Test jig for testing production PCBs

[Rajinder]

Hi all
I am looking into designing a test jig that will be used to test populated PCBs off a production line.
The PCBs will be microcontroller based with I/O lines connected to LEDs, buzzers, some analogue circuitry too.
I need to be able to test each PCB circuit functionality using a test rig abd indicate if it has passed or failed. Also at what point it failed.

My questions are
1. I think i need a a means of connecting the PCB to my jig. Do you think a pogo pin system where the PCB is placed a good idea.
2. My test routines will be using the same microcontroller. But i am not sure how test each bit of the circuit. For example the LEDs i could use a light sensor to detect if it has been lit or could i use an A/D input?
Similarly how will i test the buzzer circuit.
3. Same as q2 for power supplies within the PCB ie 3V3, 12V etc. Checking cirrect crystal operation.

I presume i will need a few test points around the circuitry i am testing, hence using the pogo pin method of contact.

I am only after a few guidelines on the best way to approach this.

Thanks in advance.

 

[Harald Kapp]

I think i need a a means of connecting the PCB to my jig. Do you think a pogo pin system where the PCB is placed a good idea.

It's the standard method. You should rovide (have provided) test points on the PCCB to access all relevant nets.

But i am not sure how test each bit of the circuit.

You need to generate a so called stimulus, a known input signal to the part of the circuit under test, then check the response of the circuit under test and compare with the expected result (good circuit). You will have to define these test routines yourself and match them to the circuit you're testing.

For example the LEDs i could use a light sensor to detect if it has been lit or could i use an A/D input?

Checking for the LED to actually light up is the ultimate test. As a standard test routine it may suffice to check for the correct voltage drop across the LED when it is stimulated to be on and no voltage drop when off.

Same as q2 for power supplies within the PCB ie 3V3, 12V etc. Checking cirrect crystal operation.

Same as above. Know the expected bahavior, compare the actual behavior with the expectation. Allow for tolerances as per your design requirements.

 

[Rajinder]

Hi
Thanks for your reply.
I have some further questions. Let me take this simple example.
I have an smoke alarm PCB that needs to be tested. It has a PIC microcontroller on it.
Various LEDs, buzzer, smoke chamber etc
I have designed a test bed with pogo test pins that fit into this system and connects to various test points that require monitoring. This test bed also has a PIC microcontroller to run various test software.
Let me take this example further. I want to test an LED is functioning correctly.
1. I presume that i will need to power only the test PCB up? The test PCB will be powered up only when i need to test a certain part of a circuit.

2. The PCB under test will not have any firmware in the PIC. Only the test PCB that runs the code.

3. For the LED example i was thinking to use a photo transistor to determine that it has been lit and then detect a edge transition or similar. Or measure the voltage on he cathode pin. This will be done by applying a pulse. Does that seem logical?

I am a bit confused and hope that you could give me some further help.

Thanks

 

[Sunnysky]

Next time do some research on Design for testability (DFT) at the board level or system level. This must be done at the same time as defined the Design Specs for Functions. But being an R&D and Test Enginerr for many decades, I know this afterthought occurs and when no headers or loop back self test or current sensing or test modes are not built into the design that a 1/8" thick FR test jig is necessary with Pogo pins and vacuum pulldown with wirewrap wired test circuits and sensors. in the 80's such a test fixtures were designed by my staff and then built for 1~2k$ by Test fixture companies with ATE headers. Then we either used one of many ICT or FT testers with Ohm's Law DMM type testing or high level servo loop frequency /phase margin bode tests. But the best designs had 4 bit slice processon dual bus, dual channel redundant I/Os with a dedicated uC for BITE ( built in self test, ) verifying every node on the board. The along came J-Tag etc.

Now you have to do it the hard way. Most defects are solder ( design or process defects) or reversed orientation parts but component functions need logic "test Vectors" to perform all the digital tests, which can be downloaded in flash prior to operation code.

 

[rfe1]

I have made a few of such test adaptors already, I always used Ingun test adaptors and pogo pins.
For LED's mostly I use LDR's if you need to test for color it will get expensive but there are sensors for this too.
If you have a test point at the LED you could also measure the voltage drop over the LED.
If you want to test a crystal I would generate an output signal with a fixed frequency from your controller and build a highly stable reference oscilator on you test board to compare this signal to. The higher the signal frequency you generate the faster the test is (few kHz should be enough that it don't bother you)

 

[mattcrow]

Here is a complete circuit you could use for detecting your LEDs lighting up. It features a latch and a reset so it will catch your pulse whenever you want (obviously all stated voltages are tailored to the system I'm using).
It's a .dwg file, you can open it in a free to use DraftSight
Drive link

Matt