ALMA Joint Correlator Meeting
Nobeyama, Japan, August 5-7, 2001
(edited by Baudry and Okumura)
Following the Europe/Japan teleconference on Future/Enhanced correlator plans of July 6, 2001 (see earlier minutes) the European, Japanese and North American teams working on the design of a second-generation correlator met in Nobeyama from 5 to 7 August 2001. Participants in the meeting were:
Europe: Baudry, Bos, de Vos
Japan: Chikada, Iguchi, Momose, Okumura, Ujihara, and Ishiguro on the
first day
North America: Escoffier
The spirit of this meeting was clearly oriented towards a high level of cooperation in view of taking a quick decision on a "Unified Design" for the second-generation correlator.
The meeting started with in-depth mutual exchange on the European and Japanese designs and on the status of the Baseline correlator. Several steps ahead were made during the Technical and Joint-Working Sessions. We briefly report below on the most important decisions and will pass to the ASAC a more detailed report on our 5-7 August meeting just before the next face-to-face meeting in Chile.
We will also send E-AEC a draft on the 3-way work plan for the second-generation correlator to be reported to the E-ACC before the next E-AEC face-to-face meeting on September 28-29, 2001.
A. Following the exchange of a number of documents and discussions on August 5 we have identified nine Areas of Common Interest where technical/theoretical details must be shared whichever final architecture is chosen. This covers a variety of topics from high-speed sampling or quantization/requantization to the location of the correlators. Nine joint reports will be edited with a first version due by the end of September, and there will be a more complete revision at the end of the year. Leaders and co-editors for each area have been identified.
B. We believe that sharing the development/prototyping tasks for a Unified Design of the second-generation correlator is feasible. Before we reach the point of a joint decision for a Unified Design not only reports on Areas of Common Interest must be completed, but in addition
C. We have discussed the process leading to industrial procurement of
the second-generation correlator in Japan. A dedicated and informal business
session was organized to prepare an industrial implementation and quotation
for both current designs.
<<Appendix >>
From: ALMA Joint Correlator Meeting Nobeyama, 5-7 August 2001
Subject: On ASAC Guidelines for the ALMA Enhanced/Future Correlator
(dated 26 July, 2001)
Date: 8 August, 2001
We have discussed on "ASAC Guidelines for the ALMA Enhanced/Future Correlator (dated 26 July, 2001)". We ask some clarification from the ASAC on several items.
ASAC Guidelines for the ALMA Enhanced/Future Correlator
(Sent by Rafael Bachiller on behalf of the ASAC: 26 July, 2001)
The following specifications and goals should be taken
into account by the European, Japanese and North American teams working
in the design
of an Enhanced/Future Correlator for ALMA.
In general terms, the ASAC stresses that the Enhanced
Correlator
developments should be guided by the goals of achieving
(a) high number of channels in wide band modes
(b) high configuration flexibility
(c) high sensitivity
(d) high spectral resolution, and
(e) power consumption as low as possible.
As in previous occasions, the ASAC strongly encourages
a tight collaboration of the different teams to optimize the design and
to select the best possible
architecture and manufacturing method within the budget
limits. What follows is a summary of the requirements and goals which were
put forward by the ASAC at recent meetings.
1.- In addition to the final total number of 12m antennas, the Enhanced/Future Correlator must accommodate the ALMA Compact Array.
Following items should be clarified by the ASAC:
(1). Maximum number of ACA antennas plus that of 12m antennas to be correlated jointly.
(2). Maximum number of 12m antennas to be correlated.
2.- A total number of about 8000 channels is the minimum required.
This seems sufficient for most astronomical observations.
Observations using multiple sub-bands and polarizations
would accordingly have less channels available per product (sub-band and/or
polarization).
A good, more ambitious, goal would be to obtain 4000
to 8000 channels per product. Nevertheless, if the total number of channels
were significantly
larger than 8000, there should be ways of selecting or
compressing them for further processing. The baseline correlator provides
4096 channels in most
modes. When used at the maximum bandwidth, full polarization,
256 channels cover 8 GHz, corresponding to a resolution of 31.25 MHz. With
one polarization,
1024 channels give a resolution of 7.8125 MHz.
Following items should be clarified by the ASAC:
(1). Third sentence: The meaning of "sub-band" is not clear whether it means (a) the passband of the Baseband Converter or (b) a continuous frequency chunk which can be observed with higher frequency resolution such as the passband of each FIR filters in the European approach or the thinly binded frequency chunk in the Japanese approach.
(2). Third and fourth sentences: The meaning of "per product" is not clear whether it means (a) "per baseline"or (b) "per baseline per sub-band and/or polarization".
(3). If "per product" means "per baseline per sub-band and/or polarization", it seems that there is a significant difference between the requirement in the first sentence and that in the third sentence. Required total number of channels should be specified more concretely from the scientific point of view, if possible.
3.- Three-bit digitizing format and three-bit (or even
four-bit) correlation format are recommended to obtain high sensitivity
by diminishing quantization
losses. In widest bandwidth, the baseline correlator
provides a two-bit digitizing format. In narrower modes, three and four
bit correlation are available (though three bit quantization at the digitizers
and FIR filter limit usefulness of the latter).
4.- A highest spectral resolution of 5 kHz is required. This corresponds to 0.05 km/s at 30 GHz, which is necessary, e.g., for the observation of lines in cold dark molecular clouds. The baseline correlator can provide 1.9 kHz single baseband single polarization; it is 15.3 kHz for full polarization single sub-band. As in example D4 of Table 1 of Memo 194, resolution of 1 kHz is possible.
Following items should be clarified by the ASAC:
(1). Total required bandwidth with the highest resolution ( 5kHz ) should be specified.
5.- A reasonable goal for the Enhanced/Future Correlator is to provide 16 sub-bands (in total, not per polarization). Note that the equivalent number of sub-bands in the Baseline Correlator is 8.
Following items should be clarified by the ASAC:
(1). First and second sentence: The meaning of "sub-band" is not clear either it means (a) the passband of the Baseband Converter or (b) a continuous frequency chunk which can be observed with higher frequency resolution such as the passband of each FIR filters in the European approach or the thinly binded frequency chunk in the Japanese approach.
(2). First sentence: Please tell us the scientific meaning of the number 16.
6.- ALMA will have the ability to be split in different
logically-independent sub-arrays, and to observe at a maximum of 4 different
frequencies. The Enhanced/Future Correlator should be able to accommodate
at least 16 independent sub-arrays, which is the number provided by the
Baseline
Correlator.
Following items should be clarified by the ASAC:
(1). We read the word "different logically-independent sub-arrays" as "sub-arrays that can be given different spectroscopic parameters and different start/stop commands or parameters of the integration". Is it correct or not?
(2). Second sentence: Please tell us the scientific meaning of the number 16. Does it mean that 16 is the product of 4 different frequencies (different spectroscopic parameters) by 4 different astronomical sources (different spectroscopic parameters and different start/stop commands or parameters of the integration)?
7.- For solar (continuum) observations, minimum integration
times of 1 msec are required. For spectral line observations, the fastest
integration period
should be made compatible with the minimum dump time.
In side band separation mode, integration times of the order of 120 msec
should be possible.
Following items should be clarified by the ASAC:
(1). First sentence: What is the maximum number of the baselines for solar continuum observations?
(2). Second sentence: What is the maximum number of frequency channels to be output and what is the highest frequency resolution in the observations which need the fastest integration period?
(3). Second sentence: What is "the minimum dump time"?
(4). Third sentence: Please tell us the scientific meaning of "120 msec".
(5). Second and third sentence: These issues cannot be decided within the correlator group. For example, the minimum period for phase switching the LO's will have impact on these issues.
------------------------------------------------------------------------------