[evlatests] New T-Data Procedure for EVLA Antennas

[Robert Hayward]

Craig,

The engineering groups do prefer to work with Excel spreadsheets since 
we want to plot this stuff up easily (very few engineers in the basement 
use Unix). However, I believe Excel can generate data files in various 
types of formats (i.e., comma delimited, etc.) so it shouldn't be much 
of a problem to export a table's worth of Tcals once the format and 
content have been defined. If not, we can run some custom code that will 
parse it any which way we want.

I haven't read VLBA Memo 51 yet but I'll look it up. As to on-antenna 
derivation of the Tcals, that would be nice. I've heard Ken talk about 
doing this before, but it was way down on his list of things to ponder. 
Obviously with the WIDAR correlator, we could do even the widest band 
receivers (i.e., Ka-Band) in about 7 slices of 2 GHz bandwidth chunks 
and have a Tcal calculated every 1 MHz or so across the band. That's 
something we can't do in the lab with our SOIDA measurement system which 
has a detector bandwidth of about 40 MHz.

One thing that we would want to incorporate in our EVLA Tcal files - 
which I believe you do well in the VLBA - is archiving the old values 
for posterity.

-Bob


Craig Walker wrote:
> Getting all of this data sounds quite valuable and I encourage it.  I do 
> have a couple of comments.
> 
> Neither of the formats for the data, Excel and pdf, is something that I 
> would like to try to crack if I were writing a program that wanted to 
> extract the numbers.  The flat text files we use on the VLBA, in 
> contrast, are easy to read.  You might see what usage develops, but some 
> sort of flat file, or maybe database, if the people who want the data 
> know what to do with such things, might prove useful.
> 
> I'm wondering if we should consider an alternative way to derive the 
> actual numbers used in calibration.  If you look at VLBA Test Memo 51, 
> in which we derive the Tsys, gain, and SEFD over the full bandwidth of 
> the receivers, you will see that the SEFD, which is basically the ratio 
> of the other two, tends to be smoother than either of the other numbers. 
>  For these data, we used the frequency dependent Tcals that the VLBA has 
> always had available.  Of we had not, the variations in Tsys and gain 
> would have been very much larger.  The fact that the SEFD, which does 
> not depend on the Tcal (it divides out) is smooth indicates that the 
> Tsys and gain are both likely to actually be smooth too, but that the 
> Tcal values are not totally accurate - or that portions of the system 
> outside the receiver are introducing some variations.  So an alternative 
> scheme for calibration would be to make measurements such as those in 
> VLBA Test Memo 51, using some strong source (accounting for spectral 
> index), and then derive a set of Tcal's that produce smooth Tsys and 
> SEFD plots.  These Tcals could be made to agree with the lab measured 
> ones at some canonical frequency or, perhaps better, in the average.  Of 
> course, such a scheme will involve more work, but would, I think, 
> produce better calibration.
> 
> Cheers,
> 
> Craig
> 
> 
> 
> Robert Hayward wrote:
> 
>> Hello All,
>>
>> The purpose of this memo is to describe our proposed scheme for 
>> handling the receiver calibration data that will be required for the 
>> EVLA Antennas.
>>
>> In the past when a new receiver is installed on an EVLA Antenna, 
>> someone from the FE Group (usually me) would send out a sanitized list 
>> of Receiver Temperature (Trx), Noise Calibration Temperature (Tcal) 
>> and Solar Calibration Temperature (Scal) values as determined by our 
>> SOIDA measurement system. I refer to this collective set of numbers as 
>> the receiver's "T-data".
>>
>> Remember that the VLA only uses the Tcal & Scal values for the 2 
>> default frequencies. The AC and BD pairs have frequency offsets of 50 
>> to 100 MHz, depending on the receiver band. For the old VLA antennas, 
>> we'd send Operations the 2 Tcal values (one for each polarization). If 
>> the receiver had a Solar noise diode, then the 2 Scal values would be 
>> added.
>>
>> The EVLA & VLA receivers are continuously calibrated radiometers and 
>> these cal numbers are used to determine the System Temperature using 
>> the backend calibration formula:
>>
>>    Tsys = 15 * TP/SP * Tcal
>>
>> where TP is the gated Total Power and SP is the Switched Power as 
>> measured at the T5 module (since we no longer have F4 modules on the 
>> EVLA antennas, the T5 is the only synchronous detector available for 
>> the new system).
>>
>> However, the EVLA will eventually be capable of handling Tcal tables. 
>> The VLBA already does this. For a new receiver on the VLBA, we send 
>> VLBA Operations the Tcal values at every 25 MHz across the band for 
>> the lower frequency receivers, and for every 100 MHz at the higher 
>> frequency bands. The VLBA software applies the appropriate Tcal number 
>> for what ever frequency is being observed. We intend to do something 
>> similar for the EVLA.
>>
>> For the EVLA receivers installed to date, we have been sending tables 
>> of  Trx's, Tcal's and Scal's to VLA Ops. As the Modcomp isn't able to 
>> utilize all of this data, Ken Sowinski would then interpolate the cal 
>> numbers for the default frequencies and enter them into the Modcomp. 
>> So all the rest of the T-data was essentially ignored. However, at 
>> some point we will have to start using calibration tables so that the 
>> Tcals across the wide (up to 10 GHz) bandwidths can be applied 
>> appropriately.
>>
>> Rather than send out an email that includes a table of T-data every 
>> time we install a new receiver, or replace a failed one, I propose to 
>> distribute a spreadsheet that contains the T-data for each of the 8 
>> cryogenic receivers on an EVLA antenna. As an added benefit, this 
>> Excel spreadsheet will contain plots of each receiver's Trx, Tcal, 
>> Scal and Gain curves. Additionally, it will contain a page that 
>> automatically calculates the Tcal & Scal for the standard default 
>> frequencies (which should make Ken happy). The worksheet will be 
>> provided in both the native Excel XLS format as well as a PDF file. I 
>> have attached the file for Antenna 14 as an example.
>>
>> I'm hoping that this will allow Ops to easily find the T-data for 
>> updating the calibration parameters but also provides the in-house 
>> astronomers with a quick way to look at such things as:
>>
>> - The Trx across the band. I often get queries from them asking how 
>> well a receiver works at the edge of some receiver band. They would 
>> merely need to look at the Trx and Gain plots to find out.
>>
>> - Ditto for Tcal numbers. As we provide wider bandwidths on the EVLA 
>> receivers, the 2 default frequency values are somewhat meaningless if 
>> an astronomer is working at a band edge (like what Rick seems to be 
>> doing a lot these days). A look at the table of Tcals would allow a 
>> better value of Tcal to be quickly found and applied.
>>
>> - Questions of receiver power levels could be resolved quickly by 
>> looking at the Gain data (i.e., how much less gain is there on a 
>> Q-Band receiver at the high end?). The Gain vs. frequency plot will 
>> allow the astronomer to get a handle on the RF band shape.
>>
>> - Finally, the plots of Trx will allow one to make relative comparison 
>> of the receiver performance on various antennas (i.e., is Antenna 13's 
>> L-Band really worse than the one on Antenna 16?).
>>
>> The spreadsheet is 33 pages long. For each of the 8 receiver bands, 
>> there are 2 pages for data tables, 2 pages for the Trx/Tcal and 
>> Gain/Scal plots, and 1 page for the Default cal values.
>>
>> For the time being they will be located on filehost. PC users will 
>> find them at
>>
>> \\Filehost\evla\techdocs\fe\T-Data
>>
>> while Unix users can access them at
>>
>> /home/evla/techdocs/fe/T-Data
>>
>> This location may change when VLA Ops has time to digest the new scheme.
>>
>> As noted earlier, there will be 2 versions. The first is the actual 
>> Excel spreadsheet which the FE Group will use to manage the SOIDA data 
>> and automatically generate the various graphs. The second is the PDF 
>> version of the Excel worksheets which just about everyone should be 
>> able to download and read. I think we should use a well defined file 
>> name convention like
>>
>>    T-Data-EA14-2000-10-03-RHH.xls
>>       and
>>    T-Data-EA14-2000-10-03-RHH.pdf
>>
>> which identifies the EVLA Antenna (EA14), the revision date (Oct 10, 
>> 2006) and the initials of the person who last modified it (in this 
>> case, me).
>>
>> Note that it will be a generic spreadsheet and will include sections 
>> for the receivers that are yet to be built (i.e., S, U & Ka-Band). 
>> Remember also that the wideband receivers are not yet in production, 
>> so the current L, C & X-Band systems are interim narrowband systems 
>> and their Trx curves look pretty pathetic when plotted against the 
>> full EVLA bandwidths.
>>
>> Henceforth each time the FE Group installs a new receiver, or swaps 
>> one out, we'll send out a notice out of the particular change (e.g., a 
>> new Q-Band on Antenna 23). The notice will include the "Default" page 
>> (in pdf) as an attachment (i.e., 
>> "T-Date-EA14-Default-2000-10-03-RHH.pdf"). Ken (or whoever is the 
>> designated guardian of the Modcomp Tcals) can use it to find the new 
>> values to plug into the Modcomp. When WIDAR is up and running, someone 
>> in VLA Ops will have to extract the table values (in whatever 
>> delimited format is required) and enter them into the WIDAR TCal 
>> database.
>>
>> Things will get way more complicated for Solar Cals once the Lilie 
>> Coupler-Fed scheme is implemented. This is because a programmable 
>> attenuator will be used used to dynamically control the noise diode 
>> power. Thus the WIDAR Scal database will require a noise level value 
>> versus attenuation setting (or, alternatively, a max noise level and a 
>> table of attenuation factors). Not much thought has gone into this yet.
>>
>> I have placed copies of the new T-data files (i.e., Excel spreadsheet 
>> and a pdf versions) for Antennas 13, 14, 16, 18, 24 and 26 in the 
>> above noted directory. They reflect the present state of all of our 
>> installed L, C, X, K & Q-Band receivers.
>>
>> As a final note, I believe that as some antennas come out of the Barn, 
>> the Modcomp may retain the Tcals from their earlier life as a VLA 
>> antenna so weird Tsys's may be reported when these erroneous Tcal 
>> values are applied to the newly installed receivers. It is probably 
>> worth having Ken (or his designate) check all the the Tcals shown on 
>> the "Default" sheet for each receiver on each antenna to make sure we 
>> have a complete set of proper Tcal values.
>>
>> Please take an opportunity to look over the new T-data files. Comments 
>> and suggestions are welcome to we try improve the EVLA calibration 
>> experience.
>>
>> Cheers,
>> Bob
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> evlatests mailing list
>> evlatests at listmgr.cv.nrao.edu
>> http://listmgr.cv.nrao.edu/mailman/listinfo/evlatests
> 
> 
>