From: "Christopher R. Carlen"
Subject: Re: 1523 joints
Date: Wed, 23 Oct 2002 08:08:14 -0700
Organization: Sandia National Laboratories, Albuquerque, NM USA
References: <3DB5C12B.firstname.lastname@example.org> <3DB5F44A.email@example.com>
NNTP-Posting-Date: Wed, 23 Oct 2002 15:07:11 +0000 (UTC)
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20020826
X-Accept-Language: en-us, en
Ken Taylor wrote:
> Hi Chris!
> I didn't realise that he tech's were mechanical, not electrical/electronic.
> That can make it difficult. However as another poster commented they may
> still react well to the challenge of the new and different!
Yes I should be keeping an open mind to such characters.
> It appears you have a set of options:
> 1/ you could do it all yourself, which it sounds like you don't have the
> resources to do;
> 2/ you could oversee someone in-house to do it, which appears to be a
> resource (person) problem also;
> 3/ you could outsource it with a sample provided for guidelines; or
> 4/ you could redesign for manufacturing ease.
> Are you able to redesign for example? Maybe by placing all the
> connectors and LED's onto another board which would be much simpler and
> could mate via some inter-board connectors (maybe even directly onto the one
> you've already designed - instead of mounting the LED's and BNC's and so
> forth on the main board, mount connectors which mate with another board
> which actually has those components on it). In this way you can divide up
> the task into the high-density soldering problem and the finicky mounting
> problem, and give the two tasks to people who can work best with each, and
> can utilise the PCB you've already designed.
> Either that, or gain some trust in an outsourcing place. Options 3 & 4 seem
> the only ones really likely at this stage.
In this case, I have a digital logic patch panel, with 32 BNCs
interfaced to a CPLD with selectable input or output buffers. The LEDs
are gloss, but we've settled on keeping them.
Because of the moderately high density of the IO buffering hardware, and
ESD protection networks, as well as a couple of expansion headers, there
isn't room on the main PCB to hold the CPLD. Thus, te CPLD lives on a
smaller daughterboard that will plug into the main PCB.
The whole point of putting the BNCs on the back side of the PCB, as
opposed to having a PCB mounted somewhere in the chassis and wired to
the BNCs, is to eliminate hand wiring. Soldering PCBs is still much
less time consuming than running wires.
But I realize your not proposing that, I'm just indicating some of the
process of evolution I've gone through in considering various ways to
So in this case an alternative might be to ave only connectors and LEDs
on the PCB that fits into the panel. This might be possible, but I
hesitate to do that due to the fact that I have taken great care to put
the IO buffering, particularly the output drivers, very close to the
BNCs so as to avoid long runs of improperly matched transmission line
impedance between the output buffers and the BNCs. Acheiving the clean
edge transistions that I expect out of the existing design, from
something that has a ribbon cable between the drivers and the BNCs,
seems improbable. Thus, I am a bit convinced that regarding the
electronic functinality, the existing design is quite optimum.
It really isn't beyond my means to build these myself. It's just a
question of whether or not outsourcing will save me time. Perhaps it
might simply be beneficial to give it a go (outsource), just to gain
some experience with that arena.
Thanks again for your input.
Christopher R. Carlen
Principal Laser/Optical Technologist
Sandia National Laboratories CA USA