[mmaimcal] B7 Change Requests

Mon Apr 23 23:24:35 EDT 2007

Hi Al,

I read Lazareff's document tonight, and have a new perspective on the 
issue.  Albeit somewhat contrived, his comparison of assembly #09 (which 
does not meet the sideband rejection spec) with a hypothetical assembly 
that just barely meets it, illustrates the concern:  By requiring the 10dB 
rejection, many of the most sensitive SIS mixers will go unused and the 
array will be less sensitive as a result.  It is certainly not easy to 
weigh these issues against one another.  Ideally, one would have two sets 
of Band 7 receivers: one for best sensitivity, and one for best image 
rejection.  While this is not practical for ALMA as a whole, can we not 
simply require that the total power antennas be outfitted with the very 
best Band 7 receivers? (i.e those that have the best image rejection and 
still have good Trx performance).  Additionally, it sounds like there 
might be another solution that would involve adjusting the bias point 
(i.e. tuning) depending on the observer's needs. See the final paragraph 
below.

Comments
========

I believe their suggested calibration scheme to promote single-dish 
sideband separation (apparently devised by R. Lucas) has more merit than I 
originally thought earlier today at our meeting. They are *not* suggesting 
that one subtract an interferometer spectrum from a single dish spectrum, 
or vice-versa. In fact, the method described may be worth attempting to 
further improve the separation, even if all the receivers gave >10dB. So 
we may want to reword our response accordingly.  However, there are some 
complications that should be pointed out at Tuesday's telecon, and in our 
response:

First of all, they do recognize the need to measure a sideband ratio at 
each spectral channel in the IF band--this is done by observing a 
continuum calibrator in interferometric mode and comparing the calibrated 
amplitude (presumably in brightness temperature units?) in corresponding 
channels of the two sidebands. I believe the weaknesses of this strategy 
are: 1) you would need to know the continuum flux density of the 
calibrator accurately in both sidebands (preferably from prior 
observations and monitoring, a spectral index may not be reliable enough), 
2) there is a possibility that more than just the receiver sideband ratio 
may contribute to the observed amplitude ratio (e.g. different paths in 
the correlator during and after Walsh demodulation/sideband separation).

The second problem comes from the phrase "Acquire the image band during 
the single-dish observation" -- I presume that they mean to retune the LO 
to place (what was) the image band into the signal band.  However, one 
would then have yet a different image band contributing to the measurement 
of the image band. If you have only 8dB rejection, you could have 
contribution from the second image at the -16dB level into the original 
signal band after the subtraction.  So, a 40K line in the second image 
could still end up as 1K in the band of interest.  So this technique may 
not work for certain sources/tunings.

However, at the end of the document, a very promising statement is made 
(in the section: "Meeting the rejection specification"):  "it can in 
principle be achieved by degrading the conversion efficiency of the mixer 
having the higher gain (we normally bias both junction near maximum 
efficiency). We have actually demonstrated that this is possible." (see 
Fig of assembly #16 at 311 GHz).  I may be naive, but it sounds to me 
like: by adjusting the bias voltage on the junction, one could give up 
some sensitivity in exchange for more sideband rejection.  If this can be 
done reliably at a variety of frequencies in the band (not just one), then 
it could be an elegant solution that fits both needs.  It would of course 
take extra time to generate two sets of tuning curves for each mixer but 
that could be done during telescope test time if they don't do it in the 
lab.  But before we put any stock in this, I think we should ask that this 
effect be demonstrated for more than one mixer assembly, and at a list of 
LO tunings.  Perhaps you could ask that this be done before accepting the 
change request?

Todd

On Mon, 23 Apr 2007, Al Wootten wrote:

> Folks,
>
> Last week we discussed pending change requests for B7m which were not 
> available on edm.  I put copies on the agenda.  They are now available 
> in edm, and a comment by Lazareff has been made to No. 171.  Tomorrow 
> morning at about 1530 UT there will be a general discussion between IPT 
> leads on these changes.  Please send me your comments as I will need to 
> formulate a science ipt response.  I think the response to 170 will 
> parallel that already submitted for a similar CRE to B9 
> (http://tinyurl.com/2nck3o) except that we have tighter requirements for 
> calibration for B7 and the atmosphere is more generally benign at those 
> frequencies.
>
> 171.  	Band 7 cartridges: Image band suppression
>
> CRE Number : FEND-40.02.07.00-092-A-CRE
>
>
> Description :
> Specification FEND-40.02.07.00-00190-00/T can be met marginally, which significantly impacts the production yield and therefore cost. We argue in attached documents that this specification
> a) Might be significantly relaxed without any impact on the scientific performance of the instrument;
> b) Is formulated in an unnecessarily restrictive way that either impacts the production yield or calls for frequent RFW's;
> c) Impacts the production yield and cost of Band 7 cartridges.
> As an indication, we propose a base level of \u20138dB image gain, with \u20137dB allowed in 10% of the frequency span of any IF band.
>
> See: http://tinyurl.com/24hr9h
>
> ------------------------------------------------------------------------------
>
> 170.  	Band 7 cartridges: IF power variation
> CRE Number :FEND-40.02.07.00-093-A-CRE
>
>
> Description :
> The pre-production cartridges generally do not meet specification FEND-40.02.07.00-00210-00/T. The first clause of that specification, i.e. ±2dB in any 2GHz portion of the IF band, is the most problematic.
> With the understanding that this specification has a direct impact on the efficiency of the digital correlator, we have designed, tested, and proposed IF bandpass equalizers that would allow us to meet the specification in its present version. We also also heard of active equalizers developed in the back end IPT.
> Because neither of these two developments has an official status as far as we know, and because we cannot commit to a specification that cannot be met (see attached document), we ask for the following relaxation of the specification:
> \u2022 ±3dB (6dB peak to peak) in any 2GHz portion of the IF band
> \u2022 ±3.5dB (7dB peak to peak) across the whole IF band
>
> See: http://tinyurl.com/2xtfdh
>
> Clear skies,
> Al
> Al Wootten
> Interim Project Scientist
> ALMA, 40 El Golf, Piso 18
> Las Condes, Santiago, Chile
> Tel: +1 434 296 0329
> Al Wootten Cell +14342424713
>
> _______________________________________________
> mmaimcal mailing list
> mmaimcal at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/mmaimcal
>

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Todd R. Hunter  | NRAO Assoc. Scientist / GBT PTCS Instrument Scientist
GB: Office 111 (304)456-2157  PO Box 2, Rt. 28/92, Green Bank, WV 22944
CV: Office 306 (434)244-6836  520 Edgemont Rd, Charlottesville VA 22903