and 240->12V 1A. The 240->12V transformer is actually used as a "step up" transformer with its "12V" winding connected to the 6V AC output from the first transformer. When used in this configuration its "240V" winding provides 120V AC.

Low power valve PSU, underside

This 120V AC output is rectified and doubled using a "full wave doubler" circuit. With the most negative point of doubler grounded, the configuration provides both the +150 and +300V DC HT outputs. The maximum allowable current from these 2 rails is a total of about 50 mA, ample for many small circuits.

The 15V transformer output is rectified, smoothed and regulated (using a 7812 regulator) to provide 12V DC at 1A. The 15V AC output is also multiplied using a ladder type voltage multiplier and generates the -130V DC rail.

In addition to driving the step up transformer, the 6V AC can be used to power nominal 6.3 volt filaments. The transformer used actually provides a little more than 6 volts (depending on load) and the output is within 10% of a nominal 6.3 heater supply for all reasonable output currents.

Two octal sockets provide duplicate access to all supply rails and a 4 pole switch allows the isolation of the -130, 12, 150 and 300 volt outputs. To turn the 6.3 volt output off you must turn the whole unit off.

Low power valve PSU, back view

The PSU works well except that the regulation of the 150 and 300 volt rails is very poor, falling by more than 25% from no load to full load. Putting two transformers effectively in series from a regulation point of view has the unfortunate effect of compounding the drop in output voltage with increasing load.

Even allowing for the two transformers, I still feel the HT regulation is poor. The smoothing capacitors are 330 μ and so at the low currents drawn are relatively huge. The resistance of the 120V output winding is 240Ω (which will appear as 480Ω due to the voltage doubler) and will therefore contribute a 24 volt drop at 50 mA output current.

I am currently considering what to do about the regulation. I may rebuild the HT circuit by disconnecting the 6V AC from the 12V winding, connecting the 240V winding to the mains and connecting the secondary to an additional voltage multiplier.

The 12V "1A" winding is actually two 12V 0.5A windings in parallel, connecting them in series would yield 24V RMS or about 35V DC per multiplier section. Using an 8 section multiplier with outputs at 4 and 8 taps voltages of around 140 and 280 volts would be produced.

The main issue here is fitting the additional capacitors in the box. Maybe there is room above the metering board, shown on the left side of the underside shot.

back to top

back to main site index