-2

The site description is given as "Q&A for electronics and electrical engineering professionals, students, and enthusiasts" and I'm interested in "QA for electronics". Though I guess QA is not explicitly engineering but it's also not reverse engineering. Maybe something in between?

However, my question is: How is the quality of electronic devices usually be assured or tested? Let's consider something everydayish like a navigation system or a tamagotchi. It's not super-complex like a mobile phone or even a laptop but provides already enough potential issues (I guess)? So, how would one investigate whether hardware and/or software is working correctly? Are there common tests or even investigations that are always performed? I would think environmental circumstances (temperature, humidity,..) are rather easy to check (though there might be strange combinations someone wouldn't bear in mind in the first place) but what about other cases (like operating temperature crashes (only) a certain software process)?

Is the topic maybe even labeled differently than QA?

Ben
  • 615
  • 8
  • 15
  • how is quality assurance not engineering? It definitely is. Your question is just too broad: there's a wealth of standardization, lectures, even study programs dedicated to this. What you need to do depends on what you want to achieve: building a device that's high-quality for usage as say credit card chip needs a different set of tools and skills than getting an operation table robot certified, or a rocket flight controller sold to the US military. – Marcus Müller Jan 05 '22 at 13:34
  • well, ok, thanks, so the topic itself is called QA? I would already be happy to know how to test a manageable device like a Tamagotchi, especially considering an interplay between hardware and software. I guess when I would like to know how to verify military rockets I would be familiar with this topic :) – Ben Jan 05 '22 at 13:39
  • wow, you realize how complex a problem it is to test something as complicated as a tamagotchi? – Marcus Müller Jan 05 '22 at 13:40
  • 2
    Start by writing down the things you want to assure. That's how all engineering happens: writing down your goals, only after *then* start working on them. Identify the things that are important about your product, and those that aren't. Start working on mathematical models for what pays to test, and what not. Then, design procedures to implement the testing. Incorporate feedback from that into your production. Update the mathematical models. … – Marcus Müller Jan 05 '22 at 13:43
  • 2
    Q&A for this site means "questions and answers" and not "quality assurance". – Andy aka Jan 05 '22 at 14:47
  • I know, it's a question about QA, so..? – Ben Jan 05 '22 at 14:55

1 Answers1

1

For measurement instrumentation what is considered a good practice is having a checklist for many tests that are performed during parts sourcing, system assembly and then at the end for final inspection.

This usually involves checking every individual parts source before assembly, running the assembled unit in a Environmental chamber for some time to burn-in sensors and to make sure the unit performs as expected.

Things like folloginw are checked:

  • when, how, from who was every part sourced
  • has it been stored properly
  • is solder specific quality and specific grade
  • is the PCB manufactured without defects
  • is the PCB electronically tested before assembly
  • is soldering process according to standards
  • has the pcb been washed after reflow
  • after powering do checkpoints have specific voltages
  • does the pcb need to be coated by conformal coating
  • does it still have the same parameters after coating than before?
  • are the screws correct material and tightened by specific torque

etc etc etc

all these parameters shall be cross-checked with measurements done during research and development process. Its not complicated, but to keep delivering consistent products the devil is in the detail.

MarcelH
  • 48
  • 6
  • thanks a lot! Can you also advise on how software issues might be related to hardware (issues)? And when one would check hardware components like does the device operate successfully at 90 °C, would one throw the device away afterward in case it works correctly or use it for further tests? – Ben Jan 05 '22 at 13:59
  • 1
    @ben 90°C is a lot and not many parts are made for that kind of working temperature. just beware before committing to this type of project. if you are testing for mass production you usually test a sample, but if you make devices which must have a very high quality you test every single individual device before delivery to customer. IMO as to the software - you usually have one set of firmware(s) for testing and quality assurance and then you have a final firmware for customer. – MarcelH Jan 05 '22 at 14:13
  • 90 °C was an arbitrary number but electronic devices reach that temperature easily, or not? I think sampling would be enough. Would you test hardware and software independent from each other? Or would you, e.g., check how the software is working at 70 °C? Well, I when testing the device it is hardly doable without the software anyway.. – Ben Jan 05 '22 at 14:45
  • 1
    Go to digikey.com and see how many parts are rated for 70C working temperature - not many. no, sw and hw are tested together, you should have additional unit tests for your software and libraries, but you must test on final hardware. you would design a testing firmware with outputs that you electronically measure - like voltages, currents etc and then put the hw into a 70C 90% humid environment for 3 days and see whether it still has correct outputs during. this does not mean it will work 10 minutes later - thats where you need to assess working conditions for the device and its criticality. – MarcelH Jan 05 '22 at 15:21
  • 1
    There are ISO and other norms that define testing for specific environments - like, for example, ATEX, which has hundreds of pages of constraints (and testing procedures) for hardware that is used in potentially explosive atmospheres. This does not include software though. Embedded software testing (including static analysis) is a huge topic. – MarcelH Jan 05 '22 at 15:25