
First of all, his cables terminate with N connectors. I initially though of changing the short RG174 cable end into a female N, but quickly realised how weak the whole system would be considering the weight of N and in-flexibilty of thick coax. I needed to protect the dongle with a box.
I picked a metallic chocolate box I kept aside some months ago and fitted a flanged N female. While I was at it I also shortened the RG174 cable to what was strictly necessary (1 metre to 10 cm): these two actions resulted in at least a couple of dB gain (or ... "less loss"). Note that 1m of RG174 at 1300 MHz looses 1 dB.
The USB hole is not something I am proud of, but it was realised in less than 5 minutes, so it is acceptable for this experiment. If the on-the-field comparison shows RTLSDR is a decent performer I will consider soldering directly into the PCB both the USB cable and the N or BNC female head.