[TowerTalk] Reliability (or not) of SMA connectors

[Lux, Jim]

On 12/5/22 2:06 AM, Jim Brown wrote:
> On 12/4/2022 11:07 PM, Michael Tope wrote:
>> I have seen it used that way in all the space hardware that I've 
>> worked on 
>
> Mike is part of the team that put our rovers on Mars, and Jim Lux is 
> part of the same organization. I know he's done some pretty impressive 
> stuff too. 


Yeah, but I've also had SMA connectors come loose (fortunately not in 
space, that I'm aware of). I've had to re-run a multiday test though.    
Mike's comment about big cables and tiny connectors is really 
important.  You'll see LMR-240 kinds of coax - stiff, heavy.  I don't 
think that's what SMA was designed for - it's more like 0.085 semirigid 
and tenth inch coax like RG-174.

We also spent months tracking down an extra 0.1 dB change in the gain of 
a radar receiver over temperature (the gain changed several dB, but 
should have been a nice straight line, but wasn't, there was a step at 
around 20C - and no it wasn't the PTFE step). We finally figured out 
that it was a SMA that wasn't torqued correctly on the input to a 
Ku-band LNA, and the tiny change in match due to CTE mismatch in the 
semi rigid cables moved it just enough.  It flew, since the system 
calibrated itself - but the concern was "what if it's 0.1 dB today, and 
10 dB in flight".  The connectors were all staked with a blob of 
something, and it was buried so deep the risk of breaking something else 
was higher than the risk of an idiosyncracy turning into something more.

I think anyone who has done precision RF measurements has had similar 
experiences. There's a reason you see people tape the cables down to the 
bench.  That, and keeping the big cable from pushing the tiny hardware 
onto the floor.  Hooking up a NanoVNA to 1/2" LMR-400 with adapters is 
always sort of a tail wagging the dog exercise.