I wish I were smarter

We're doing an ISO implementation at work.

It's resulted in managers sparring over email and CCing dozens of people, each citing different sections of the new ISO docs to support what they think is right.

It's just going to be a multimillion dollar metric shaped certificate on the wall in the end.

Yeah, I wish I was (were?), too.

That must be a serious switcher. I recently designed a board with a boost converter where there is no Kelvin connection on the current sense and it is currently geared towards 15A peaks. Ground just goes to the plane. It could probably do more but that's where the inductor would lose its cool.

When I do a design I always start with the module spec on the word processor. Before anything gets put on the schematic. This contains a layout guideline section which is a permanent part of the document. This is Rev X1. When I debug any changes, cuts, jumpers are entered into the prototype work section which is the last section in this document. Once it's all said and done it gets worked into the design, the rev level goes to Rev A and the prototype work section is deleted.

We write the manual before we design a product, and it's the design spec. Sometimes we put details in the manual that are private design notes; they will be deleted from the manual that the customer gets. But we do archive the prelim manuals. We also keep whiteboard photos and notes from reviews and any just plain design notes. Every product gets a folder on one of the server drives, and that gets backed up and left up forever. Terabytes are cheap.

We do assume that the first PCB will be rev A, and that we can ship it. That happens about 95% of the time.

Lately we are doing official first article testing, to make sure the gadget meets all specs over temperature and power supply voltages and such. That generates a report.

Same here because I grew up with a need to keep a design history trail. A requirement in med-tech. But I am surprised how few engineers actually do this. Most dive right into generating schematics, under the assumption to do the documentation "one of these days", as grandpa Kettle would have said. Then it either never really happens or only half-heartedly.

Yes, but if there is only one resistor value change it is not the same assembly rev level as the original design. The bare PCB always gets Rev A here as well but not the assembly, only after debug.

Very good. I just had one case where it all worked fine here but not after shipping to the client. And the box didn't get dented. Weird.

I'm shocked at how many undated, untitled, no author or rev noted, PowerPoint cartoons I get from other organizations. With helpful file names like "Spec". Or incomprehensible context-free spreadsheets. Or two product specs, one from engineering and one from purchasing, same date, same rev ID, different number of pages.

Some people seem to think in PowerPoint and Excel. Well, "think".

Oh. We always keep the assembly and etch rev the same. Otherwise we'd burn up the alphabet fast.

Temperature testing is brutal. We should always do it.

The topper are published (!) datasheets that say "Uncontrolled Document".

I go Rev X1, X2, X3 ... until it's all done, then Rev A.

That, and susceptibility testing. Which can result in very surprising effects.

I bought a $4K Keithley 2401 SMU. When I grabbed a test lead, the output shifted about 1%.

I sent it back to the distributor. They demanded that I justify the return, so I sent them those pictures. If they actually CE tested it, they must have Volkswagen'd it.

The Agilent unit is great.

The Keithley 2100 DVMs are junk too, rebranded Array boxes.

Keithley=Tektronix now. Sad.

is

Did it actually "work fine" when tested? Or had somebody signed off that it did?

My first job at Cambridge Instruments was getting the "voltage constrast" a dd-on package that we bought from Lintech Instruments (also in Cambridge) w orking with our electron microscopes.

It took us three months to sort out all the issues, and we created a 20-poi nt test sheet to run through on the assembled machine to make sure that eve rything worked. We gave a copy to Lintech, and they agreed to run their pac kage through the same tests before they shipped it across town to us.

Thereafter we always got a signed copy of that test sheet with the packages they shipped to us. I don't think that any of the packages actually passed all the tests when we ran it through - the drop-offs were always trivial, but there was always at least one ...

Some people take quality control seriously, but it takes a fairly robust ch aracter in the quality control department (not necessarily the boss) to mak e sure that it continues to be taken seriously.

Yes, it did and it was tortured a bit to make it not behave. But it behanved.

Yours truly did.

Well, I believe I am such a robust character. Before any prototype leaves to Fedex it has been heated to the limits, shock-cooled, smacked onto a tile, et cetera. Oh, and then it was transported to Fedex on the luggage rack of my road bike and along a totally dilapidated abandoned old highway (Lincoln Highway, left to the elements for more than half a century).

Back when I was first designing alternator regulators Ford told me it had to pass the "box car test".

WTF ?:-)

Turns out a crate of regulators is shoved off of the floor of a box car onto the train station platform below... helluva shock ;-) ...Jim Thompson

Once I needed to make sure that prototypes for a hi-rel system were really sturdy. So I took them for a 3h ride on my mountain bike. This is one of the local routes but filmed by another rider:

It brutal out there. Chains last 1500mi if you are lucky and rear tires

The luggage rack of a road bike may constitute additional shock testing - b ut not as severe as being drop off the tray of a truck.

There are all sorts of elastic elements in the suspension of even a road bi ke.

We got an informal lecture once -over beers - on building Telex machines ro bust enough not break when they fell over, but not so robust (too expensive ) to allow them to be shipped without shock-absorbing packing - polystyrene foam gives enough to convert 100G shocks on the case to to sub-10G shock a s the packaged machine, and it's cheap.

But then you exposed it to a customer.

The Lincoln Highway was the first paved road that ran coast-to-coast.

Around Truckee, it is now called Donner Pass Road. I drive a bit of it from the cabin to Sugar Bowl, and it's spectacular.

And just before the Rainbow Bridge overlook, you can park and hike up into the old transcontinental railroad tunnels and snow sheds.

Anybody out here should check it out. It's not far from Reno.

Not really. With 110psi in the stiff Continental Gatorskin tires it's a brutal ride. It isn't about single shocks, that gets tested by dropping things from the top of our garage steps. It is about constant jarring and vibrations. A road bike is a great test stand for that.

Better to build them so they withstand such shocks. Because when the company moves the old cardboard boxes will be long gone and they'll just stash it somewhere in the truck or container.

Tour of Flanders was today

-Lasse

-Lasse

The spots of orange dirt - the remains of the fallen?

Joe Gwinn

Sounds about the same as when ups knocks the box off the back of a truck. (My lesson in why to install heavy power supplies firmly.)

George H.