Automatic "manual" testing

[Saimoun]

Hi

That title is very contradictory I know 

I have a product which I am soon'ish making 1000 of. https://www.nextpcb.com/ is the PCBA house, and they already all sorts of automatic testing both on the PCB and on the soldered components. But I thought it would be nice to have a quick (=cheap) "manual" test of the assembled boards, since I intend to send the products directly to the end customers.

All I need is to make a test device that I can easily "plug" in the board to test it, then that device can have a "test" button that will do all I need - that way it's only a few seconds testing per board.

My problem is the "plug" part - the board has two female TRS connectors, two DIN-5 connectors, and a USB type-B connector, so either my test device has 5 cables but that will take forever to plug and unplug, or it could have one big "bar" where all the connectors are together so they can be plugged in at the same time - but that sounds dodgy and the "bar" of male connectors has to be strong enough to endure 1000 plugging+unplugging.

So I was thinking routing PCB traces from the connector pins to a IDC-like connector on the board that can be easily accessed. Then the soldering of the actual connectors will not be tested but it's through hole so the chance of them not being soldered correctly is pretty low I think.

I understand having a panel connector would be the best (I have seen this ) but I have no experience in doing panels and I left that job to the PCBA house, so I am thinking for now testing it per board.

What do you guys think? Am I on the wrong track all together to automate final testing? If not, which connector should I aim for?

Thank you!

Simon

[ajb]

It's perfectly valid to have a dedicated test connector for situations like this.  Whether that's an actual connector component installed on the board, or a set of test points depends on testing needs.  As you say, it won't test the soldering of the IO connectors, but sometimes that's acceptable.  Attempting to combine a bunch of separate connectors into one big monster plug has a lot of mechanical challenges, because the exact location of the mating interfaces of a connector as soldered to the PCB have some tolerance based on the mechanical tolerances of the connector itself as well as the installation and soldering.  If the mating test plug isn't designed to accommodate that (probably by allowing the connectors to float by some amount in two dimensions) then it will be extremely difficult to plug and unplug from the DUT and may actually cause damage. 

For your situation, a simple header with all of the required signals could work, or a set of test pads that you can contact with a test plug made from pogo pins (plus a couple of fixed pins with matching holes in the board for alignment).  If you have a lot of signals that can't be easily aggregated into one connector, a bed of nails test fixture would be a good way to go, but more complicated to build.  In some cases you may even be able to get pogo pins to contact connectors on the board rather than test points, which will allow you to test the connector itself which can be nice.  TRS and DIN connectors wouldn't be practical, but I've done it with micromatch connectors and male headers down to 1.5mm (JST ZH).

[Saimoun]

Thank you for the reply ajb!

There are not too many pins to be tested, in fact I realized with only 8 pins I can do a very decent test!

So yes a simple header would be fine, I was thinking making it an IDC connector to avoid the test operator to plug it wrong - do you think an IDC connector would last 1000 plugs and unplugs? Or should I stick with simple 8-pin header?

Something like that:

[tooki]

Why not make a test jig with pogo pins?

[wizard69]

For a small run like this try to divide and conquer.   I'd use the fast connects (USB) as is and create a bed of nails fixture for the rest of the signals.   With a little thought and some shop equipment this can be done pretty cheap and dirty.

The big problem with test headers or card edge connections (you did consider them right) is the additional routing and space requirements.   Sometimes it is easy to add such and other times you don't have the room.

Quick and dirty bed of nails can be made with pretty cheap mechanical components and you can even us a space PCB for a drilling guide.   The idea here is to place pogo pins on center with existing vias.   If need be add a Few vias.    For a few pins plus a USB connection this can be done with rather basic tools.

[tszaboo]

For a production run of 1000, sometimes it is faster and easier to just do the testing.
I've seen engineers spend days to make a few hour job faster. False economy.
For connectors, you have to ask yourself, what's the reason to test it? Bad soldering? Bent pins? Can we ship the product and exchange it in case broken?
There is also some low cost test equipment that might be good for this sort of jobs.
https://be.eurocircuits.com/shop/offtheshelf/product.aspx?ad=19804&ano=ec-test-mate&an=ec-test-mate&s=elektor

For 1000 board, this would add 2 EUR/board, you have to decide if your budget could allow it.

[Saimoun]

Thank you all for helpful answers

I did not hear about pogo pins before - I just googled it, but it seems a bit insecure... One guy even said you had to press the pins quite hard otherwise it might not make a connection.
Bed of nails seems to be the same, correct?

In any case the board is not very dense and I can easily route the signals I need to the side.
I wanted to have an IDC connector like this:



But I do not have the space for that connector, so instead I went with a normal pin header male, but with one missing pin to make sure the IDC cable from the tester can only be plugged one way (I'll stuff one of the holes on the cable).



And yes for the connectors the idea would be to test if the soldering did not go well - bent pins I did not even think about it 
But yea hopefully neither of these happen very often, if they do the customers will simply have to send the product to me for repairs

[tszaboo]

And yes for the connectors the idea would be to test if the soldering did not go well - bent pins I did not even think about it 

I was investigating failures with a JVC television control panel. The SCART connector had some 0.5% failure rate. So what happened is that the operator removed each and every PCB from the production line, hold it in the air by one hand and popped in the SCART connector by the other hand. And some pins sometimes didn't find the holes...

[tooki]

Thank you all for helpful answers

I did not hear about pogo pins before - I just googled it, but it seems a bit insecure... One guy even said you had to press the pins quite hard otherwise it might not make a connection.
Bed of nails seems to be the same, correct?

Well of course they need the correct amount of pressure. That’s why they’re spring loaded and used in jigs. They’re extremely widely used in electronics production. I suspect that the vast majority of commercially sold electronics are tested using them.

[penfold]

For smallish numbers of pins, there's the Plug of Nails option from https://www.tag-connect.com/product/tc2070-idc

[Saimoun]

Thank you everybody again - nice to know all that stuff

@penfold not sure I get the idea of the "plug of nail"...? The signals still need to be routed, so what is the difference with using a pin header and an IDC cable?

[tooki]

@penfold not sure I get the idea of the "plug of nail"...? The signals still need to be routed, so what is the difference with using a pin header and an IDC cable?

Another header is another part to buy and assemble. So it adds extra cost. (Not a ton, but non-zero.) Using a “plug of nails” (or other compact pogo pin jig) means zero added unit cost, since no parts are added to the board.

Oh yeah, another zero-cost option is to make the test connector a card-edge connector, made right out of the PCB itself. Then use either a PCB-mount or IDC card-edge socket to connect.

[Saimoun]

Ok thanks tooki, makes sense 

And yes I was also thinking using a card edge like Dave does on his video - but I will have to do it on individual boards (and not the panel) and I cannot really place it anywhere without making the board bigger, which will increase the size of the enclosure since it has to fit in there as well

[penfold]

@penfold not sure I get the idea of the "plug of nail"...? The signals still need to be routed, so what is the difference with using a pin header and an IDC cable?

Another header is another part to buy and assemble. So it adds extra cost. (Not a ton, but non-zero.) Using a “plug of nails” (or other compact pogo pin jig) means zero added unit cost, since no parts are added to the board.

Oh yeah, another zero-cost option is to make the test connector a card-edge connector, made right out of the PCB itself. Then use either a PCB-mount or IDC card-edge socket to connect.

Yeah, zero cost, takes up less real estate (no plastic connector body taking up as much space), zero  insertion/extraction force (not zero, but its a progressive spring pressure compared with the IDC), if you were to make a custom version of it (the the tag connect ones are a bit limited in form, I was only really pointing you there for inspiration because I was short on time to write more of a response) then you can design it around whatever pad placement gives you best routing. The advantage of the tag-connect ones is that they've taken a bit of care over designing in some compliance in the alignment pins so to accommodate for misalignments (the alternative I suppose is to design around larger pads and sloppier tolerances).