DEC bus transceivers

allison ajp166 at verizon.net
Wed Oct 26 21:19:17 CDT 2016


On 10/26/2016 12:40 PM, Noel Chiappa wrote:
> Re: DEC bus transceivers
>     > From: allison
>
>     > Actually since about 1987 I've used about 1200 pieces of the 8641 alone
>     > repairing boards at the commercial level.
>
> Well, that's over almost 30 years - and your total from that period is about
> 4% of the remaining stock (and in a commercial operation, to boot, not
> hobbyist)...
Its also a lot of boards fixed and sold.  It was a good second income.
that has consumed about 40% of what i have and that doesn't include
the rare like DC03/4/5/10 and others some more common like DLART,
T11, DCK-11-series, and w95xx boards.

I had the gift of being a central engineering milrat at DEC.  So many
of the design protocols for many of the DEC buses are hardwired in
my brain.  They were building systems not just boards.

>     > If you going to build a board or three maybe even 20 its not a big deal
>     > but its not a reliable source of predictable quality.
>
> Sure, but try looking at it from our perspective: we either use an
> out-of-production part, or have to design something (almost certainly from
> discretes) that meets those specs; and we actually looked at the latter (viz.
> Dave B's design). However, after some pondering, and taking everything
> (including all the below) into account, we decided to go with the original
> chips, since they were still sorta available.

If you use the DEC part you don't have to design, its plug and go. 
Design means
you evaluated something and its application and went with it as it met the
defined need.

I've done the latter and have many tubes of parts that work quite well.
All the designs I did that were unique ground up used those as the packages
work as expected.  The key is thresholds and current sink.

Having measured parts both dynamic and static I have a good feel for
what will
work in all but the pathological cases.   For the pathological cases the
biggest
issue is current sink. But like I said there are systems I run and do
not to try
and make meet DEC supportable configuration standards which means
things like third party cards and one-offs  and those that are museum
level stock to the bone.

> Which is why both we and Guy have stocked up on them, at the start of the
> process: we don't want to crank out boards designed for a certain part, and
> then not be able to get the out-of-production parts the boards were designed
> to use.
>
> If we were designing something for serious production, that wouldn't be an
> option, but for limited-volume hobbyist use, it is. The choice of an
> out-of-production part does have a down-side, but it's minor (and mostly
> alleviated by the pre-buying), and the other options were (in overall sum)
> worse.
>
>     > If you get to the bridge your talking redesign in reality or an
>     > expensive buy from unreliable source then testing them in bulk.
>
> But, but... I'm _already_ buying them from unreliable sources, then testing
> them! :-)

Therein lies the problem.  My purchase was back when the largest network
was still owned by DEC.  and it was the year they were just EOL'ed
(every one
was doing last time buys with lead times of twenty plus weeks.  I've
dealt in
the grey market and  for other things I can say often the profit taking is
extreme and they know the barrel your over.

>
> But to be serious - if the demand for QSIC's, etc, runs the well of DS8641's
> dry, yes, we'll probably have to re-design. In other words, we'd be right
> where we'd be today if we decided not to use out-out-production parts.
Likely the redesign would be more along the line of different package
(maybe smaller) and the odd inversion or not.


Allison





More information about the cctech mailing list