Some Comments on Clark's NMA Cost Memo
Sandy Weinreb 27may01

Here is a copy of my May 27 reply to Barry Clark. It has no secrets --- you can distribute it as you wish. Bottom line is that we need to see how the ATA comes out and whether SKA development work leads to a low-cost 100 GHz antenna in the 12m range. A good development job for NRAO would be to design a 10 to 50 GHz or even 100 GHz log-periodic feed in a cryogenic vacuum dewar.

From: Sander Weinreb
Re: Multielement station for eVLA


Thanks for the excellent memo on eVLA station cost. I hope you do not mind if I forward it to Larry Daddario and Durga Bagri. Larry and I are planning to present talks on the SKA cost equation at the Berkeley SKA meeting; I will cover the antenna and front-end costs and Larry will do the data processing cost. I hope you are planning to attend this meeting, July 9-12, and perhaps will also give a talk on costs.

Some comments on your memo:

  1. Antenna costs: I think that the ATA pedestal and drives will increase the reflector cost by 2 but the increase in cost due to needing more accuracy may only be 1.3 so your cost equation $130*D^3.1 may be OK.
  2. Electronics costs: I think your $240K estimate is too high even with the 2*8 GHz bandwidth. The 7-50 GHz and 80 GHz receivers should be combined in one dewar and with MMIC LNA's I think $45K would cover both. We should be thinking at least 3 years in the future and Moores Law will help the digital costs by a factor of 4. My guess is $120K but we need much more detail work to support an estimate.
  3. The minimum total cost with antenna cost going as D^3.1 is at the antenna cost 1.82 times the electronics cost. At $240K electronics and using your antenna cost equation this is a total of around $3.1M with 4 x 12.5m antenna; at $120K electronics this is $2.5M with 6 x 10m antennas. These compare with around $3.8M for the 1 x 25m antenna.
  4. An array makes much more sense for an SKA station which will need a lot more collecting area than a 25m antenna. Yet the optimum antenna size could turn out to be as large as 25m. To meet the Ae/Tsys SKA figure of merit goal of 20,000 we would need around 2000 x 25m and each would need to cost under $300K - an order of magnitude less than present 25m costs.

We need better cost estimates and we will be getting them from the ATA experience and SKA studies that are planned. It is important to find out how the cost of hydroformed antennas goes with diameter and upper frequency and what is the cost of an on-site assembly plant. We also need to assess how factors other than cost effect the question of antennas per station - field of view, dynamic range, and maintenance.

I have attached a .pdf version of your memo and a brief .xls file of my cost analysis using most of your figures but also the $120K electronics cost.

-- Sandy