The fixture was supposed to be for a customer's phone. That detail mattered more than I realized at the time.
P.O. 3210 looked routine. In September 2022, I was handling prototype assembly orders for a small design-services shop. The job was to build a test fixture that would hold a phone board and connect it to a measurement board. Not rocket science. The carrier board and the controller board were supposed to talk through a small socket strip from the Samtec SSM family. I had used those parts before. I figured the order would take about ten minutes.
It didn't.
Before I go further, here's some context: I've been handling component orders and bench-level rework for about eight years. I've personally made, and written down, twelve significant mistakes worth roughly $23,000 in wasted budget. I now keep the team's pre-order checklist. P.O. 3210 is the reason that checklist starts with 'read the drawing.'
The part number looked close enough
The drawing called for a Samtec SSM socket that had appeared in previous versions of this fixture. I didn't open the manufacturer's datasheet. I didn't compare the recommended PCB layout to our board file. I looked at the internal BOM, saw Samtec SSM, and assumed same series meant same footprint. I ordered enough sockets for the whole build.
Maybe I should explain this for anyone who hasn't spent many evenings staring at connector datasheets. Samtec board to board connectors are grouped into families, but each family has many variants. Different stack heights. Different terminal styles. Different footprint options. Samtec SSM was the right family. It was not the right variant.
Where it fell apart
The assembled boards came back. The first stack-up test failed before the phone board was even placed. The Samtec SSM sockets seated on one board, but the mating connector on the other board didn't line up. I tried pushing harder. Never do that. I heard a crack. It turned out to be an alignment pin, not the PCB, but it was enough to make my stomach drop.
The part I ordered was the same series but had a taller body than the original. That pushed the carrier board too high for the pogo pins that had to contact the phone's battery terminals. On paper, the stack height was off by maybe 0.8 mm. In a fixture, that's a mile.
That's when I made a second mistake. I saw the 3.3 V rail acting strange and blamed the little ceramic capacitors near the connector. If one had cracked during my forced fit, it could pull the rail down or cause noise. I reached for my multimeter without thinking carefully about what I was measuring.
How to test a capacitor with a multimeter
If a power rail misbehaves, a capacitor is a reasonable suspect. But testing one while it's still soldered to the board can give you a reading that has nothing to do with the part you're testing. The first rule I learned: isolate the capacitor if you want a trustworthy number.
Here's the short version of the process I use now:
- Power the board down. If the capacitor is large or might be holding a charge, discharge it through a 1k resistor for a few seconds.
- Remove the capacitor or lift one lead before measuring. In-circuit readings can include other components in parallel, which makes a good capacitor look bad.
- Switch the multimeter to capacitance mode. Connect the probes and wait for the reading to settle.
- Compare the reading to the value printed on the capacitor. A stable reading near that value is usually fine. A reading that stays near zero often means the meter doesn't see enough capacitance.
The capacitor I suspected was a 0.1 uF ceramic part. Out of the circuit, it read 0.09 uF. That's a healthy part. I had almost replaced a good capacitor for no reason.
What the rework actually cost
The capacitors were not the problem. Once I ruled them out, I finally read the connector datasheet end to end. There it was: the recommended mating height for the original assembly did not match the body height of the connector I ordered. The fix required correct parts, a second board assembly run, and a few uncomfortable conversations.
The total cost was around $2,860 when I added up the replacement connectors, the extra assembly work, the scrapped boards, and the expedited shipping. It was not the most expensive mistake I have made, but it was the most avoidable.
The lesson I still use
I still use Samtec board to board connectors all the time. They are reliable parts. You just have to treat the datasheet like a contract. Samtec SSM still shows up in our fixtures. The difference is that I now treat the series name as a starting point, not a complete specification. The drawing defines the stack. The drawing defines the footprint. The drawing wins.
- Check the current drawing before ordering, even if the part number feels familiar.
- Verify stack height and footprint together. Changing one connector body can change both.
- When a board has a weird electrical symptom, don't guess. Test capacitors the right way: isolate the part, use capacitance mode on your multimeter, and ignore readings taken through a parallel circuit.
- Never force a board-to-board connector. If it doesn't seat, something is wrong.
P.O. 3210 happened in 2022. The specifics are a little dated, and Samtec's current catalog has more options than the one I was looking at then. The habit isn't dated. What was optional in 2020, checking every part against the drawing, is basically mandatory now. The fundamentals haven't changed; the execution has changed.
Now I read the drawing first. That habit has probably saved us more than P.O. 3210 ever cost.