On 20/02/2012 16:08, David WE Roberts wrote:

It is official - I am a complete dip switch.

Reset the SCSI ID from 9 to 2 and it all sprang to life immediately.

Thinking about it, I am not absolutely sure why it wouldn't work before.

If you had it on an external narrow bus (i.e. 50 way connector) then it
would not have seen the right address even when the host was polling the
higher numbers.

My first reaction (after seeing the SCSI card software scan for IDs
0-15) was that if it was working on a 3 bit address it would have the
range 0-7 or 4 bit 0-15 so putting the address below 7 would put it in a
valid range for a 3 bit address.
This is standard for 8 bit (narrow)SCSI.
This seemed to work.
But in that case why have a dip switch with values 0 through 9?

Count yourself you lucky you did not need to do it in binary ;-)

Cheap implementation of a single dip switch which covers 10 of the 16
values in a 4 bit address?

Yup... either a hex thumbwheel type or a row of DIL switches to enter
the address in binary seem more common.

Cheap part which can be set from 0-9 but shouldn't be set above 6?
Doesn't seem rational on a semi-professional scanner which cost £1k when
you could go round the world on the Queen Mary for 2/6p and still have
change for the cab.

Anyway, pleased (and relieved) that it is working.
My friend is really chuffed because the test scan of a colour slide was
very good quality, and she got a semi-professional scanner for £10 plus
the cost of the SCSI card.
She will also have to cough for the VueScan software but that looks to
be money well spent as the test version drives the scanner really well
and nobody else seems to do Windows drivers for it.

You may find Silverfast do a version (at a price!)

At some point I may have a play under Linux but the Windows solution
fits in with her other software and all in all she has a really good
scanner for not a lot of money.

Oh, and I tried it without the terminator (just to see) and the SCSI
software reported "insufficient termination" which showed that the
terminator was doing something.

Which is a more informative error than you often get ;-)

The card was (IIRC) an Adaptec 29160N with a 50 way D type connector
externally.

There are a few versions of that IIRC... (have got the LVD version in
one machine, and the older 80Mbit/byte version in this one)

Termination of the card end of the bus is handled automagically.
The scanner had a matching 50 way Centronics style SCSI port, plus
another 25 way D type (which was used to terminate with a short 25 way
- Centronics style cable plus a gender bender passive centronics style
terminator) which also suggests 8 bit SCSI (see above).

16 bit would quite often use a 68 or 80 way connector IIRC...

So the final puzzling part is that it had one connector which could
connect to Wide, but the other connector for Narrow.
Then again SCSI I supports 25 pin D-sub and 50 pin Centronics so this
points to a SCSI I device.
So I say again why the 0-9 ID dip switch?

Just annoyed with myself that it took so long to fix a trivial
error/misconfiguration.

I recall spending hours once getting some trivial data link working...
caused in the end by whoever made the kit applying the label for the dip
switches round the wrong way on the box!


--
Cheers,

John.

/================================================== ===============\
| Internode Ltd - http://www.internode.co.uk |
|-----------------------------------------------------------------|
| John Rumm - john(at)internode(dot)co(dot)uk |
\================================================= ================/