Single Tube SOTA Transmitter
A single tube transmitter is developed for portable and Summits on the Air (SOTA) use. The SOTA transmitter had to be relatively lightweight, run on a small battery and rugged and small for backpacking. The transmitter was interfaced to a QCX receiver, T/R switch, keyer, and RF monitor. And the high voltage is provided by small, efficient switching mode power supply.
This post is part of the series about a backpackable vacuum tube based transmitter teamed with a QCX receiver for portable SOTA activation. Other posts are the overview, concept and results, and detailed descriptions of the high voltage supply evaluation and QCX interfacing, modification and antenna switching.
The inspiration was an article from the March 1967 Electronics Illustrated edition, '1 Bottle Xmitter for 40', which had potential. The vacuum tube used is a 6AQ5A for which a 12AQ5A could be substituted for operation from a 12 volt battery. The schematic is very simple and made to be assembled with little expense. The circuit depends on inter-element capacitance for output matching and positive feedback for the crystal oscillator, this does not provide reliable operation. The transmitter will be modernized a decade or two for more dependable operation.
SOTA Transmitter Build
The design and build of the transmitter was split among three boards: Radio Frequency (RF) Oscillator and Amplifier, Low Pass Filter (LPF) and RF Monitor, and High Voltage (HV or B+) Control.
RF Oscillator and Amplifier
The oscillator and amplifier is implemented in the single pentode tube using an electron coupled oscillator design. The schematic of the oscillator and amplifier is shown in figure 2.
There are significant changes from the inspiration circuit (figure 1) to provide more reliable and enhanced operation. The changes include addition of C1 and C2 divider for oscillator feedback, RF isolation of the cathode (pin 2), biasing and bypassing of the screen grid (pin 6), high voltage (B+) blocking to the LPF by C6, isolated solid state relay (SSR) keying of the cathode, options for straight key or keying from an external keyer, and a 12 volt filament for direct powering by the battery.
In the electron coupled oscillator design, the oscillator involves the cathode (pin2), grid (pin 1) and the screen grid (pin 5). There is isolation and amplification by the plate (pin 5) from the cathode and grid. The electron flow from the cathode to the plate is the common element between the two stages, giving the circuit its name. The role of the screen grid is to provide isolation between the oscillator and the output at the plate, reducing the pulling of the oscillator's frequency with vary loads such as during keying.
To understand how the circuit works, remember the role of the bypass capacitors (0.01 uF) and the inductors (2.2 mH). The bypass capacitor is, in simple terms, a short at 7 MHz and an open at DC. Likewise, the inductor is a short at DC and an open at 7 MHz. Inductor L2 blocks RF from entering into the B+ (high voltage) supply and L1 keeps the cathode above RF ground. Capacitor C4 holds the screen grid at RF ground while capacitor C5 bypasses any remaining RF to ground, blocking it from entering back into the high voltage supply at B+.
For oscillation to occur, positive feedback is required. In this circuit a Colpitts oscillator is implemented with the voltage divider consisting of C1 and C2. The screen grid is RF grounded, the same potential as the bottom of C2, providing the path for positive feedback to the capacitor divider. In order for this to oscillate, the cathode can not be at RF ground; this is accomplished by L1. The crystal, Y1, sets the frequency of oscillation. The crystal is loaded by the series combination of C1 and C2, usually around 18 pF to 22 pF. The grid leak resistor, 47k R1, provides bias to the grid and a bit of leakage for the circuit to go into oscillation. The value for R1 is not critical, values between 33k and 68k may be used. The socket for Y1 is an old fashioned HC6 socket but the two pin header J5 allows a modern crystal to be used as well.
The oscillator is enabled by grounding point SK. This may be done by a straight key plugged into J3 or through an isolated SSR U1 or U2. The SSR is turned on by an external keyer connected to J4. The circuit board was laid out so that either U1 or U2 may be used depending on availability. Either SSR is enabled when grounded through J4. In the SOTA setup, the keyer output from the T/R switch is connected to J4 so that the QCX keyer is used to key the transmitter. Capacitor C3 bypasses any RF from the SSR or straight key and also provides some shaping to reducing key clicks.
The resistor at the plate, 100 ohm 1 Watt *R1, has 7 turns of #22 wire wrapped around it and connected at each end of the resistor. This is a small inductor in parallel with *R1. All the tube transmitters I have built has always had this resistor, often referred to as the VHF suppressor. It is designed to be lossy at VHF.
Capacitor C6, 0.001 uF, is critical for safety. This capacitor blocks the B+ high voltage from the output. Use at least a 1 kV rating for the capacitor and buy it from a reputable vendor.
The output goes to the LPF and RF monitor.
The LPF went through a couple of iterations to the final pi-L configuration. The RF monitor uses a modification of the W8DIZ tandem match SWR meter, and LM3914 and LED bargraph for a touch of 1980s and 1990s technology. The schematic for the LPF and RF monitor is figure 3.
The function of the low pass filter is two fold, not only does it filter harmonics but also it transforms the plate impedance to the 50 Ohm output. In a solid state final amplifier the voltage is relatively low with high current giving a low impedance output. The LPF for this case needs to be able to step up the final amplifier's impedance to the 50 Ohm output. For a tube final amplifier, there is high voltage and low current giving a high impedance output. The LPF in this case needs to be able to step down the final's impedance to the 50 Ohm output. For tube final amplifier a parallel tuned tank, a pi or a pi-L network is used to step down the impedance and attenuate harmonics. Though many vintage transmitters use a parallel tuned tank or a pi network, the second harmonic attenuation is not adequate to meet modern requirements.
But to design the output network, the LPF started as a parallel tuned tank to provide operation for measurement of basic characteristics so the pi and pi-L networks could be designed. The design of these networks followed the description given in chapter 11 of the 22nd edition of William Orr's (W6SAI) Radio Handbook. This is a classic reference that, even though it is over 40 years old, is still relevant today. Grab a copy of the W6SAI Radio Handbook if you are able to find a copy.
After prototyping the parallel tuned tank, basic characteristics were measured so that the pi-L network could be designed. The pi-L network that is used is L1, L2 and capacitors C2 - C12 in figure 3. The layout is a series of headers so that the capacitor combination, that tunes the network to minimize the plate current and maximize the power output, could be chosen by inserting jumpers on the headers. Using jumpers instead of variable capacitors is an experiment to eliminate variable capacitors that are physically large and eliminate tuning at each set up.
Though the header layout worked in operation of the transmitter, the value of capacitance selected by the jumpers was different from the calculated value. While in prototyping the transmitter calculated and actual values where close in value. The variance was due to the header layout that added a significant amount of capacitance. A better design would be to have pads available for a couple capacitors to be soldered in at each side of L1. Initially variable capacitors would be tacked in to get values needed for a tuned and loaded into a dummy load or antenna of choice. Then remove the variable capacitors and solder in fixed capacitors.
For an in-depth discussion of the design of the three LPF variants, please refer to the Appendix.
A means to monitor the RF output is useful to track the operation of the transmitter and loading of the antenna. A tandem match coupler is used to sample the RF going to the antenna from the LPF. The resulting forward measurement is displayed on a LED bargraph.
The tandem match coupler is T1 and T2 and associated components in figure 3. The tandem match coupler was invented in 1966 by C.G. Sontheimer. The forward RF power is sampled from T1 and is rectified and smoothed by D1 and C12. A voltage, VFWD, proportional to the forward RF power is developed across R3.
VFWD is applied to U2, the LM3914 linear display driver. The LM3914, in dot mode, drives a 10 LED bar display, BAR1. The transmitter, as built, drives the greatest LED at full power. The sensitivity of the RF monitor can be modified by changing R3 and the LM3914 configuration. The LED bar display is mounted perpendicular to the PC board so it may be monitored during operation.
High Voltage Control
The High Voltage supply, described in another post, is mounted on a perfboard. A SPST switch, SW1, is used to enable the high voltage supply. A header jack, J2, allows for measurement of current being delivered by the high voltage supply. During normal operation a shorted header is connected to J2. Battery voltage is delivered to the other boards by J3. The high voltage is delivered via J4 to the RF oscillator and amplifier board.
Final Assembly and Operation
The SOTA transmitter was assembled into a small, lightweight wood box. The RF oscillator and amplifier (figure 5) and the LPF and RF monitor (figure 6) are on two separate printed circuit boards (PCB). The two PCBs and the high voltage perfboard were mounted on separate sides of the wood box. Interconnects are used between the boards in the wood box and to the T/R switch board mounted on the QCX. There is enough room in the wood box for the QCX and T/R switch to be stored during transport.
Figure 5 RF Oscillator and Amplifier Figure 6 LPF and RF Monitor
The Gerber Files are available at the project Github page.
The SOTA transmitter was carried in a backpack with the T/R switch and QCX, and a battery to SOTA summit W6/NC-402. The summit was successfully activated on 40 meters with seven contacts, including two summit to summit contacts. Figure 7 shows the SOTA setup for operation on the summit. The SOTA transmitter had previously been tuned to the trapped EFHW antenna that was used on the summit. No retuning of the transmitter was required during the SOTA operation. Figure 8 is a short video of the SOTA transmitter in action, the audio is the QCX keyer sidetone. The operation of the RF monitor can be seen in the video.
The chasers did report that there was some chirp on the signal but did not keep them from being able to copy the transmission. It reminded some of their early novice days. A discussion of the possible source of the chirp is in the appendix.
Figure 8 Video of Transmitter Transmitting, the Audio is the QCX Sidetone
An important parameter for the design of the LPF is the impedance that the LPF sees from the tube plate circuit. The DC resistance of the plate, Rdc, will be plate voltage, Vp, divided by the plate current, Ip. For a tube operating in class C, the RF plate load resistance, RL, seen by the LPF is approximately one half of the DC resistance.
Rdc = Vp/Ip Equation A1 RL = Rdc/2 Equation A2
Estimating the plate voltage is straightforward during design but estimating plate current is more difficult. To design the LPF, a value of 40 mA for plate current was used. Prototyping the circuit, even with a LPF that is not exact, will provide the opportunity to measure the plate current to better hone the LPF design. Using the 40 mA estimate, the RF plate load resistance is about 2700 Ohms.
The parallel tuned tank is shown in figure A1. In this circuit C1 and L1 form a resonant circuit, their reactances are equal to RF plate load resistance divided by the Q of the circuit. For decent harmonic attenuation the Q should be greater than 10; there is little improvement beyond a Q greater of 15. Since this step is just to measure plate current, a Q of 10 was used.
XL1 = XC1 = RL/Q Equation A3 XL2 = 50 Ohms Equation A4
An advantage of the parallel tuned tank is that it is easy to design, only equations A3 and A4 are needed. For operation at 7.04 MHz, Q = 10, Vp = 215V and Ip = 40 mA; C1 is 84 pF, L1 is 6 uH and L2 is 1.1 uH. The circuit was protyped with this LPF, it provided about 1 Watt into a 50 Ohm dummy load. The plate current was measured to be about 28 mA when C1 was tuned for minimum plate current and maximum power out. This value for Ip will be used to design the pi network.
The pi network LPF is shown in figure A2. Capacitor C1, the tune capacitor, is tuned to minimize plate current and C2, the load capacitor, is tuned to maximize power output. The pi network has better attenuation of harmonics than the parallel tuned tank but requires an additional variable capacitor and more design steps. The pi network was used as the output LPF for many vintage tube rigs. For high power the capacitors needed wide spacing between plates for the high RF working voltages. Though this increased the cost of the network, leading rigs of that period used the pi network.
Designing the pi network LPF is a bit more work. Even though the pi-L network is used in the final design, reviewing the pi network design steps is worthwhile to better understand the pi-L network and appreciate the use of pi networks in vintage rigs. The pi network can be evaluated as two L networks with a common inductor L. Two series inductors, La and Lb, will be substituted for L (La at C1, Lb at C2). La and Lb are part of separate L networks - C1 and La, Lb and C2 as shown in figure A3.
XC1 = XLa = RL/Q Equation A5
XC2 = -Ra*SQRT(RL/(Ra(Q^2 + 1) - RL)) Equation A6
XLb = -Ra^2*XC2/(Ra^2 + XC2^2)
For class C operation and RL greater than 3000 Ohms, a design Q between 12 and 15 is suggested by the Radio Handbook. A Q of 13.5 was used for this design which gave C1 = 81 pF, C2 = 543 pF and L = 6.8uH. A T80-2 toroid was used for L in the prototype build with the pi network. The power output to a 50 Ohm dummy load was about 1.7 Watts, better performance than the parallel tuned tank. The harmonic attenuation is better than the parallel tuned tank, estimated to be between -35 dB and -40 dB for the second harmonic. The pi-L network will have even better harmonic attenuation.
The pi-L network is shown in figure A5, L2 is added to form the pi-L network. Like the pi network, capacitor C1, the tune capacitor, is tuned to minimize plate current and C2, the load capacitor, is tuned to maximize power output.
The pi-L network is evaluated as three L networks as shown in figure A6. An image resistance Ri is used as a load/source for the second and third L networks. A resistance between 300 and 700 Ohms is recommended for Ri. An image Q, Qi, is also used to calculate the final L network parameters.
Qi = SQRT(Ri/Ra) Equation A8
XC1 = XLa = RL/Q Equation A9
XC2a = -Ri^2 * SQRT(RL/(Ri(Q^2 + 1) - RL)) Equation A10
XC2B = -Ri/Qi Equation A11 XLb = -Ri^2 * XC2a/(Ri^2 + XC2a^2) Equation A12
XL2 = Ri/Qi Equation A13 XL1 = XLa + XLb Equation A14
XC2 = XC2a +XC2b Equation A115
For class C operation and RL greater than 3000 Ohms, a design Q between 12 and 15 is suggested by the Radio Handbook. A Q of 13.5 and Ri of 550 Ohms were used for this design. This gave C1 = 82 pF, C2 = 347 pF, L1 = 8.7 uH and L2 = 3.8 uH. The power output to the 50 Ohm dummy load was just under 2 Watts, better performance than the pi network. The second harmonic attenuation should be about -55 dB.
The Excel worksheet for pi and pi-L network calculations is available at the project Github page. The above examples all assume purely resistive loads. To evaluate complex loads a Smith chart may be used. The SimSmith and SimNEC software may also be used to model the networks.
Some of the chasers during the SOTA activation commented that there was some chirp on the signal they received. It was not a serious chirp where copy was not possible. But it was still noticed and let some OMs reminisce about their novice days in the prior century.
A chirp is a slight change in frequency when a dot or dash is sent by the transmitter, sounding like a bird's chirp. Simple, single stage transmitters are prone to chirps according to the 1979 ARRL Radio Handbook. It further states that if the stage is lightly loaded and followed by isolating buffer amplifier stage, the chirps are reduced or eliminated. That is the course taken in most vintage transmitters with a clean CW signal. For this build, adding another tube as a buffer would add substantially to battery load due to the additional filament current. And also increase the size and complexity of the transmitter.
Regulating the high voltage supplied to tube plate and screen grid was done in vintage transmitters to reduce chirps in addition to reducing frequency drift. As noted in the high voltage module evaluation, the high voltage supply is well regulated. Probably better regulation than the VR tubes used in vintage transmitters. Also some vintage transmitters did not key the oscillator but keyed a later stage, another way to reduce chirps.
In the electron coupled circuit used in for the SOTA transmitter, there is isolation between the screen grid and the plate. The screen grid is acting as the oscillator's 'plate' with amplification and buffering to tube's plate, the output of the circuit. Greg Latta, AA8V, has an excellent electron coupled oscillator post. In another post, AA8V describes how a 6CL6 tube was found to have superior isolation between the screen grid and plate according to a March 1950 QST article by W1JEQ.
A different tube, 12AQ5, is used in the SOTA transmitter. The quality of the screen grid to plate isolation is not known for the 12AQ5. In addition, the screen grid bias is derived from the plate supply through resistor R2. Even though the plate voltage is well regulated, the voltage drop across R2 will vary as the screen current changes - such as during keying. Since the screen grid is the oscillator 'plate', voltage variation will induce a frequency pull. And the screen grid voltage is going to be close to the plate voltage since the screen current is relatively low ( < 10 mA). The 12AQ5 data sheet (figure A8) recommends a screen grid (Grid-No. 2) voltage close to the plate voltage.
The oscillator circuits used by AA8V and in the W1JEQ QST study all had a screen grid voltage substantially lower than the plate voltage and separately regulated using cold cathode VR tubes. Two 0B2 108V VR tubes were used in the AA8V circuit, providing a regulated 216 Vdc screen grid voltage; much less than the 350 Vdc supplied to the plate. The W1JEQ test circuits used 250 Vdc plate voltage and a regulated 150 Vdc screen grid voltage. My 1970s home built two tube novice transmitter, with little or no chirp, also used a regulated screen grid potential much less than the plate potential. The 6CL6 data sheet (figure A9) recommends a 100 volt difference between the screen grid (Grid No. 2) and plate potentials. According to W1JEQ, lower screen grid voltage reduces current through the crystal. Reduced crystal current will improve the keying characteristics of the oscillator. Greater difference between the screen grid and plate potentials may contribute to better keying due to reduced crystal current and better screen grid to plate isolation.
The SOTA transmitter chirp is due to the screen biasing scheme and perhaps the relatively close screen grid and plate potentials. The SOTA transmitter could be improved by better regulating the screen grid potential. Rather than using cold cathode VR tubes, high voltage Zener diodes could be used. The screen grid potential needs to be at or less than the plate potential so the the regulating circuit must accommodate this restriction. Even though the data sheet does not recommend it, the screen grid potential could be lowered for the 12AQ5 to ensure that the screen voltage stays below the plate voltage. Voltage regulation derived from B+ will require a dropping resistor, whether using Zener diodes or cold cathode VR tubes. Depending on difference between the screen grid and plate voltages, substantial power dissipation may occur in the dropping resistor. This will add to battery load, potentially impacting the portability of the SOTA transmitter. Portability of the SOTA transmitter is a key (and essential) difference to the other transmitters referenced in this discussion.