I have been updating the QO-100 set-up again.
I picked up a 110cm dish at a nearby radio ham who had used it for QO-100 before upgrading, so he could transmit DATV over the wide band transponder.
This, with the 4-patch feed antenna, makes my SSB signal quite readable. There was, however, a problem with the audio frequency spectrum of the SSB signal, low frequencies were too high, and the higher frequencies too low. An external microphone with a tone control helps, but I suspect that some other solution needs to be found, either a full realignment of the transceiver (which may be needed for the frequency accuracy, anyway
The frequency stability of the up-converter appears to be quite good, though the precision lacks a bit. less than 1kHz offset is not too bad for a non tuneable TCXO. An external reference will remedy this.
I found, however, that making the modification for an external reference requires very good SMD soldering techniques, so I might go to someone with capabilities for this, as I do not think I have a sufficiently fine tip, and the fine motor skills to do it myself. The mod requires moving a "0-Ohm" resistor.
On the receive side I have used a very simple solution for an external 25MHz reference for the LNB. It is placed indoors in a foam insulated enclosure, but the temperature drift is more than I like, about 1kHz per degree Celsius around 25kHz. I can slow the drift with the foam insulation, but the drift is there none the less.
I am actually impressed because the oscillator is a simple 5V DIL-oscillator in a small metal casing, no TCXO or other stabilization (except for the crystal, of course). This has provided very good performance otherwise. The oscillator results in an offset somewhere around 5 - 15kHz, too much for easy re-setting of the receiver frequency after being absent from the receiver.
I have used the same oscillator for controlling a simple LNB (without dish or a further horn antenna) for some beacon monitoring.
The last few days I have tested a low cost Chinese 25MHz TCXO - claimed 0.1ppm stability. The stability was excellent, but the resulted was poor sensitivity (in spite of a sufficient output to the LNB(s)). Further, I observed some spurious responses 200 and 400kHz offset from the wanted frequency. A check with a spectrum analyzer showed, as expected, clear sidebands 200 and 400kHz to each side of the nominal signal, both about 60-70dB down. No wonder I had spurious signals, when that modulation is multiplied by almost 400.
Further, I suspect that the sideband noise of the oscillator (which I cannot measure) is too high, so the LO signal on 9750MHz is spread out, reducing the signal-to-noise ratio. So right now I am back to using the simple DIL oscillator.
With the simple DIL oscillator, what can be done to stabilize that further? I can think of 2 things:
1) Making a simple "oven" to stabilize the temperature, then adjust the supply voltage to generate the correct frequency.
2) Build a TCXO with a 25MHz Xtal
3) Build or find a VCXO for 25MHz and incorporate a PLL circuit to lock it to a 10MHz reference (OCXO, Rubidium standard, or GPSDO
4) Use a GPSDO set to 25MHz
5) Take a 10MHz reference signal, square it, divide-by-2 (5MHz) and extract the 5th harmonic with a bandpass filter and/or lowpass filter.
Because I need a 10MHz reference frequency (for the uplink converter and other microwave stuff) anyway, I should probably try (3) or (5) first, as it is the simplest. If that works to my satisfaction I am all set with respect to frequency stability, and I can thing of other refinements and other projects on frequencies high and low.
