engineering lore

Sure, but it's purely a financial decision. As you noted earlier, EC wires, done right, aren't unreliable.

Eliminating EC wires does make the easier to build. But whether to throw away built boards is, or should be financial. Though $4000 will buy a lot of wire. A $4000 BOM cost is pretty impressive!

No, that's the selling price!

But the decision has to be made on the cost, not the price.

al

ow

at

so

ct

At the end of the day, what part numbers describe needs to reflect what a b usiness needs to record & differentiate. And what a business needs to speci fy depends on the business. Using a bunch of numbering one doesn't need is counterproductive - as is failing to use enough numbering. So any proposed system can take varying levels of trimming to suit the company.

I also prefer to use primarily brief verbal descriptions of products follow ed by variant numbers rather than just product numbers. What works best all depends on what you do.

NT

OK, but which is on the BOM? Can you post an extract?

It's not clear why you need the two numbers, unless it defines a bin location or something.

But it would get a different unique stock number, or not?

We do the same thing. There is a four digit number for everything purchased from outside. PCBs, ICs, Rs boxes, machined parts, screws.. etc. The number is meaningless with out the data base. (Well higher part numbers are from a project came later in time... generally.)

Our system leaves alot to be desired...

George H.

Yes, it would have a unique stock number and corresponding stockroom bin, so that manufacturing always pulls the correct one.

We might make them in-house, or order them from Digikey or some cable house.

The random number thing is OK, as long as there is a way to, say, view all 0603 resistors in order. If not, it will be a nightmare to design things.

We used to have a 5-digit system, but we converted it to 7-digit telephone numbers, where 132-xxxx is an 0805 resistor, and they are in order. It would be nasty to mix resistor reels with chassis pans in adjacent bins. Also nasty to have to do a lookup to discover where a bin is.

My innovation was to put color dots on the bins with corresponding dots on bags and reels, as a crosscheck that the parts are in the right bin.

I have a program that searches our parts database and finds all resistor combinations with a given ratio range and a given Thevenin range. That saves me mountains of calculations.

Hey, it just occurred to me that it will also find pairs of resistors that when paralleled hit some value. It won't do series pairs.

All such systems do that!

I could imagine a consulting business that sets up part numbering and stockrooms for companies. Lots of startups and growing companies make an unholy mess of that stuff. I quit one job because the material control manager (the owner's wife) refused to let me organize the part descriptions. They were always buying the wrong stuff, and had piles of unfinished kits and product around, waiting for the right parts. They of course swiped parts from some kits to finish others, multiplying the chaos.

Oh my, don't get me started on the search-ability of the database... nor the descriptions that go with a part... disaster! I can only complain so much. I had no part in it's creation. I'm forever going in a changing descriptions... to add things like part numbers! (Seems to me a 2N3904, should contain the string "2N3904".)

To be fair, the major task of the database is so production people can lay their hands on the right part.

That's nice, a visual parity check.

George H.

Hi, John -

I noticed that your link is to Amazon-uk. Is that correct? I found the book, but you seem to have made and extraordinarily low priced purchase. If you have additional details on how to find such a purchase, I would very much appreciate it.

Thanks.

Disregard, please. I found one for $10 and free shipping.

Otherwise it just makes a bunch of extra work for people.

NT

yes there should be a universal part number system where the "part number" for a 2n3904 for example is

TRANS NPN 2N3904 (optional text here)

Computers these days can easily handle long alpha numeric strings that are also human readable. Why do part numbers have to be N digit numbers? m

" for a 2n3904 for example is

e also human readable. Why do part numbers have to be N digit numbers?

maybe it made more sense when computers had only k of RAM & couldn't cope w ith spaces.

One of the issues with such systems is they're often unnecessarily limiting . When perfectly practical I design circuits to work on a range of transist ors, particularly for jellybean jobs. I can see no reason to restrict trs, in many cases, to just one type.

But as I mentioned, it all depends on what you need. A customer approved de sign you can't change versus a little undemanding widget require very diffe rent levels of specification, part numbering etc. As do mil spec manufactur ers versus say novelty goods.

NT

How would the bins in the stock room be ordered? Would 2N3904 be a bin number? Would the adjacent bin be a 2K419 power transformer? Would transistor BCX70J be on a different shelf far away?

Assigned numbers let like parts be located together, in an agreeable order. All transistors together in little bins, all resistors together on reels, all sheet metal parts together in gigantic boxes.

In a former company I devised a concept where part names consisted of substrings that took care of such ordering and were (mostly) mnemonic, like:

R.4k7.1206 C.100n.0805.50V.X7R T.2N7002 M.S.M3*5...... (mechanical, screw, ...) Part like cables, metal panels and other specially made parts got similar prefixes and then the project number where they where first used: PCB.P01234b.RevB (second PCB in project P01234). Sometimes a descriptive text was appended/inserted. (PCB.P01234b.kbd.RevC)

You can play with the ordering of the substrings to get your physical ordering. Parts also had a ID-number. It was assigned consecutive by the database system and was used internally in the database to reference parts, e.g. for BOMs. Part names were changeable, IDs not to keep references intact. The only problem was, that sometimes people created part names with swapped substrings, resulting in duplicates. But this seems to be nearly unavoidable. Often you can spot this when looking at a list sorted by part names.

We use a 7-digit number, 123-4567. There is a master document that has rules for assigning the numbers; we had some heated debates over that!

The software has good search capability, so it's easy to find a part by mfr, part number, or description. I can search for opamp+jfet+2x. or res+0805+.05%

There is an item-per-line text file too, which people can sic Python or whatever onto for special cases. I have a PowerBasic program that finds resistor ratios.

Try this one: