About a year ago I purchased the NooElec Ham It Up v1.3 bare bones HF Up-converter (HFU) to use with their NeSDR SMArt dongle. The plan at that time was to get an SDR-based QRSS Grabber running online 24/7. For various reasons this never happened, and it’s only now that I have got the various bits and pieces all connected and working. The HFU is needed to extend the NeSDR frequency range down to the lower ham bands, for example the 80m, 40m and 30m bands.
My initial results were encouraging. With the various bits (NeSDR + HFU + 8 port USB hub) simply laying on the bench, I was getting good results, but the SDR was slowly drifting in frequency. The receive frequency moved about 60Hz overnight. Now for normal use, where you’re tuning across the band, listening to different QSO’s etc. this setup is fine and works really well, but if you’re monitoring QRSS signals 24/7 and the receiver drifts 60Hz that’s more than 25% of the whole QRSS band segment!
My immediate thought was to place all the bits in a box to protect against any draughts etc. that might be causing this problem. The results were not good… it made things worse! Checking against my 10MHz GPSDO I could see the line on Spectrum Lab tumbling at around 3Hz per minute! So was this due to the NeSDR dongle, the HFU or both? The NeSDR does run warm, so by putting all the bits in the confined space everything was warming up. I knew the specification of the NeSDR TCXO was 0.5ppm so my suspicions were with the HFU. To be sure I checked the NeSDR on its own and with the HFU switched in-line. As I expected the NeSDR on its own was basically “rock solid”, the trace on Spectrum Lab was straight, horizontal and didn’t appear to move. With the HFU switched in-line the trace was falling… so the 125MHz TCXO fitted to the HFU was suspect… or was it?
When I purchased the HFU I assumed that the board came with a TCXO fitted. On closer inspection and after revisiting the manufacturers specifications it soon became apparent that was a false assumption on my part… but not all was lost! Re-visiting the NooElec website I discovered that they offer a 125MHz 0.5PPM TCXO upgrade module. You can now also buy the “Plus” version that has the TCXO fitted as standard. Needless to say an order was placed for the TCXO module. 🙂
I tried attaching a large (6mm) nut to the can of the existing 125MHz crystal so that this would act as a “thermal mass” and potentially reduce drift due to minor changes in temperature. A thick piece of foam insulation was also added to the bottom of the board to try to stabilise the temperature surrounding the quartz crystal. Initial results showed that this was all helping but it was still not good enough for long-term monitoring of QRSS transmissions.
One week after ordering the TCXO an envelope dropped through the letter box. The installation of the TCXO was fairly straightforward; the original surface-mount crystal package is probably best removed with a hot-air tool and tweezers. I used a normal soldering iron with a home-made saddle-shaped bit that enabled me to heat all four connections at the same time. The only other thing that needed a few moments of careful thought was the orientation of the TCXO board.
Once the new TCXO was fitted, the next stage was to fit the foam insulation back in place and to run some tests to see how well the new setup worked. I used the 10MHz output from my GPSDO as the signal source and Spectrum Lab to view the frequency variation. The following photo shows the trace over a 10 minute period. I adjusted Spectrum Lab to have the trace running fairly close to the top of the screen to help show that the trace runs parallel to the border and is not drifting up or down in frequency.
My tests so far would suggest that what was already a very good SDR combination, one that would probably suit most people’s needs, is now even better with the frequency stability needed to run as a QRSS Grabber or work with other digital-modes.
- To run a QRSS Grabber 24/7.
- Having moved the QRSS Grabber receive hardware away from relying on an expensive multi-band transceiver to a relatively low cost SDR platform move the PC requirements to a Raspberry Pi 3.