[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Picking a Mark IV Lens
- To: firstname.lastname@example.org, "Griffing, Dan" <DAN.GRIFFING@kla-tencor.com>
- Subject: RE: Picking a Mark IV Lens
- From: Tom Droege <email@example.com>
- Date: Fri, 16 Jan 1998 16:35:12 -0600
- Old-Return-Path: <firstname.lastname@example.org>
- Resent-Date: Fri, 16 Jan 1998 18:06:32 -0500
- Resent-From: email@example.com
- Resent-Message-ID: <"xI_NDD.A.ECC.EG-v0"@kani.wwa.com>
- Resent-Sender: firstname.lastname@example.org
Thanks for your comments. One reason for putting this up is
to give a pitch for engineering considerations in designing a
science project. Most of my friends in science don't make this
kind of analysis. The biggest error that is made in experimental
design is usually to push too hard for the best solution. This
leads to engineering problems, and delays. When all is done,
less data is taken, and it is of lower quality than it would have
been with an easier specification.
At 10:14 AM 1/16/98 -0800, you wrote:
>Aren't you mixing two separate issues. Star magnitude
>affects the stars you can see in the whole field of view,
>even in the center where its clearest. Your 70% estimate
>of good area on the Mak affects the width of the useable field
>which will affect the number of units you will need to scan the
>same area of sky, but not the end result.
Hmmmm! Seems the result is the same either way, less
data per dollar.
>I also wish to caution you about the big factor of risk of
>committing yourself to a course of action involving a
>large unknown, i.e. the development of a "home-built"
>system in which you will have to be successful in
>so many areas for the assumptions of your analysis
>to be true. With the Mak, nearly everything is
>known about it. For the Mak unknowns, you can
>build a single prototype without being committed to
>using the Mak for the whole project.
You bet! I thought I pointed this out pretty strongly.
But I think if anything, I have understated what the better
lens will do for us. But it *is* high risk! That is what I
get paid for. Making decisions like this. I succeeded
in making a number of critical decisions like this for
the CDF experiment. Funny, you can get away with
making the vital decisions, but the trivial ones are
critiqued out the wazoo! You can propose some critical
analog scheme for sampling the data, and the eyeballs
just glaze over. But talk about piping in water for cooling
something and the debate rages for days. For several
years now we have been having a debate on a
project about whether or not cowboys will shoot holes
in our water tank detectors. Much more critical problems
go un discussed.
>On the other hand, the home-built must be designed
>and constructed first which is a significant percent of
>the project's cost which must be spent even to test
Yep, it is a high risk decision. But one that I note that
the scientists in the group immediately endorsed. You
bet I will look very carefully.
>My bottom-line advice is that it is usually more
>financially prudent to buy something "off-the-shelf"
>than to design it yourself, unless you have an
>overwhelming justification to do otherwise.
I could not agree more. So far, I have not found the
right thing on the shelf. I keep looking.
>On the other hand, just wanting to design and build
>something yourself is part of what amateur astronomy
>is all about and it is such a justification, especially if
>its your own money. But the home-built option is a
>more risky and less financially conservative path.
Yep, but I am in production. That is I plan to make a bunch
of these. We have talked of a run of 40 lenses. But still high
risk. I, of course, proposed 1 + 40 in a two stage procurement.
But it looks like one will cost half as much as 40, and would not
go to the same vendor. So you can't escape the 40 stage risk.
>I do applaud your financial analysis, however, in evaluating
>the end goal and how much each incremental unit of quality
>will cost. It is quite proper to evaluate ends versus means
>especially with respect to those who will have to be putting
>up the means.
Means is me here. So I really worry.