Bed of nails test jig.

[BrendanP]

I've been soldering on a molex 70 pin .5mm pitch connectors to pcbs with the trusty fry pan.

I've got the application of the paste down to a fine art now and dont get many bridges.

Still, I need to check that all is OK before I load the rest of the board, picaxe, V reg, R's caps etc. I have been going around with the MM but its slow going.

I've heard of bed of nail test jigs. I am familiar with gold plated spring probes that I could use as the contact points.

The way I see it Id draw up a pcb that would have probes mounted on it that would correspond with the 28X1 DIP pads , R pads etc when the board to be tested would be lowered onto the bed of nails.

The board to be tested has 2.4mm holes in it for mounting bolts. I figure I could have some rod machined to this size and mount it on the test jig so the board to be tested would slide down the rod on two points so it would correctly align with the jig underneath it.

Then if there were bridges in the molex connector Id have an arrangement where LEDs would light that would indicate which pins on the connector were shorted together.

Am I on the right track?

 

[BeanieBots]

The principle is OK but I've never seen a bed of nails which had a pin spacing of 0.5mm. If you can get the nails to connect at some other point on the board, then that not need be a problem.
The probes as you describe, are sprung loaded. You don't need many before there is quite a substancial load on the PCB. The most common counter action is to hold the board in place with a vacuum over a non-conducting mask which also supports the pins.
Now a days, functional testing is rarely done and quality testing (eg solder bridges) is checked visually by camera.
Only you can make the call regarding early failure detection, functional testing and the time/expense involved vs the volume and value of your boards but unless you produce more than 100/wk or each unit is priced over £1k, then bed of nails testing is probably not worth the effort.

Don't forget "design for test". It might be more cost effective to include a "test pad" or even connector on all your boards specifically for testing. That would allow you to have just one expensive test jig which can test ALL your boards.

 

[BrendanP]

Thanks BB, I was intending to test at the DIP pads for the picaxe and a couple fo other points, as you say the .5mm pin spacing is just too fine for testting there.

I had thought to make a inspection camera set up, I saw an article on using a 50mm old slr lens and bare board CCD camera to do so.

Problem is the molex connector is a gull wing sort of design and its possible to get bridges up inside the connector where they can't be visually detected.

 

[BeanieBots]

How about simply having a test lead which plugs into the molex or are we talking such high volume that a manual plug/connect is not an option?

 

[hippy]

Don't forget "design for test". It might be more cost effective to include a "test pad" or even connector on all your boards specifically for testing.

BenieBots beat me to it. That's how I'd do it. If you take the connections to one or two rows of SIL pads on 0.1" pitch you should be able to do a basic test by just holding a molex pin block in the holes. If it passes then it's okay, if it fails there's either a problem with the board or poor contact. If you take the tracks to edge of board you may be able to plug it straight into an IDE socket, as used for floppy disk connectors.

 

[BrendanP]

Manual would be OK, (if it was such high volume it would be someone else's problem probably in Taiwan!)

Hippys idea of a pads on the side is a good one.

The problem is that the molex connector is good for only 50 connect/disconnect cycles so the male connector side would have a short service life

 

[hippy]

I'd expect a molex to actually last longer than 50 cycles and most connectors are going to have similar issues. The worse case is more false failures; keep a known working board and also make the contact loom socketed to a final loom tail while keeping a pre-built spare and a swap when contact failure is suspected should reveal if that's the case or not.

 

[BeanieBots]

50 insertions for a molex sounds about right.
However, I've done similar with euro connectors (rated for 32 insertions) on test rigs and they've lasted for years with 20+ insertion daily.
The insertion rating is that which will gaurantee the spec'd contact resistance and current capability. If you are close to those limits, then you should be considering a different connector anyway.
One of the few occaisions when I would suggest ignoring the spec and go for it anyway unless you have a very good reason to keep to it.