Gain Stability Meeting FE/BE/SE/Sci IPT telecon 15 July 2003 Scott, Janes, Sramek, Tan, Cunningham, Wootten, Tan, Haupt, Payne Not present: van Dishoeck, Richer, Holdaway, Baudry Is Allan variance the correct quantity? General agreement, though this is not specifically what Jack said in his memo. Since we must deal with differences, Allan variance seems like the right measurement. Memo No. 466 has some issues for the science group to study. The atmospheric numbers used in that memo are based on an old set of data; Holdaway can provide better data now. Basically, use of the old memo numbers would result in a more pessimistic assessment of sky variations at high frequency than would be warranted. Science: The critical science is on the fly continuum, a total power measurement. It is best to have stable receivers on all antennas. It is possible that if it is not possible to reach the critical level of stability, the on the fly continuum science could be performed by a subset of dedicated antennas. Such a subset can be identified as the four dedicated total power antennas, operating in concert with the ACA and equipped with nutators. For spectral line on the fly measurements, sensitivity is needed, a primary reason for having built 64 antennas. All 64 antennas will be used in total power on the fly spectroscopy. However, for spectral line on the fly measurements, the stability is not so severe as the bandwidth is more limited. We have no current specification for this. Holdaway is currently performing simulations for total power observations with ALMA in an effort to understand the coupling of receiver gain variation, sky variation and other factors. This should be reported for the ASAC meeting. I think that Larry's proposed testing scheme is a good basis for discussion. Thacker and I tried to do similar measurements on the preprototype 3mm receiver in GB in May though of course we had to use the GBT Spectrometer and aips++. Wide range of 1/f noise is measured for many devices. None for the three stage CV amplifier for Band 6, but LD predicts this should be a bit below 10-4. Thermal stability influence gain also and may dominate. With our refrigerator the time scales approach that of the cycling. IRAM rx have excellent stability with a hybrid cryostat. This suggests that temp stability is a critical factor. Welch addressed this directly. 2 -3mK is achieved with our cryostat. GHT thinks temperature stability effects on the SIS mixer are not dominant. Tests have not been done with temperature stable Front End. CC says T stability of Band 6 measurements is comparable to what well see with Sumitomo cold head. CC says most of this comes from temperature fluctuations. Can one really control the temperature on one second timescales. Some question on this. GHT thinks that over minutes this may help but on one second timescales he has doubts. LD you have to arrange not to have 10s of mK variations on short timescales. With small parts this may be possible to control. If one can regulate on long timescales then on short ones it will be easier; but this depends on many things, including detailed understanding of what causes the thermal fluctuations. Every part of the system needs detailed attention. Backend stability. Brundage discusses this and notes that variation in system temperature also enters. He knows temperature coefficients of components and is trying to control temperature to control stability. 2nd LO has been thought of also. The sensitivity of the second mixer with conversion loss is treated in the ICD. The spec includes downconverter gain variation .1 dB per dB. Pretty much means run the mixer saturated. GHT: First LO is another concern; not so much saturated. LO power makes about half of the gain variation of the Band 9 SIS mixer, I recall. LD: Operate at optimum and then this won't have so much effect. But with all these in the field, some will not be set at the optimum point and there will be some sensitivity of gain to LO power. CC: I wouldn't accept that it is second order. LD: It goes as the square of the difference of the LO power from the optimum. CC: There is a stability spec on the first LO. .03% is the value. LD: How is that measured? CC: I don't know; you wrote the spec Larry. This is the variation in power that one could expect as provided to the SIS mixer. RS: Is a constant temperature load feasible? GHT: It is on the edge but it is probably achievable with care. CC: Band 6 noise floor should be known but we should investigate. Call Effland? GHT: We have some test results; we need to sort them out and present in a structured way. AW: Present Larry's suggested tests. GHT: Needs to have a number for the ESO contracts asap. We need a figure that there is consensus that we can achieve; we cannot push groups in some direction without such consensus. RS: Contracts are best effort; why can you not just say here the specs are, make your best effort to meet them. In several months, the project will have to accept what there is or accept the delay or the achieved performance. GHT: We just don't want to scare the potential contractors off. LD: At higher frequency bands, which are the EU ones, there is no way we'll get to thermal noise owing to the atmosphere. Requirements at high frequencies may be relaxed, though what is needed to achieve gain stability at each band is quite similar. Room temperature components are common. If we can achieve this at low frequencies then we should be able to achieve it at high frequencies, given the atmospheric instability. Science group will evaluate the atmospheric contribution. CC: If we do come up with numbers which are realistically achievable will this result in a product which is acceptable to the users. GHT: Well, just go with 1 in 10^3 AW: Absolutely not! That is just a copout. We need to try to maintain stability with great viligance at each step of the chain, measuring in accord with a set standard, for instance that proposed by Larry. Another meeting proposed for Weds 27 August. 1430 UT.