Actions
Forum Thread Details
  • Replies 38 replies
  • Subscribers 287 subscribers
  • Views 7292 views
  • Users 0 members are here
Related
See a helpful answer?

Be sure to click 'more' and select 'suggest as answer'!

If you're the thread creator, be sure to click 'more' then 'Verify as Answer'!

How do you start up an untested circuit board?

fmilburn

I've been known to throw a circuit together, apply power to it, and fry something.  Or the circuit just doesn't work and the problem could be anywhere in that mess of wiring.  But I used to work on complex and potentially hazardous projects where we couldn't afford those types of mistakes. There were mechanical, electrical, civil, chemical processing, and instrumentation disciplines, and towards the middle of the design specialists in commissioning and start-up were brought in.

I recently sent out a PCB design for manufacture without prototyping it first.  The schematic is shown below but it is just an example.  My question is more general to starting up any untested circuits.

Schematic

Time for me to get my act together.  I put together a plan with the following approach:

  • Break the circuit into logical sections that can be tested sequentially
  • Describe the test plan for each section with the expected outcome
  • Solder it up a section at a time and test the section before moving on

The actual plan used is shown below.

Commissioning and Startup Plan

One lesson from this was that the test plan should have been made before the PCB was sent out which would have allowed for dedicated test points.  Not shown in the plan (but they should have been) are the bench power over-voltage and over-current settings.

Fortunately, this board went together and started up without a problem but it got me thinking.  What did I leave out?  What tips do you have for planning and starting up an untested circuit?  Do you know of any good resources on the web?

  • A pretty big topic.

    • Usually you want to start by building and testing the power supplies.
    • Then check that all IC power pins are properly connected to the appropriate supplies.
    • Then check that no outputs are shorted to rails or other outputs.
    • Re-check that no connectors have been reversed. 
    • Then complete assembly or assemble by section.
    • Then apply power using a severely current limited power supply and see what gets hot.
    • Once everything is cool, check that signals are doing what they should.
  • in reply to dougw

    "They should be on all power rails, and all signals of interest"

    Well stated :-)  The signals of interest is the part I struggle with.  Making a test plan made me think about it and narrow the locations down - experience would be of great help here. One nice thing about through-hole parts is that it is pretty easy to probe almost anything.  SMD not always so much.  

  • in reply to fmilburn

    Those are the ones I'm talking about as well.  The link you provide errors for me: incorrect link or not available in your country.  It may be worth posting the full URL.

    I always position them in pairs with one connected to ground and the other to the signal of interest.  The spacing I set as sufficient to allow me to use a probe and pigtail inserted into the loops, useful for minimising loops if checking for, e.g. ripple.  The through holes are sized to take those test points but are also just big enough to take the probe tip of my Siglent probes.  I've also found they are big enough to take a bootlace ferrule which is stopped by the ferrule 'flare' and makes a good connection without soldering.  The advantage is that the probe and pigtail insert into the ferrule and are held in place, even when placed vertically.  Ferrules are also dirt cheap.

  • in reply to fmilburn

    I've never been able to get an answer to that concern either!  So far, I've not blown anything.

  • in reply to fmilburn

    I had the same questions on test points, although it isn't clearly stated.  How many, where to put them, and what to measure is the main reason I developed the test plan.  I did some internet searching before making this post and what I found wasn't that enlightening.

    KiCad has a number of footprints in the library.  A couple of them are shown below.

    Test Point Footprints

    I have tested a few PCBs in the RoadTest program with the loops and really like them for their ease of use.  Farnell / Newark carry them.  For example, there is Buy Now  in different colors.

    Here is another link if the one above doesn't work for you: TEST-1(BK) - Multicomp - PCB Test Point, Pin, Black (newark.com)

  • in reply to fmilburn

      What do you think about the point Andrew raises on measuring continuity with a DMM and ICs in place?


  • With the small footprints on ICs, one can often measure correct solder joints on digital IOs or generally all pins that have protection diodes built in. With the DMM set to diode forward voltage mode, putting the positive test lead on the IO pin and the negative test lead on a respective Vcc pin is going to show ~300mV if the IO pin and at least one power pin is soldered correctly. Same can be done with the positive lead on GND and the negative lead on the IO pin. Even correctly soldered BGA balls under the chip can be tested like that. I usually do this as soon as I have soldered an IC to minimize false results measured from more than one components connected to a specific trace (pull-up/downs or an IC pin at the other end of the trace). An open solder joint would show no forward voltage and measuring a second IO pin would rule out the Vcc or GND pin not being soldered.

  • in reply to fmilburn

    Hi Frank, 

    For digital logic 33 ohm is often typical (at the output end of the wire), apparently that's good to prevent reflections bouncing.. often I won't over-think it and will just put in a 100 ohm resistor, and see what happens. It will limit current to 33 mA max with 3.3V logic, so it's unlikely to damage chips if an output is accidentally shorted or two outputs are mistakenly connected together, but will hardly slow things down either.

    For surface mount, 1206 size is pretty good to select for polyfuses, there's a nice selection then at current ratings which are typical for semiconductor circuits. For ferrite beads, there are curves to indicate the expected impedance at different frequencies, but some datasheets only specify one frequency, in either case I just go for one with high impedance at that frequency... 0805 size is handy, and 600 ohm @ 100 MHz is a typical value feasible for them.. they may have 10 ohm or so at 10 MHz from memory. I used this one: HZ0805E601R-10 but I was running low on them and they are out of stock, so this time ordered PE-0805FB601ST which is similar. They can be treated like a zero-ohm resistor on power lines, but act as a higher value resistor for all the noise on the power line, so they are nicer than a real zero-ohm resistor for power lines, and the same/similar price.