NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/sys/amiga/hardware-part1/2909

Internet Message Format | 1996-08-05 | 8.7 KB

Path: sdd.hp.com!inn
From: Jeff Grimmett <[e-mail]>
Newsgroups: comp.sys.amiga.hardware
Subject: Re: A3000 SCSI
Date: 29 Jan 1996 18:03:54 GMT
Organization: Hewlett-Packard Company
Message-ID: <4ej26a$[e-mail]>
References: <4crkgh$[e-mail]> <4djffa$[e-mail]> <4dlre0$[e-mail]> <4e0amr$[e-mail]> <4e0jru$[e-mail]> <4edjsc$[e-mail]> <4egdq5$[e-mail]> <4eglri$[e-mail]> <4egrbq$[e-mail]> <4eh2tp$[e-mail]>
NNTP-Posting-Host: hpsdv330.sdd.hp.com
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 1.2N (Windows; I; 16bit)

[e-mail] (Michael van Elst) wrote:

>>Fine.  Go make a printer cable work with a SCSI drive.
>
>Might even work. After all there are things like the ZIP drive that
>also uses DB25.

"Might even work" doesn't fly.  MAKE it work without hacking it. A3000.
To, say, a DJ1200.  Or an Epson Stylus.  GO for it.

>>Without hacking
>>it.  Standard 25-pin to 50-pin SCSI cables are the subject at hand.

>And if you take a look, these are mostly 25 wire cables (that lack
>the right number of ground lines, don't avoid crosstalk, have the
>wrong impedance, etc...). Some even don't have all the ground pins
>on the 50-pin side connected to a wire.

(a)  The number of wires doesn't make a difference if they're not used
right, and usually make matters worse.

(b)  Crosstalk, impedance: what do you think one of the functions of the
terminators are, after all?

(c)  If you don't have all the "ground wires" then where, exactly, would
you connect the extra pins on the far end?

(d)  A lot of this is irrelevant if you get the good cables, anyway.
Sure, they don't have 25 twisted pairs going through the cable, but if
that's done incorrectly to begin with, you can twist your little heart
out and it won't make it magically work better.

>If you get 50-wire twisted-pair cables with DB25 and 50pin Centronics-style
>connector for $15 then you are pretty lucky.

Oh, yeah... me and a million other potential customers in San Diego.
That's the going price for good-quality "mac-scsi" cables around here.
And the $66.00 high-density cable was bought at exactly the same place
and was the exact same make and build quality.

Something you should know: I have to tear any cable that I consider using
APART to prove it's usefulness.  We can't afford to accept a
manufacturer's claim to actually have implemented shielding or whatever
properly, we have to verify it ourselves.  Price was not an object, I
bought and tore apart over a dozen makes and models of these things. I
know exactly what's in there.  The AVERAGE 50-pin to 50-pin SCSI cable is
no more "correct" than the average DB-25 to 50 pin SCSI cable. They make
the assumption that ground is ground and tie both ends to the shield,
instead of sending 25 twisted pairs over AND a shield.  In the former
case, your precious 50-pin to 50-pin cable is no more effective than one
with a DB-25 connector on the host end, and in a lot of ways is worse
since noised caught on the shield is actually conducted onto the return
lines in the host AND the drive.  At least with a DB25, the host is
built to contend with that.

>>Incorrect.  As I POSTED, and as you conveeeeeeniently cut out in your
>>reply, the same cable made by the same company with the notable exception
>>of the connector on the host end costs $15.00 for the DB25 version,
>>$60.00 ($66.00 to be precise) for the high-density version.
>
>Have a look wether the DB25 version actually mets SCSI specs if you
>ignore the DB25 connector (that obviously violates SCSI specs).

Perhaps you should prove that it isn't.  Either viewpoint is equally as
valid.  Even better, prove that the OTHER cable IS compliant.

>And yes, you might see higher prices for high density cables. Others
>do not. Good SCSI cables cost between $40 and $60 here. And that's
>independent of high density connectors. And no, neither of these use
>DB25 connectors.

>>By your own inexplicable reasoning, then, the A3000 is a non-SCSI machine
>>because it uses that connector.
>
>It does not fully adhere to SCSI specs. Correct. I wouldn't go so far
>to call it non-SCSI because it is still close to the specs.

Close is not good enough.  It is compliant or it is not.  I WORK in the
regulatory testing field, Micheal.  This is what I do.  I've had it up to
here with "mostly passing" from engineers, it doesn't work for them and
it doesn't work for the A3000's SCSI either.

>>It (a) violates SCSI specs and (b) does
>>not work reliably (as quoted from [e-mail] (Michael van
>>Elst)).
>
>But it does work reliably.

That is not what you said.

>>>>Generally, it does.
>>>It always does.
>
>>Dream on.
>
>That's not a dream. That's _personal_ experience.

Ah.  So, I get the picture.  Since it works in Micheal's machine in a
certain fashion, the laws of physics must suspend themselves in order to
bring the rest of the universe into alignment with it?

No.

I WILL accept "it works for me," as long as YOU do not contend that "it
works for me" MUST universally apply to every other 3000. I'll be the
first to admit that that's the case in my situation, and have clearly
stated so on several occasions.

>>>>For a non-theoretical piece of machinery, it MAY make sense.
>>>It makes perfectly sense for real-word machinery.
>
>>Dream on some more.  Go out and buy a few modems and prove it. (laughing)
>
>Weren't we talking about SCSI and the fact that adhering to specs produces
>reliable machines ?

Right now, this very moment, we're talking about the fallacy of assuming
hardware works according to spec simply because it CLAIMS to. There are
many v.34 modems that don't live up to the v34 spec.  Don't take my word
for it, there are quite a few reviews available on the web these days.

>>Of course, only the ones that you personally agree with could be right?
>
>Of course not. But if such a bulletin contradicts personal experiences and
>knowledge I may have reasonable doubts about its validity. You don't ?

Only if I have proof that I know everything.  Since I don't, I'm willing
to admit that the manufacturers of a particular piece of equipment MIGHT
be a little more knowledgable than I am about it.  If you claim superior
knowledge over the designers and builders of the machine, I am afraid
your credibility is going to take a dive.  The whole point of a technical
bulletin is to point out something that needs correction in the field.

>>Perhaps you should stay out of this one,
>
>Yes, maybe. It's a long time that I played juvenile games.

Well, be assured that you have not lost your edge due to inactivity.

>>>These have to be handled. But not in the way you suggest.
>
>>I suppose we just close our eyes, and wish all the big nasty problems
>>away?
>
>What for ? We try to understand the problems. People with better knowledge
>than me or you did this before, some of them wrote the SCSI specs, and it
>seems that these people were perfectly right.

.. and other people with more knowledge of the specific MACHINE wrote a
technical bulletin stating that thier machine under certain circumstances
did not quite adhere to that spec.

>>There you go again.  The A3000 isn't the physical embodiment of the SCSI
>>Standard and EE theory.
>
>It is close enough to what the SCSI standard describes and of course it
>can be perfectly handled by EE theory.

Several paragraphs later, "close enough" still isn't going to fly.

>>To say the A3000 is imperfect is not to say
>>either of the others are.
>
>You misunderstand. The A3000 _is_ imperfect. The theory provides an explanation
>and solutions. These solutions do even work, for me and others. Actually they do
>for all people that I met that had such problems. The technical bulletin you
>are referring to does not provide a reliable solution. If you now insist that
>your method is still right then you simply disagree with facts.

It's not my method.  You are not reading everything I say on this matter,
obviously.  Specific revisions with specific quirks.  The people that
arrived at this solution are more qualified than you or I to make that
judgement.

>>>_No_ A3000 _requires_ non-standard termination configurations.
>
>>Oh, sure, if you say so.  Wonder why that's not convincing my 3000 of
>>anything in particular?
>
>Probably because you lack some knowledge about your A3000.

That's rich.   Next thing you'll be telling me is that modems work better
based on the owner's sex appeal.  The hardware doesn't care.  It doesn't
care what I know or don't know.  It DOES work within the framework
established by CATS and/or CBM engineers for that particular revision of
main PCA.  And it doesn't care what your opinion on the matter is,
either.