name resolution

[Gene Heskett]

On Sunday 26 November 2017 06:28:33 Liam Proven wrote:

> On 26 November 2017 at 09:41, Gene Heskett <gheskett at shentel.net> 
wrote:
> > And after you'd fixed that again , it was still the termination.
>
> Ex-ACT-ly!
>
> > 99% of
> > all the termination problems with scsi is that the designers of the
> > interface cards specced a schotkey diode for the powerline isolation
> > diode, but the fscking bean counters changed it for a std si diode
> > on the way to the production floor because it was $3 cheaper, with
> > its .7 volts voltage drop across it, which in turn reduced the logic
> > one noise margin to zip, so the bus errors were at least once per
> > write. One company that makes cards for the amiga even drew up the
> > artwork for resistive terms bass ackwards, so in addition to putting
> > a schotkey diode in it, I had to cut the traces and reverse the 5
> > volt and grounded ends of all 3 of its soldered in terminators. That
> > card never again had a bus error that wasn't a fading psu. The
> > termination was finally fixed. Logic one resting voltage was finally
> > at 3.3 volts, the design target for a scsi bus. But watch for fading
> > psu's, long before the logic or anything else complained because the
> > 5 volts was down to 4.7, screwing the logic one noise margin, the
> > buss errors were raising their heads again.
>
> That's... terrifying.
>
> I am, sadly, almost totally incompetent at electronics. Stuff like
> this was well over my head.
>
> I've certainly seen machine/drive combinations that Just Would Not
> Work, though, when both tested fine with other devices -- and now you
> make me wonder if it was electronics/component-level problems.
>
It was. the original scsi  bus had a max cable length spec in excess of 
100 meters!  Because it was envisioned as a transmission line. But to 
function with that length of cable demanded near perfect terminations on 
both ends.  Thats not possible without higher priced custom components 
that aren't made in Orville Redenbachers production lines. So by the 
time the bean counters edited the BOM as it went by towards the 
production floors, it was doomed to be a problem child. I do not know 
the "characteristic" impedance of the later, much finer pitched scsi-III 
cables that used the miniature d-sub connectors, but I can easily 
believe that it was just as fussy if not moreso. The so called active 
terminations were actually a miniature power supply regulator set at 3.3 
to 3.4 volts, that could also sink some current if the line should 
bounce above, but whose output network consisted of a single 120 ohm 
from its output to each individual signal line. It even worked fairly 
well against the usual resistive termination of a 220 ohm to the 5 
volts, and a 330 ohm to ground, which by ohms law for parallel R's was 
132 ohms, a hair high for the cable, said to be 120 ohms. But no other 
pair of std r's in parallel even came that close so there was bound to 
be some ringing on signal edges. 2 active terms facing each other from 
the ends of the cable were close enough to tolerate even the more 
cheaply made cables fairly well. Then came the ide which made the 
terminations automatic and poorly done, limiting those cables to about 
18" in total length.

Even todays sata is not just a wire, but is in fact a transmission line. 
Making sata do an automatic fallback to as low as 1 gigabaud when it 
detects the first sign of a poor termination, silently unless you 
are 'tail'ing the log, has made sata a quite acceptable thing to the 
masses as to them, it Just Works(TM). In that sense, it really is a 
roaring success.

> > But that was the 2nd time in my life as a journeyman C.E.T., which I
> >  am, that I called the maker and gave him hell, telling him the best
> > part must have run down his mothers leg. Totally unrelated but
> > important in a production room for a tv station where we had quite a
> > few amiga's doing video work, and several of the A4000's had 68060
> > cards in them, I found that some production people at Commode Door
> > had installed all the 6.3 volt electrolytic bypass caps backwards on
> > that $1500 card.
> >
> :-o
> :
> > Took about a
> > year before they started blowing out the tops. With the soldering
> > tools of the day, fixing that wasn't my idea of fun. But by then
> > there wasn't anybody answering the phone at Commode Door.
> >
> :-(
>
> The early Amigas were a premium product, but had to compete in a
> mass-commodity market.
>
> > Just before I retired in 2002 as the CE there at that tv station, we
> > bought some Apple G5's to replace the failing amiga's, but they
> > should have come with fire extinguishers. None of them lasted 6
> > months.
>
> Wow!
>
> That's horrific.

Any slowing of any fan was deadly. Air was so carefully steered around by 
the plastic moldings used to direct the cooling air within them was 
deadly and you wound up with a motherboard that was well and truly fried 
to a crisp. They did of course usually last the 90 day warranty. I think 
later G5's must have had much better fan bearings. But imo at nearly 6 
grand a copy, they should have been first class fans in the first place.

When that nibbled on apple wanted nearly 3 grand for a populated 
motherboard, then took almost 90 days to deliver it was when our IT guy 
started building our own machines. Quality motherboards, running linux, 
are still running, albeit with fresh drives since 15k rpm drives aren't 
long lasting drives, are still running a decade later serving up 4 hidef 
video streams while recording two or 3 in the background. He intended to 
build a machine that Just Worked, and has been wildly successful doing 
it. Raid arrays of course. A drive can fail, flip the lever and slide in 
a new drive, close the lever. It can rebuild that drive while still 
serving up 4 channels worth of hidef tv. Do that times 3 and it never 
fails to deliver. But the owner died 2 years ago, and it was sold last 
spring for an unbelievable price, and the new owners want to replace all 
that with windows crap. How many ways can you spell stoooooppiiiiddd.

> Time did show that the G5s *were* pushing the PowerPC envelope, but
> still, they were _beautifully_ built machines. I still have one, over
> a decade old and working fine.
>
> > Which, if dealing with a sata thats flooding the syslog with bus and
> > drive resets, and the sata cables are that pretty hot red color,
> > replace them with ANY other color of cable, that plastic dye
> > converts the copper its in contact with into powdered copper oxide.
> > Very poor conductor.
> >
> :-o
> :
> > That plastic dye was known to be a problem by the early 1970's as
> > cable production moved to the J.A.Pan company, but all the cable
> > makers could see was a booming business selling replacement cables
> > as it was always the red wire that failed. See it, replace it with
> > any other color.
>
> ... Wow.

Liam, it turned out at the end of the day, that we were closer than some 
might have thought from watching our exchanges on quite a few things. I 
like it that way. :)
> --
> Liam Proven • Profile: https://about.me/liamproven
> Email: lproven at cix.co.uk • Google Mail/Talk/Plus: lproven at gmail.com
> Twitter/Facebook/Flickr: lproven • Skype/LinkedIn/AIM/Yahoo:
> liamproven UK: +44 7939-087884 • ČR/WhatsApp/Telegram/Signal: +420 702
> 829 053

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>