Quality & assurance: not to be taken lightly

//Quality & assurance: not to be taken lightly

The Problem

Designing hardware is one thing, making sure that every single piece of hardware is properly tested before it leaves the production facility is another. Many hardware manufacturers only test one in x produced devices but we are convinced that when going for 100% customer satisfaction every device needs to be thoroughly tested.

During quality assurance it is imperative that all possible functionalities are tested which requires a complete test platform. In this article we won’t discuss the entire platform but rather show you how we’ve created an efficient hardware fixture to test our modules.

During previous production batches we learned that devices with screw terminals and non-removable connectors (like our Output module) take a very long time to connect to the test platform. The functional test itself only takes around 25 seconds (this includes programming the ICs and testing bidirectional communication, relays, buttons, LEDs etc.). Because of all the screws that need to be fastened (for each relay) the time for each test easily increases to 3 minutes. As you can tell this is basically a waste of time (and money) and needs improving.

Using our new test fixture described below we’ve reduced the total added time from some 3 minutes to only a few seconds.

The Solution

The principle we use is fairly simple: Every Output module leaves the assembly line without its bottom plate. This leaves all necessary hardware connections on the PCB exposed for the fixture. In this case we used test needles (Pogo pins) to connect to the exposed PCB and enable the Output module to be tested while already assembled. This ensures that no additional issues can arise from assembling after the functional test.

First, we purchased several types and lengths of test needles. In this fixture two types of diameter are used: 1.67mm and 2.36mm (the diameters for the pogo pin receptacles).
We also ordered the following drills to suit the test needles: 1.7mm and 2.38mm.

To fix the needles at the correct position, we used 8mm thick acrylic.
The best way to have the holes perfectly align with the pins is to use a laser cutter but since we didn’t have access to one we used a standard column drill.
Positioning of the test needles was done by drilling through a blank Output module PCB fixed to the acrylic so they couldn’t lose their position relative to each other (as seen below).

Here you can see several holes have already been drilled through the PCB and acrylic.

When drilling through acrylic, the drill heats up and can melt the plastic causing the finished bore diameter to be too wide. Great care needs to be taken to prevent this so don’t work too fast and remove plastic residue after each hole.
Also note that the rim of each hole needs to be as clean as possible since the test needle receptacles have a lip that needs to be flush with the acrylic and decides how far the receptacle goes into it.

After all the holes have been drilled the receptacles are mounted in the acrylic. The 0.02mm tolerance between the drill diameter and the diameter of the receptacles is perfect to ensure a tight fit.

With all the test needles in their receptacle you get this:

So far so good, but now we need to align the bottom PCB of the Output module with all these test needles.

For this we used a bottom plate to be placed over the test needles with 2 modifications:

  1. Drill holes where all the test needles need to pass through (no need for a tight fit so we made them large enough
  2. Remove the 4 fixations on the side

As a result the Output module can now be aligned with the test needles and onto the fixture easily!

All the wiring that needs to be soldered to and from the test needles will be discussed in a future blog post.

As you can see the two bolts will touch the PCB and cause short circuits so we taped them over as a quick (but temporary) fix.
The DIN rail is used to keep some distance between the bottom plate and the acrylic but would be better replaced by more acrylic of the correct thickness. If the bolts are tightened too much the bottom will bend.

Some thoughts

  • There were no spares of the Output module v3.2 PCBs so we used an unstuffed v3.1 instead. Unfortunately not all connectors’ placement was identical to the new version so some improvisation was required.
  • Like stated before, don’t overdo the drilling or the acrylic will melt.
  • Make sure to double check each hole before drilling, even missing by a little bit will probably cause you to start over.

Future improvements

  • Currently the DUT (Device Under Test) has to be held down by hand, a holding device needs to be added.
  • Fixation of the bottom plate to the acrylic is not yet perfect due to the DIN rail and 2 bolts.
  • The PCB has the tendency to fall out of the bottom when handling the DUT, we may switch to a bottom up positioning.
  • An additional test needs to be added (i.e. testing the 5VDC after the convertor).
  • LED functionality is currently at the mercy of the test operator, this will need to be automated.
By | 2017-11-17T16:26:33+00:00 September 1st, 2015|Categories: general|Tags: , , , , , |0 Comments

About the Author: