6JH8 ONE TUBE DIRECT CONVERSION RECEIVER

This strange circuit is the result of my fascination with beam deflection tubes, also known as sheet beam tubes. They were designed to be used as balanced demodulators, and are noted for having an exceptionally linear transfer characteristic between the deflectors and the plates.

While not wanting to get into too many details of this tube's operation, the most important thing to note is that any signal applied to grid 1 will be amplified and appear as an in-phase (common mode) signal at the two plates. On the other hand, any signal applied between the deflector plates will be amplified and appear as an out of phase (differential mode) signal at the two plates. This allows the different signals to be separated and redirected as necessary. This unique characteristic allows reflexing the audio back through the tube without it breaking into oscillation.

To work as a direct conversion receiver, the RF signal passes through a tuned RF stage consisting of T1 and one section of the tuning capacitor, and is applied to grid 1. A signal from the local oscillator running at the same RF frequency is applied to the deflectors. The result of this mixing operation is an out of phase audio signal at the plates. The out of phase audio is detected in T3 and from the secondary is routed back to grid 1. This audio signal is amplified and appears at the plates as an in-phase signal which passes through the primary of T3, this time with no effect, and then to T4 where it is impedance matched to the headphones.

Feedback for the local oscillator is taken as a differential signal between both plates through a DC blocking capacitor to the primary winding of the oscillator coil T2. Half of the secondary winding forms a parallel tank circuit with one gang of the tuning capacitor, and feeds back to one deflector. The other half of the secondary winding provides an equal but out of phase signal for the other deflector. For best operation the deflectors need a slight negative bias. With a B+ of about 160 volts, the optimum deflector bias is about -9 volts, and is provided by a standard 9V battery permanently wired into the circuit. As there is no current drawn, the battery will last as long as its shelf life.

Because the oscillator circuit operates on differential signals, it seems to be extremely immune to influence from the incoming RF signal which is common mode. The result is that the oscillator hardly pulls at all, even when tuned near a very powerful local station. This in fact, is almost a detriment, because the tuning has to be precisely on frequency before the oscillator locks in. To help, the tuning capacitor has a 6:1 reduction drive, and an additional 5 pF variable capacitor in parallel with the oscillator section of the tuning capacitor, is mounted on the front panel for fine tuning. On the other hand, the stability of the oscillator, allows accurate calibration of the dial and provides immunity from adjacent stations.

For RF regeneration, feedback for the tickler winding of T1 is taken from the cathode circuit. By taking the signal from the cathode, only common mode RF should be present, and none of the oscillator signal (which should be entirely differential).

All of this sounds good in theory, but how does it work in practice? I have to admit, I wasn't expecting this to be a very well behaved circuit. I had expected that a certain amount of unbalance in the differential circuits would lead to some spurious common mode audio cropping up where it wasn't wanted and would lead to oscillation. However, I also figured that this could be cancelled out by proper phasing of the secondary of T3 so that it would be fed back negatively and disappear. It turned out that the circuit was surprisingly well behaved, and worked pretty much according to plan. Maybe better. There was a difference in operation depending on the phasing of the T3 secondary, but it was hard to tell which way was actually better. By the time I had desoldered the transformer leads, flipped them and resoldered, then powered up the receiver, listening conditions would have changed enough that a proper comparison could not be done.

Finally, I wired up a DPDT switch to reverse the polarity on the fly. The result was that in one position (which I will refer to as negative mode) the audio signal level was at a reasonable level and frequency response was quite flat. With the switch in the other position (positive mode), the audio level was a bit greater, and there was some discernible positive audio feedback which did not necessarily lead to oscillation. It actually provided a bit of extra audio gain and a bandpass filter effect which sometimes helped intelligibility of the received signal. With the volume control turned all the way up, it could go into oscillation. However, the transition was very smooth as the audio bandpass became very narrow and began to ring a bit and then finally oscillate. This behavior was affected by the amount of RF regeneration. With the regen set at a modest level, the operation was as described. At the high end of the band, it was possible to turn up the regen quite a bit to get more RF gain, at the expense of audio stability. As a result, the audio gain had to be reduced somewhat when the RF regen was cranked up (especially with the audio phase switch in positive mode). (It is important to make sure that the regeneration is strictly working to provide amplification, and not detection.) In the end, I decided that the DPDT phasing switch should remain in the circuit permanently in order to pick the best operating mode according to signal conditions.

Regarding the RF regeneration, only a single turn was required for the tickler winding, but in retrospect this probably makes sense considering the degree of coupling that comes from using a toroid core. In operation, there was a very abrupt increase in gain as the resistance of the regen pot was increased just a bit above zero. With further rotation of the regen control, the gain remained about the same and then decreased just a bit before finally breaking into oscillation. The optimum position remained the same for frequencies below about 1400 kHz. And so I normally left it in this position and didn't touch it again. Above 1400 kHz, if I was trying to pick up and especially weak signal, I found that the regen could be turned up quite a bit higher to get more gain without breaking into oscillation.

In operation, I found this receiver to be comparable in performance to my previous homebrew receiver which is a two tube superhet with reflexed audio. The new set was a bit more complicated to operate however. It was also somewhat more susceptible to overloading from powerful local stations. Additionally, my antenna matching unit which has a significant effect on the amount of received signal, tended to be more critical with this receiver. As the antenna matcher was adjusted near to the optimum position, it could cause the receiver to break into oscillation. I suspect this may be due to the matcher (being just slightly off frequency) causing uneven attenuation of upper and lower sidebands which resulted in unbalanced signals passing through the receiver, and messing up the common/differential mode signal behavior. This was somewhat annoying, but not an insurmountable problem. With a bit of practice, I got the hang of tuning the antenna, and found the receiver quite easy to operate. I just had to make sure to keep the audio gain turned down until the everything was tuned in.

Being a direct conversion receiver, there was never any question about being on frequency. There was always a clear beat note as each station was tuned in. Even in situations where no station could be heard, there was almost always enough of a carrier from some distant station, to get an audible beat note, and hence there were markers every 10kHz virtually across the entire band. (I was able to use a second receiver with digital tuning to check the frequency when necessary. The second receiver would easily pick up the local oscillater frequency.)

For headphones, Sony MDR-W08 Sport Walkmans were used. Initially, a pair of homebrew piezo phones were tried, but these required the use of an extra matching transformer which added enough insertion loss to negate any benefit of their extra sensitivity. I could have wired the output transformer as an autotransformer to get a better match, but I wasn't crazy about having the B+ that close to my ears. I've come to the conclusion, after using the MDR-W08s in a couple of DX contests, that they are surprisingly sensitive, not at the level of sound powered phones, but seemingly better than standard high-Z magnetic phones. To use the piezo phones in a tube circuit like this one, I've found that the output transformer needs to have about a 2:1 ratio. A probable future modification will be to find a good 2:1 output transformer, and go back to the piezo phones.

The antenna used in the contest was a relatively short long wire, about 35 feet strung out in a horizontal L (with each leg about equal length), running above my deck on the shore of Lake Katepwa. Being almost right on the water, seems to help bring in those DX signals. There was also very little local interference, so I was able to make out quite a few weak signals that would have been otherwise obliterated.

The antenna matcher is an L network consisting of a tapped coil wound on a ferrite core, and a 100pF variable capacitor. An SPST switch allows connection of the variable capacitor to either the antenna or the receiver side of the coil. The matcher makes a significant difference in the amount of received signal, especially when the antenna is as short as this one was.