Since I’ve purchased Mini VNA Tiny, my first ever VNA analyzer, I have completely reconsidered all my beliefs around how cables, connectors and how impedance mismatches in general behave around the circuits. The situation, I noticed, is not as ‘innocent’ as it seems. Most recently, amongst many other cables, I have been measuring the loss of a very high quality Hyperflex-10 cable by Messi-Paolini. This is currently the lowest loss cable on the market on the 10.3 diameter group. The cable loss, measured with Mini VNA Tiny, didn’t come close to the manufacturer’s specifications, although this is a top of the range cable. An absolute turn down for me, as I’ve purchased 75m of that expensive cable to split it into 25m sets for 3 different antenna setups! What I’ve discovered after many hours of measurements left my mind in real shock: Performance of connectors and adapters joining the coaxial cable with the VNA were the cause of additional loss to a big extend!
So, below I am showing to you what I’ve discovered. The adapters were introducing a significant impedance mismatch, masking the cable’s high performance. I’ve isolated the adapters used in my measurements and I investigated them closer up to 500MHz. The adapters are as follows:
- SMA-M to UHF-F (unknown manufacturer)
- UHF-M to UHF-M (Amphenol, type: http://www.amphenolrf.com/083-877.html)
- UHF-F to SMA-F (unknown manufacturer)
All linked in series in the order listed above to create a 70mm long ‘adapter-block’. See, the adapters below:
The impedance performance of the adapters were compared against a standard 50 Ohm WiMo load kit using Mini VNA Tiny, as shown on the picture above. First, the S11 performance was investigated. When only the 50 Ohm standard load was connected directly to the DUT port, the S11 response (dB) and series resistance (Rs) were as shown below:
As can be seen from the above plot, the return loss is excellent, less than -40dB, all across up to 500MHz, with the resistance flat to 50 Ohm value, an indication of a proper impedance match between the DUT port and the 50 Ohm standard load. Now, let’s go a step further and examine how the adapter-block is influencing the above ‘ideal’ plot.
This was done by inserting the adapter-block between the VNA port (DUT) and the 50 Ohm standard load as shown below:
The measurement of the S11 response (reflection loss in ‘dB’) together with the series resistance seen by the DUT port towards the adapter-block, is shown below:
It is really mind dazzling as to what has happened to our ‘perfect’ 50 Ohm flat line! The port of our VNA is now seeing a completely different resistance towards the adapter-block across the spectrum! Anything above 50MHz deviates seriously from the 50 Ohm line, with the return loss reaching a disastrous -10dB value in the UHF range while the resistance is reaching a value as low as 27 Ohms!! This is obviously a sign of serious impedance mismatch caused by our adapter-block! In other words, these ‘little monsters’, although innocent on first place, are the cause of some impedance mismatch trouble across the RF chain. No wonder then, that my high quality Hyperflex-10 cable couldn’t perform as expected – the adapters were introducing an impedance ‘kick’ that was masking the coaxial cable’s behaviour.
Going a step further, let’s investigate the loss introduced (insertion loss, i.e. S21) by this adapter block. This measurement is shown below:
The insertion loss goes as high as 0.8dB, which might not seem much indeed, but worse is that this varies on a ripple-fashion, as shown above, a sign of clear impedance mismatch caused by the 3 adapters. When a coaxial cable is measured with such adapters connected on the VNA ports the ripples are superimposed with the coaxial cable’s response creating a completely difference plot than what’s expected – and this can lead to misleading conclusions about the cable’s performance.
So, no matter how perfect load you have at the end of the line or how good the coaxial cable is, any poor quality adapters used in the interconnections during the measurements are a source impedance mismatch along the system, which inevitably will generate the variations seen above. Next time you use such adapters, be very suspicious – they can’t be ignored! Use the highest quality adapters you can afford if you want to get accurate results in your measurements.
As a concluding remark, it has to be noted that the UHF adapters aren’t really performing well for anything above 30 MHz, see references [1,2] for details. This applies also for the UHF connectors as well, which although popular amongst the ham radio community, are degrading anything across them, if not to a big extend in the VHF band, at least on the UHF band. So, note down this information since this is something that can’t be ignored when you aim to cut down every single bit of unnecessary loss in your ham shack.
Note: Someone might attempt to estimate the loss that each of the 3 adapters is causing on the overall figure. Unfortunately, this is not straightforward process, only a guess can be made. The specific Amphenol UHF-M to UHF-M adapter used in the measurements above is stated to work up to 300MHz, yet the manufacturer does not state the insertion loss whatsoever, it only classifies this connector as a ‘non-constant’ impedance type adapter. I suspect though, that the major loss is caused by the ‘unknown’ adapters SMA-UHF type. So, if we assume that the Amphenol is introducing a loss of a typical ~0.1 dB for this class of adapters, then the other two connectors share a worse case scenario loss of ~0.7dB together, or about ~0.35dB each assuming an equal share (which of course is not true, but for the shake of simplicity lets take it). But, apart from this loss, the fact that the resistance is deviating significantly from the 50 Ohm value, means that these adapters aren’t any good, at least on the UHF band.
PDF Files of measurements:
 ‘UHF’ Connector Test Results: http://www.hamradio.me/connectors/uhf-connector-test-results.html
 The UHF type connector under network analysis: http://www.qsl.net/vk3jeg/pl259tst.html