[Oberon] Oberon-specific Spartan-6 FPGA board

[skulski at pas.rochester.edu]

Chris:

> Consider the size & cost of RS-232 for functionality, compared to SD.

SD card is a disk, RS-232 is a serial communication device. Two different
things entirely. Cannot be compared.

This discussion does not seem to be going anywhere.

If one wants to design and make a board then a few things need to become
clear.

1. Board design is a labor intensive process. We at SkuTek have our
estimates based on extensive experience: about three man-months on
schematics & layout is a fairly aggressive estimate for a full featured
BGA-based board of a complexity similar to the BeagleBone (*). Twelve
weeks of skilled engineer's time costing $100/hour means $48k of loaded
salary (I do not have to explain what loaded means?). In this discussion
we are not seeing volunteers willing to donate this kind of effort. We are
not seeing funding, either.

*) I just completed such a design, so this estimate is based on experience.

2. A prototype costs roughly $1k per piece, if you order two pieces for
testing. Even is everything turns out just perfect, you still need to
start with a proto run. A sane designer will not order 100 pieces of an
unproven design.

3. Then you need to start programming. The FPGA Oberon System, as great as
it is, is targeting a particular board that is ancient as of today. It is
not advised to stay with this ancient hardware. A newer FPGA, newer
interfaces, newer memory, or a combination of these, will require quite a
bit of HDL redesign based on thorough understanding of the FPGA
architecture. Lots of testing is involved, HDL simulations, etc. (Usual
FPGA stuff.) Here again, a couple man-months is not unreasonable. We are
talking of perhaps another $20k worth of the engineers' effort.

We went through some of this. Our conclusion was that the Oberon System
HDL needed some tweaking to achieve "timing closure" and to integrate it
with the rest of our system. So here again I am speaking from some
experience.

4. Finally, after all is done, one needs to start selling the boards. This
community will buy perhaps a dozen. Who is going to buy the rest of the
batch of a 100 boards? (You need to make that many to achieve a reasonable
cost.)

Since we are talking of quite substantial effort plus a substantial
investment in the initial batch, the next question is how it could be
funded or recovered? Below are a few possibilities, not all of them
realistic.

a) A volunteer angel steps forward and does all of the above for free.

b) The said angel pays for making 100 bards, sells a dozen, and stores the
remaining 88 boards in a freezer.

c) The Swiss Gov't donates about $70k in order to promote a great Swiss
invention. (We at SkuTek will accept such a grant and we will develop the
invention.)

d) An entrepreneur figures out who needs a few thousand such boards, and
what for, and recovers the development investment by launching a good
business.

As you will probably agree, while the miracles (a, b, c) might happen,
only (d) is realistic in this cruel world. Unfortunately, I am not seeing
much convergence towards (d) in this discussion. In particular, I am not
seeing a good answer to a nagging question "if FPGA Oberon is so good,
then why is it still a small niche?"

It would help if we realize not only why is Oberon so great, but also what
is it missing? Also, what is this System good for? What applications it
can do better than any other system on the market?

If we figure out these questions then maybe someone will do good business
with the FPGA Oberon. Otherwise, this community will be stuck in endless
discussions as we always have been.

Thank you,
W.