Test Philosophy

To ensure it *survives* ?

You need to design it right.

No amount of testing will make a poor design *survive*.

Be especially aware of thermal considerations. They will kill electronics faster than anything if the sums are done wrong.

Graham

I assume you are asking about test of prototypes to prove design not test of the finished product to show it was manufactured correctly

Two types of test to consider Functional and environmental

Functional tests require all ranges of inputs and outputs be simulated / checked

Environmental:

Simulate or calculate with worst case ranges of variables e.g. resistors, gains

Test over expected temperature range

Test over primary power range

Test for susceptiabilty to electrical noise. If you don't have fancy noise generators use a relay with contact wired to coil to make a relay buzzer and capacitively connect coil to inputs, power and outputs

Test for immunity to RF by using a hand held VHF radio in transmit mode holding within a foot of the unit under test.

If you have access to an RF generator capacitively couple generator to inputs and sweep generator thru its range.

Key a cell phone to talk while holding cell phone near the unit.

Dan

Hey...

I'm curious what is considered good practice when designing and planning for the test cycle of low volume products.

In particular, I am curious what most designers consider neccesary.

Lets take a test case ...A PCB with a micro graphic display some memory a keypad and a serial port. You intend to make 100 pcbs. What would you presribe for a test to assure this design survives (along with your reputation)

Obviously at 100 piece builds you cant do everything so what would you do?

Anyone have to any planning for this situation?

That ought to be a standard test for IEC whatever !

Aside from EMI, static discharge tests can give some interesting results.

Graham

I found that switching on and off a Weller soldering iron, the old fashioned one, with a transformer in the holder ,will crash an 8051 breadboard with remarkable regularity

martin

"Henry and Marisa" schreef in bericht news:4364cf8d$0$15138$ snipped-for-privacy@news.optusnet.com.au...

for

A very conservative design (hardware *and* software) and no testing at all. Or maybe look for a COTS terminal that fits the job, and avoid reinventing the wheel althogether.

Nope, not enough paperwork involved !

And its free, we can't have that in this " new world economy" can we?

What is really required is a Simple set of tests, (Thatcher-ism) that will be adequate for the majority, rather than Blair-ism tests that mean immmmense amounts of paperwork for next to no extra intrinsic value, except to the EMC test company

Anyway, I hope DNA had a good party, and left his Gfreud to do the washing up

Open Source EMC TESTS.........

martin

I can break anything.