Programming connection

Here's another option:

-Martin

I've been looking into this too. I want to add a 1x PCI-Express edge connector on my board for programming. Programming and JTAG testing will be done by inserting the board into a PCI-Express 1x slot on my test bench. Anybody else doing this?

Petter

--
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail?

I've thought of something similar, but probably with a 'bed of few nails' rather than an edge connector.

If you standardised in a 'footprint' of pads on the bottom of the board you could use a standard test rig with moveable top/edge locating strips.

The problem with the edge connector is that you're left with a protruding connector on your PCB.

Nial

In my application this is not a problem. I've worked on PCI and PCIe designs, but I have never seen the cost of actually cutting the edge connector.

Petter

--
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the
most annoying thing on usenet and in e-mail?

I thought of various edge connector ideas. For some reason I don't like it, although it could be the most sensible.

Here's another thought, take a SOIC 8 test clip, like these:

and trim the plastic such that the pins that would normally contact the IC can go through holes in the PCB. One could lay out a very compact (narrower than 0.15in SOIC) hole pattern that the clip could go into. If you only need four pins you could place the holes near to the PCB edge and have only one side of the clip engage the holes while the other side simply clamps against the outer edge of the PCB (the holes being about 0.1in from the edge).

In fact, in the four contact case I would probably opt for an 8 pin DIP clip simply because of it being a lot more rugged.

No decision yet, just going through various permutations and ideas now.

-Martin

You can look at a closeup of a spring-pin connector on page 6 of the DesignCon 2005 paper

Precision Interconnect appears to be the folks that worked with Agilent to get the "soft touch" Logic Analyzer probe up and running.

Another paper I saw suggests the spring pins are good for "up to 500" contact cycles though spring pins in general can have much better mating life, so please double-check the suggested mating cycles in any of your solutions.

Because you're so concerned about cost, I'll assume you have a very high production. I'd suggest just going straight to a good spring-pin source and making your own probe block. By using the appropriate crown or point style of pin, you can avoid the added expense of gold-plating your board as may be needed for many of the rounded-tip spring pins. The pins are available with replaceable contact points so you can have your pre-assembled contact block with fixed receptacles and just replace the individual spring-loaded points as they wear.

These "pogo pins" are available from many sources but Interconnect Devices -

- is the source that I've seen in ATE for decades. These pins are available through distribution as well so getting them shouldn't be a problem.

It is a shame there isn't a market of simple connector headers like this pre-made, providing a simple ribbon cable connection. I've wanted contacts like these for all those unpopulated programming connectors on production (or near production) boards. Since my own needs are just the occasion board trouble-shoot rather than programming, I've considered putting something very rough together at various times but never managed to make it happen.

If you want some custom stuff done for you and you're willing to go for the expense now for the savings in hassle and cost later, places like

do custom work and have experience with the pogo pins and the mechanics needed for a robust solution.

Good luck,

- John_H

While I am not high production, the extra cost becomes significant for some assemblies. If we get away from FPGA's and enter into the world of small inexpensive microcontrollers in equally small boards, adding a programming connector is expensive in more than one way. Typical connectors used for JTAG programming can dwarf microcontrollers and expand the PCB. There's also the wear-and-tear on the programmer side of the ribbon cable if you are doing a few hundred boards at a time.

I guess I am looking for a solution that we can adopt across new designs and keep life simple. Although the financial cost of connectors might not be all that significant, there's a lot more beyond raw cost that makes them undesirable.

Thanks for the link to the paper describing the "soft touch" test connector. It sounds like a very good solution for test and troubleshooting needs.

-M

Also don't assume that gold plating increases the board cost significantly. We've been using ENIG (electroless nickel/immersion gold) on all of our production boards for some time now. We initially changed to ENIG due to process problems with white tin, but stuck with it when we realised that the cost differential was small. This is not the same sort of plating you would need on an edge connector finger. Connectors need at least 15 microinches of gold plate to be reliable. The ENIG plating is very thin, but enough to increase the board's shelf life if you don't build them up right away. And the "pogo" style pins are non-wiping so you should still get several programming cycles out of the ENIG's minimal plating.

Regards, Gabor