test strategies

quote: This involves integrating Parametric ICT with AI-driven analysis. end quote

Are they selling AI ? or 'nails' that word "nails" is metioned many many times in that article

n AI ls

Usually putting a nail / scope against a RF circuit changes the impedance etc at that point. Seems a bit of a babble article.

I was reading today that humans have more brain cels than AI and humans only need milliwats.

So, human (your, mine, people's ) experience counts

FCT testing sounds really broad and expensive, requiring such things as an environmental chamber, a shaker table, a one ton rack full of programmable instrumentation.

Might be easier to incorporate built-in test.

We usually have some BIST, but we need to test products, at the customer connector pins, against calibrated test equipment.

Design proving tests (did the designer correctly implement the requirements) and manufacturing tests (does this thing have no manufacturing defects) are sometimes confused by management. I've seen design proving tests used as production tests when it was not sensible to do so.

You test for where you expect to find faults.

In manufacturing, you want to assure yourself that the product has been built "as designed". That the correct components have been used, configured as intended, assembled reliably, etc.

After delivery, you are interested in other potential faults: has something failed from use/abuse? is some associated device not connected/operating as expected, etc.

A lot depends on the recourse you have available at the time of test. You can always (?) rework a device if it fails "final test". But, once it's out the door, your options are considerably more limited.

It also depends on what you can rely on *beyond* your device's scope. I.e., is there someone "qualified" to perform a test, evaluate its results, take remedial action?

I run a BIST in my devices during manufacture which is intended to assure the manufacturer that they have built devices that meet my goals.

I run a POST in devices as they power on -- to verify that they have enough functionality to perform a more extensive BIST. And, depending on those results, load more exhaustive diagnostics to *qualify* the performance if a repair/replacement might be needed (because dispatching service staff is insanely inefficient).

I'm currently doing a FA test, a first article test, on a new delay generator. The FA proves that the gadget works, hardware and software. We have a formal FA test procedure and report template.

After FA, production only has to assure that it was built correctly and does i/o and is calibrated. That's a little different from the FA test. FA is mostly manual, whereas production testing is fully automated.

First thing I tried today didn't work! I set a trigger divisor to 1 and it came back something else. Software bug. Production test doesn't have to look for software bugs.

But bed-of-nails sounds so last millenium to me. I think that a lot of old-timers write articles for the electronics mags.

Indeed. And humans need to be able to learn from said experience. Many people are incapable of learning from their own mistakes [1] and repeat them over and over. I'm sure everyone here in sci.electronics.design can think of several examples from real life.

Machines never learn from experience, because they don't have experi- ence. They are and always will be /stochastically ignorant text gene- rators/.

I keep saying that, because so many people /believe/ the machines are sentient, and have thoughts. The aren't and they don't.

I don't know about you, but I don't think I'd trust an L.L.M. generated P.C.B. [2] because it doesn't have /experience/, and it doesn't know how to test it. And it will never learn from mistakes.

A human written algorithm to design P.C.Bs. on the other hand sounds like a useful tool; but still, what do you do if the algorithm doesn't know how to make use of a new component?

Best wishes, and happy hand drawing your own circuit boards.

The main reason for using printed circuit boards was stated to be the elimination of wiring errors on the production line. An incidental advantage was slightly greater productivity because components could be soldered-in quicker - but this was offset by the time and extra processes needed to make the boards. The disadvantages were the difficulty of fault diagnosis (the circuit was difficult to follow) and the difficulty of making a reliable repair (the tracks lifted off the board when re-soldered).

As the manufacturing processes improved and wave-soldering became the norm, assembly became much quicker and fewer girls (they were nearly always girls) were needed on the production line at a time of increasing wage costs. Being able to *print* the circuit was mainly driven by consideration of mass production, there was no inherent advantage in asesembling circuits that way until transistors, miniature components and higher frequencies made it inevitable.

Drawing the boards by hand on tracing paper was a way of preparing for production and was actually a very wasteful process compared with hand-wiring. For prototypes and individual pieces of equipment, hand wiring was still used. I was still building 'rats-nest' prototype circuits in the late 1960s and only transferred them to printed circuit when they were finalised.

For a few months the main BBC TV broadcast transmissions from Crystal Palace in London were passed through a full height rack full of hand-wired Veroboard to encode Teletext data. The pcb version followed later. The Veroboard was the plain type just used as a mechanical support for point to point hand wiring sleeved with silicone rubber Hellermann sleeving. Logic was mostly 74S and 74 series TTL. John

Yes, and if it is for a one of project, I have my doubt about peeseebees

Funny, I was repairing the Chinese remote control for my local satellite TV receiver, I started to work intermittent. Replace the batteries, worked OK, next day no longer worked at all, IR Led? Measured some voltages, seems OK, changed for a new IR LED, it then worked and then a bit later no longer worked. Old IR LED was OK..

Put my glasses on for a closer look at the PCB Broken or damaged track? I noticed one chip, on that PCB, a sort of flexible thin PCB with many pressure switches for the buttons on it. That chip's soldering looked like coming lose. Soldered it again (just heated / melted the contacts in fact)

Now OK and works better than ever, like new. So the pressure on that thin PCB when pressing buttons on the remote caused it to bend and after a few years the chip's soldering gave way.

Had not seen that sort of problem before! So, PCB thickness doee AI know about that? ;-)

I seem to remember building one of those Ceefax boards, cannot remember the name of that company. But that was in the seventies.

It is bad thing they dropped Ceefax. Almost all German and Dutch channels still have it, with lots of program data and ads.

AI pcb layout is here and will get better. Nobody lays out non-trivial ICs by hand any more. Or designs buildings on paper with slide rules.

There are lots of pcb layout AI companies now. I don't trust *any* pcb layout, no matter how it's done. I check and tweak my boards.

A new component needs to be added to the schematic and PCB libraries, now or with AI layout. AI can (maybe) read a data sheet and do most of the work, and check things.

We visited a company near here that uses AI to check schematics and layouts, not actually do them. Allspice. That makes a lot of sense to me.

PCB than too.

We used black tape on mylar. It was called "the artwork". I have a few around still. Just checking a typical layout against the schematic would take two people two days. I fell in love once doing that.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

I one of the written exams I did, for electronics technician NERG here one question was 'here is a transistor circuit, draw a PCB for it'. So on paper with a pen in a few minutes... there were many more questions to answer. Was fun, easy, passed,

Had been doing that for years at home with a black pen on copper and then etching.

Dremeling is better.

You get a solid ground plane, matched impedance traces, SMA edge-launches, and places to write.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

As most of my designs, early in my career, were digital, wire-wrapping was the norm.

But, I quickly discovered most technicians were sorely lacking in that skill. I eventually opted to just do my own prototypes as familiarity with the design often made it easier for me to know which connections were to be made without consulting the schematic so I could proceed at a faster and more reliable pace.

Once I started working on my own, I quickly opted to prototyping in foil as it made it *so* much easier to "get to Done". It made economic sense to buy layout tools instead of relying on someone to physically tape a board and then having to manually check the artwork against the schematic. I *rarely* find problems after-the-fact with this process.

Plus, having multiple copies of a prototype (instead of one "hand-made) allowed me to distribute them to other stake holders: manufacturing engineer to prepare for that hand-off, another to the test engineer, packaging guy, documentation, etc. It trimmed a lot off the schedule (and thus cost). And, took pressure off folks wanting/needing "my" prototype!

Yes, some look really nice!

Yes SMA

Dremeling.. I do have one, always makes so much dust on the table :-)

There are some nice fan/filter things on Amazon.

The wirewrap/perfboard thing, parts on one side and connections on the other, always confused me.

also they reason power and ground planes used to be in negative, a lot easier to put a circle everywhere you don't want a connection and invert it all, than it is to tape everything but the circles black

I have a nice fan to blow smoke away when I am soldering. Do not always use it, grew up with soldering and its smoke.. But does not dremeling on copper clad PCBs throw copper everywhere? I do not want that on my tables full of microchip stuff.

Wirewrap, yes that was long ago, done some :-) No, not really a problem. Soldering was faster, not so many pins sticking out. and nice to be able to solder SMDs on the veroboards.