1

So you went through weeks of schematic design and PCB layout design and Sent files over to fab house.

Now, you got the PCB from FAB house.

What is the first step you do when you get PCBA? What are typical tests tes that you follow for testing?

Assuming we have some sensors, some boost circuits, lipo chargers, antenna , MCU and I2C outputting sensors which have sensitive analog input on the board.

[EDIT] I hope the question is not to vague. intention here is to learn about best practices from people who are working industry for years.

Curious KP
  • 383
  • 1
  • 11
  • 1
    But approx how many test points are in your PCB? I think you should have a test plan with expected results before you fab it. In other words, it is too late to ask now. – tlfong01 Sep 19 '20 at 02:10
  • 1
    Well you want to verify function and performance 100%, and measure things that might indicate something is a bit 'off', such as power supply current and voltage for all rails. Maybe you've designed a test fixture with pogo pins and a rack of test equipment or and/or bespoke circuitry to provide and measure inputs and outputs. Maybe you've built in some self-test routines that need to be run.. lots of things are possible. – Spehro Pefhany Sep 19 '20 at 04:47
  • 1
    Test fixture will be used in later stage of development to check every unit is functioning in same manner. I am wondering more about validation stage of cycle. – Curious KP Sep 19 '20 at 05:01
  • 1
    I would say the primary goal is safety with functionality being a secondary goal. Copies of you board may be used by various other engineers at your company as they develop code or do whatever using copies of your board. What you don’t want it is your board blowing up or frying other equipment. So I agree with Spehro in that you should test everything 100%. Really put the board through its paces to ensure it is safe and does what the specs indicate. – Joe Mac Sep 19 '20 at 05:25
  • 2
    I can’t think of any App Notes or rules of thumb. So I can’t give specific guidance. Really test the PSU and then gradually get each piece of the circuit working in the order that seems appropriate. In design, be sure to add test points to nodes that matter. Extra headers on certain signal lines (e.g. SPI) are helpful for attaching probes. Don’t ignore anything that’s odd to you because it may be an issue that is much harder to solve later. You should be VERY familiar with all the data sheets at this point and the components should match their specs. So if something is off, figure out why. – Joe Mac Sep 19 '20 at 05:32
  • Thanks Joe, this itself is very good answer for thinking process that goes into it. – Curious KP Sep 19 '20 at 16:29

2 Answers2

2

So you went through weeks of schematic design and PCB layout design and Sent files over to fab house.

Now, you got the PCB from FAB house.

Basically you learn from your mistake. That mistake is: don't ship any design out to a PCB manufacturer without simulating the circuit to death and re-fixing the schematic/PCB as you find design/performance/functionality errors.

What is the first step you do when you get PCBA? What are typical tests tes that you follow for testing?

If you'd have simulated the design as advised, you would have already developed a series of tests (as guided by the ways in which the simulation tests developed) and, you would apply those tests right now. That's a professional approach to the situation; you already know the testing based on simulation - the sim would have uncovered the potentially weaker areas and those are what you should focus on now.

But because you didn't you are left scratching your head when in fact this could have been sorted out weeks earlier. Simulation saves between an average of 1 and 2 board iterations per design in my experience. It's a no-brainer - get used to your circuit before committing to copper. There is no economical alternative given that simulators are basically zero-cost these days.

intention here is to learn about best practices from people who are working industry for years.

Well, you asked!!

Andy aka
  • 434,556
  • 28
  • 351
  • 777
2

If you have the board in your hand, it's too late.

Even before or without simulation, you will have prototyped the riskiest parts of the circuit and know how they work and what to test. The less risky parts are less risky because you already know how to test them.

Therefore you already have a test plan that enables you to test that the board matches the prototype / your design intent / the simulation.

As part of the test plan you have placed a test point on every net on the PCB (or at least, every net that doesn't have a pluggable connector if you're trying to save money at the expense of more test time to plug in all the connectors. A good PCB fab house may ask for this as part of their process anyway.

Then, while the board was away being assembled, you set up a test fixture (like a "bed of nails" with Pogo pins connecting to those test points. This gives you access to every net on the PCB to test it. (Alternative : a generally rather expensive "flying probe" tester.)

This allows two levels of testing :

  1. With power off : that every component is correctly fitted, there are no short circuits on the board. Testing some components may be difficult because of other components fitted to the same nets affecting the impedance measurements.
  2. With power on : functional testing, beginning with voltage and current checks, then (for a MCU) clock frequency and amplitude, then the ability to upload test programs e.g. to exercise each pin and peripheral, then the actual app code.

Here's a very simple bed of nails fixture for functional testing and ISP (In System Programming). For such a small PCB I do the first level testing by inspection (except for PSU short circuit which can be done on the bed of nails)

If you have the board in your hand, you'll be soldering little loops of wire to various device pins so you can attach scope probes or DVM leads without slipping and blowing things up. (Yes I have done this too)

  • Hi Brian, thanks for great insight. I am aware of bed of nails and "Power On" testing, but "Power Off" testing is something new that I heard. Can be done by combination of visual inspection and just multimeter probing? Since there is no power, oscilloscope probing probably wont work right? – Curious KP Sep 19 '20 at 17:32
  • Kind of tedious with a multimeter, and generally inaccurate since components in parallel interfere with each other's values. But possible for gross errors like solder blobs between V+ and GND, or mistakes the assembler made on earlier boards (check all the other boards, repeat mistakes are common!) –  Sep 20 '20 at 19:00