Dear Dark Energy Camera Pundits,

Here fyi is a very informative message from Paul Jorden at E2V.    It should give us a basis for discussing CCDs for this project on Monday.

Alistair

Hi Alistair,

It is good to hear from you, and interesting to learn of your new project-it sounds ambitious and exciting.  I had a brief dialogue with Tim in July about a large mosaic, but he asked me to keep it quiet, which I did. 

Please give Tim my best wishes- I hope he is fit soon.

I’ll try and respond to your queries:

1. We have completed three major contracts for ground-based large mosaics-

CFHT Megacam      (40 chips) completed on schedule; instrument now in use.
SAO Megacam        (38 chips) completed on schedule; instrument in assembly.
ESO VST                 (40 chips) completed this year; instrument nearing completion.

We are also part way through a contract to supply 60 chips for the Kepler space mission, and on schedule.

Of course, we have also supplied chips in various scopes over the years for other projects.  We have started work for ESA’s GAIA mission which will have hundreds of large chips.

I attach a copy of my San Diego 2003 SPIE paper about focal plane mosaics.

2. We generally say 18-24 months for ~ 40 chips, and so ~24-30 months for 70 chips would be reasonable.  This is for an existing design, made at a reasonable rate.  This is how it works- after contract signature we initiate manufacture, so there is an initial manufacture time ~ 6-8 months to complete the first ones.  Then we start continuous manufacture and testing, at a rate of ~ 3-4 delivered chips/month. Of course, we have to test more devices than we end up delivering, and so our test capability can be a limiting factor; it is possible to do it faster, but perhaps at higher cost, or with less contingency.

If your selected device is a completely standard product, we may have some in stock; we would also expect to be able to supply setup devices somewhat ahead of science ones.  But for a large build, then the manufacture and test time is always needed.

For our large (ground-based) projects we have taken contracts with a ‘global’ cosmetic spec. ie specify total number of defects, rather than demand every chip reaches a grade-2 spec (say).  this gives us more flexibility on delivery.  Certainly if you demand high specs on all parameters, then this is a big yield factor that impacts price.

3. There is nothing much to choose between ccd42-90 or ccd44-82.  If you prefer 15 um pixels, then I would go for this.  The CCD44-82 chips have a slightly lower noise also.

A ROM (budgetary) price would be ~ 3m USD. for 70 chips covering grades 1-3 (say).  it is unlikely to be much below this level.  make your own guess about exchange rate drift & inflation! This assumes that the specs are not demanding, and based on our standard device type.

4. We have made and supplied quite a few deep-depletion CCD42s and CCD44s.

These are 40 um thick (cf 16 um std) and so have better red response. 

There are some data sheets on our web site.

We have also recently made some developmental ‘high-rho’ devices

(similar to LBNL) of nominal 300 um thickness [small TV-format test samples].  These offer even better far-red response. I attach a note on typical QE of these various types, just as a guide. AR coating needs to be matched to the desired spectral range; obviously a wide range is hardest.  If you only wanted red (eg > 700nm) then a good optimisation would be expected; although at the expense of lower blue. Many ‘survey’ mosaic scientists would not want to compromise on blue response, but in reality if you sacrifice this it will allow a better red device, and could be cheaper- since good blue response takes extra work.

For d-d types we normally charge a 20% premium on price; we may review this in the context of a large order, but take this as a guide for now. We have not yet estimated the delta price for a high-rho type, since these are only at the prototype stage. Assume at least 20% extra. Schedule is probably about the same for all types, except that we have not yet designed and laid out a high-rho version of a large-format chip.

5. We have recently designed and made a 4k*4k, 12um chip.  I expect us to be selling this to astronomers next year.  This would also be an option for your project I would imagine, if you can live with the 12 um pixels. We would expect to sell a 4k*4k chip for somewhat less than a pair of 2k*4k ones.  However, for a large focal plane the fill factor probably does not differ much and maybe a monolthic 4k*4k does not offer enough advantage.  [I suspect that the main market will be for echelle spectrographs of small FOV imagers].

In order to make a 4k*4k chip with 15 um pixels we need to develop our technology, which is in-hand.  However, we do not expect to be in this position until 2005, so maybe this is a little late for you.

This is how I would estimate the timescale-

• commission camera (on telescope?) in mid 2007

• instrument builder should have all chips by end 2006- latest; and this may be tight.

• chips should be ordered by end 2005 (calendar year)- latest

• contract discussions to be done in 2004; these usually take some time [> 6 months?]

In terms of risk-  standard silicon or our deep depletion type would count as low risk.  The high-rho is newer, although I hope that within 2004 we will know a lot more about this.

As you probably appreciate, based on past experience, we are comfortable about quoting to supply this sort of quantity, and delivering to a schedule with guaranteed performance and price.

I guess this is enough information for now, and perhaps will be useful for your meeting next week.

I’ll be happy to discuss this further with you.  It sounds like a meeting bewteen CTIO and e2v would be valuable sometime.  Are you, or any colleagues going to the Cozumel mtg next month?

Best regards,

Paul 

Dr Paul Jorden,  CCD Sensors, e2v technologies
Waterhouse Lane, Chelmsford, Essex, CM1 2QU, UK
Tel:       44 (0) 1245 -453458 (direct), -493493 (switchboard)
Fax:      44 (0) 1245 453224
http://ccds.e2vtechnologies.com/

-----Original Message-----

Dear Paul -

We are about to issue an Announcement of Opportunity for people to propose to build a major new instrument for the Blanco telescope, to be commissioned around 2007-2008.   Tim Abbott may have mentioned this previously to you, but he is off on medical leave at the moment and I have not seen any correspondence on this matter.  We have had some initial discussions with a group that is very interested in building, in partnership with NOAO,  a prime focus imager with a FOV 8-10 times larger than our present 8Kx8K mosaic Imager.  Considering the instrument, the data pipelining, and an associated 5-year project that the proposers would do with a substantial grant of observing time, this is a ~$20M project.

With the time scale as above, this would be a rather fast-track project and one which in my opinion any R&D should be minimized, and the risk areas clearly identified. The focal plane is as always one of these.  I am talking to the possible proposers next Monday (note - they have lots of instrument building and silicon experience, but know little about CCDs) and would like to lay out the options, and so have a number of questions for you.  Your answers to these will in no ways be considered binding, and you are at liberty to not reply, or conversely to supply answers to questions you think I should have asked!

1. Could you list the “big” contracts you have successfully completed recently, - e.g. Megacam at CFHT, VST camera, etc etc.

2. For a camera with about 70 CCDs (assuming 2Kx4K) can you give a rough estimate of the time needed to deliver such a number of CCDs from initiation of a signed contract.  Is the delivery rate approximately linear?  Assume “standard” 42-90’s or 44-82’s.  Probably Grade 2?  Comments on grade choice are welcomed.   I think we would prefer 15 micron pixels over 13.5 microns.

3. Are 42-90’s and 44-82’s considered to be the same from the point of view of supply?  Or is one to be preferred over the other?   Would a focal plane with about 70 Grade 2 CCDs be worth in the vicinity of $3M, or would you want me to write a formal RFQ before answering this?

4. The science drivers may well push to have as high QE in the red (700-1100 nm) as possible.  Can you comment on your capabilities in this area - AR coatings, deep depletion devices etc.     Separating out what are options, and what would involve extra cost and schedule would be useful.

5. If you are developing new devices on a 1-3 year time scale that might be relevant to this effort  then I’d be interested in hearing about them.

best regards