[evlatests] S5/S6 investigation

[Ken Sowinski]

On Fri, 11 Jul 2025, Paul Demorest wrote:
> hi all,
>
> Earlier this week, Cameron pointed out several antennas that are 
> reporting unexpected values for S5 and/or S6.  Ken and I investigated 
> some this week, and Hichem sent some useful info.  Here's a report for 
> the record, the short story is that a few of these switches do appear 
> stuck/broken, but the state they are stuck in is correct for normal 
> operations and there is no need to make any changes or repairs.  More 
> details for those interested:
>
> These "transfer" switches have the effect of swapping the RCP and LCP 
> signals before the T304 inputs.  S5 swaps the A and C IFs, S6 swaps B 
> and D.  Each switch can have two states, which I will call "normal" and 
> "swapped".  In regular operation, all are in the normal state; the 
> executor is hard-coded to command them to normal at the start of each 
> scan.  The only way to temporarily set them to swapped is via a manual 
> MIB command from device browser or telnet.  I suppose these switches 
> were intended to help troubleshooting(?), but this functionality is 
> basically never used AFAIK.

Some historical context.

Yes, it was mainly for diagnostic pusrposes and was frequently used
for that in the pre-EVLA days to isolate problems to before or after
the xfr switch.  As Paul writes it has rarely been used since then,
mostly because we never think of it, and have no easy access to it.

It aso had an astronomical justification.  Milller ocassionally did Zeeman
experiments using the xfr switch to alternately route RCP and LCP into
the same IF and sampler to preclude bandpass differences between the
two IFs that would otherwise require difficult bandpass calibrations.
I imagine it must have worked because he did this many times.  Emmanual
(cc'ed here) may remember/know more than I about how effective it was.
If the main purpose was to deal with the differing  analog filter 
bandpasses, that is now moot because of the use of digital filtering.

Ken


> The two switches are combined into a single mon/ctrl point, 
> M301_SW.SW_S5_S6_MON/CMD.  There are only four allowed values:
>
> 0x0 = both normal
> 0x4 = S5 swapped, S6 normal
> 0x8 = S5 normal, S6 swapped
> 0xC = both swapped
>
> Given that the switches are never exercised, we expect all antennas to 
> read 0x0 at all times.  Mostly this is true, except we see ea04=0x8, 
> ea14=0xC, ea17=0xC and ea27=0x4.  Ken and I both tried sending various 
> commands but it seems that for these antennas one or both switches are 
> stuck in the swapped state:  S6 for ea04, S5 for ea27 and both for 
> ea14/ea17.  For ea04 and ea27 the non-stuck switch does respond to 
> commands as expected.  For one normal antenna (ea01), both switches 
> responded to commands as expected.
>
> Regarding the effect on data, it is easy to see if RCP and LCP are truly 
> swapped on an antenna:  On a typical low-polarization source, 
> parallel-hand correlation amplitude (vs the other normal antennas) will 
> be low while cross-hand correlation is high.  This is NOT seen during 
> regular observing so that either means: these stuck switches are 
> reporting an incorrect state; OR maybe someone has worked around the 
> problem by also swapping the input or output cables (thus cancelling out 
> the swap in the switch).  For practical purposes it doesn't really 
> matter too much which is the real explanation.  The current setup is 
> working fine and there is no need to mess with it.
>
> I tested exercising the switches on ea01 and ea04 while observing a 
> bright source.  Both ea01 switches acted as expected:  The signal moved 
> from parallel-hand to cross-hand correlation when the relevant switch 
> was commanded to swapped position.  And the switches all revert to 
> normal position at the start of a new scan.  For ea04 I was able to swap 
> A/C via S5, but not able to get S6 to change its state.  So the 
> interpretation from looking at the data was consistent with the M&C 
> readings.
>
> I did not check any antennas other than the ones mentioned here so it 
> might be possible there are others that are stuck in the "normal" state.
>
> One final thought is that since we don't really need S5/S6, if these 
> could be repurposed to replace any of the other failing band/LO switches 
> that would be fine in my opinion.  I'd guess this is unlikely since it's 
> a different style switch though.
>
> Cheers,
> Paul