[evlatests] Noisy C-band antennas?

Rick Perley rperley at nrao.edu
Wed Aug 30 11:12:19 EDT 2023 

I was also away when Pedro showed this material.

Something has definitely changed.  I have on my bulletin board a 
spectrum generated by Bob Hayward and me, back when we were doing the 
initial commissioning tests of the new C-band receivers.  It is labeled 
as from ea24, and I have it annotated with 'Mangus to Gray Hill'.  But 
... the frequency and width don't match that listed in Pedro's table.  
The RFI 'wedge' is centered at 6004 MHz, with a width of 30 MHz.

Also, in the copious amount of data that I've perused over the years 
since we implemented the wideband system, I have never seen any evidence 
of compression.  The RFI 'wedges' were always present in the region 
around 6.1 GHz, but they were never strong enough to cause saturation 
affecting spectral windows outside those where the RFI was present.

Rick

On 8/29/23 17:10, Pedro P.B. Beaklini wrote:
> Hi Guys,
>
> I am sending attached the source of the problem. A map of the 
> microwave link that crosses VLA. I am also attaching my presentation 
> at the last VLA/VLBA test/commissioning meeting, where I briefly 
> updated those issues. The problem can be solved using a mix of 3-bit 
> and 8-bit samplers instead of the usual 3-bit samplers.
>
> As Lorant mentioned in today's meeting, on the next OPT release, we 
> will provide the users with an alternative resource to be used at 
> C-band during A and B configuration. We will also add an explanation 
> about what is going on to the user's guide. I am also sending a draft 
> of the text that will be added attached; it is still being revised by 
> the team to reach its final version.
>
> I hope that solves your major concerns. But of course, we can still 
> work on new tests if a deeper explanation is needed.
>
> Cheers
>
> Pedro
>
> On 8/29/23 16:26, Rick Perley via evlatests wrote:
>> This problem must be highly time-variable.  I have eight wide-band, 
>> 3-bit, C-band observations on hand, taken between 22 June and last 
>> week -- and *none* of them show any problem in the calibration gains, 
>> or any hint of any problem.
>>
>> We really need to find the source of this problem...
>>
>> Rick
>>
>> On 8/29/23 16:23, Bryan Butler wrote:
>>>
>>> Middle of E and W arms.  See attached.  Drew tells me they see it 
>>> "quite a lot" - not sure if it's always the same antennas.
>>>
>>>     -Bryan
>>>
>>>
>>> Rick Perley via evlatests wrote on 8/29/23 16:14:
>>>> Where are these eight antennas located? In previous tests, there 
>>>> seems to be little spatial correlation for the affected antennas -- 
>>>> in Pedro's note (below), W36 and N28 are far, far apart.  Hard to 
>>>> see how a single source can affect both so badly.
>>>>
>>>> Also note that I claimed on June 29 that the effect was completely 
>>>> missing.
>>>>
>>>> So -- do we really understand what is going on?  Perhaps a more 
>>>> careful review of data taken at C-band may be needed.
>>>>
>>>> Rick
>>>>
>>>> On 8/29/23 15:52, Bryan Butler via evlatests wrote:
>>>>>
>>>>> Yesterday, I got a pipeline reduction report from the DAs on a 
>>>>> science observation at C-band, observed on July 27.
>>>>>
>>>>> The situation seems really grim - upon digging further (well, 
>>>>> asking for more information from James), this effect is seen 
>>>>> strongly on 8 antennas (1, 4, 5, 9, 19, 25, 26, 27).  Roughly 
>>>>> 1/3rd of them.
>>>>>
>>>>> Shouldn't we be telling our C-band observers to observe in 8-bit, 
>>>>> instead of 3-bit, to avoid (or at least lessen the effect of) 
>>>>> this? If it really is compression, it's unclear (to me at least) 
>>>>> that this is recoverable at all, and certainly not without a much 
>>>>> larger uncertainty on the flux density scale.  If I had known it 
>>>>> was this bad, I certainly would have observed in 8-bit, and taken 
>>>>> the sqrt(2) loss rather than the 1/3 (potential) loss.
>>>>>
>>>>>     -Bryan
>>>>>
>>>>>
>>>>> Pedro P.B. Beaklini via evlatests wrote on 7/11/23 16:29:
>>>>>> Rick, and all,
>>>>>>
>>>>>> I did a test last Thursday on this. The thing is that the noise 
>>>>>> depends on the antenna location. The compression is due to an RFI 
>>>>>> source in the direction of stations W36 and N28 and often can be 
>>>>>> seen in N32, N36, and W32 stations. For some reason that we are 
>>>>>> still investigating, ea17 is more affected than others when it is 
>>>>>> on the RFI path. But the problem can be seen in the other 
>>>>>> antennas in the RFI path.
>>>>>>
>>>>>> On the current configuration, there is no antenna on W36 and N28. 
>>>>>> EA17 is far away now, at N64, and no compression was seen in my 
>>>>>> test on ea17, in agreement with your data. However, on the 
>>>>>> antenna closest to W36, ea05 at W40, I could see a C-band 
>>>>>> compression on 3-bit. Please, can you take a look at ea05 on your 
>>>>>> data? I also saw a new noise source leading to compression on the 
>>>>>> east arm ( peak on EA04 at E48). That is new to me and needs 
>>>>>> further investigation, but the effect on data is similar.
>>>>>>
>>>>>> If someone wants to take a look at pdif varying in time on my 
>>>>>> tests and past data, I put some shorts movies at. Each frame is 
>>>>>> 5s, time on top in minutes
>>>>>>
>>>>>> /lustre/aoc/sciops/pbeaklin/tsys/C-band-issue/2023-06-06/pdif/
>>>>>>
>>>>>> As a last comment, I was not the first one to notice the issue. 
>>>>>> The DAs, SRDP team, and the support telescopes group have been 
>>>>>> aware of that for a while. I only decided to take a deeper look 
>>>>>> at it :-)
>>>>>>
>>>>>> Cheers
>>>>>> Pedro
>>>>>>
>>>>>> On 6/29/23 14:55, Rick Perley via evlatests wrote:
>>>>>>> Some months ago, Pedro noted that certain antennas were 
>>>>>>> especially noisy (low SNR) for some antennas and spectral 
>>>>>>> windows, when using 3-bit samplers. From my own data, I 
>>>>>>> confirmed this.
>>>>>>>
>>>>>>> The recent data referred to in my last circular allows a check 
>>>>>>> on this situation -- the data were taken with 3-bit sampler.
>>>>>>>
>>>>>>> The 'noisy' effect is completely gone!  All antennas, in all 
>>>>>>> spectral windows, have exactly the expected sensitivity and gains.
>>>>>>>
>>>>>>> Pedro plans to take some data, to confirm.
>>>>>>>
>>>>>>> Rick
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> evlatests mailing list
>>>>>>> evlatests at listmgr.nrao.edu
>>>>>>> https://listmgr.nrao.edu/mailman/listinfo/evlatests
>>>>>>
>>>>>> _______________________________________________
>>>>>> evlatests mailing list
>>>>>> evlatests at listmgr.nrao.edu
>>>>>> https://listmgr.nrao.edu/mailman/listinfo/evlatests
>>>>>
>>>>> _______________________________________________
>>>>> evlatests mailing list
>>>>> evlatests at listmgr.nrao.edu
>>>>> https://listmgr.nrao.edu/mailman/listinfo/evlatests
>>>>
>>>> _______________________________________________
>>>> evlatests mailing list
>>>> evlatests at listmgr.nrao.edu
>>>> https://listmgr.nrao.edu/mailman/listinfo/evlatests
>>
>> _______________________________________________
>> evlatests mailing list
>> evlatests at listmgr.nrao.edu
>> https://listmgr.nrao.edu/mailman/listinfo/evlatests

More information about the evlatests mailing list