Godamned 0603

And resistors. I like the 0612 for current sensing.

Best regards, Piotr

I think a key point on units is whether you need to convert them or not, and how they are compared to different measurements. If you need to convert things into real lengths, weights, or whatever, then metric is the only sane choice. But often you don't need conversions.

It doesn't matter if a .44 calibre bullet is 0.44 inches wide or long,

0.44 kg in weight, or whatever - it's just a name, and as long as you match up the name used on the gun and the ammo, you're fine.

It doesn't matter what width a 19" rack is - it just matters that everyone follows the same standard size. You don't measure the height of the rack in centimetres - you measure it in "units" because everything that goes in the rack is an integer number of "units" in height. If you want to know if your new 4 unit server will fit in your rack, you count the number of units of space you have left - conversion to millimetres or measuring with an inchtape would be silly.

On the other side, the size of your pcb tracks and footprints, or screw threads, or mechanical drawings, all need to be as accurate as practically possible, and all need to be specified in a precise scale - metric.

Some cultures use additional parts of their hand (or even the rest of their body) in counting.

Base 12 has been used for many things - that's why we have "a dozen" and "a gross". Base 20 has also been popular (hence "a score"). And mixes of bases have been used historically - the Sumerians and Babylonians alternated between 12 and 5, giving 60 per digit pair, from whence 60 seconds in a minute, 60 degrees in a triangle.

But for larger numbers, a consistent base is a lot easier. Either 12 or

16 might have been a better choice than 10, but it's hard to change now!

I'd prefer base 16 - it makes calculating the digits of π easier :-)

That is an error. There is nothing more precise about metric than imperial units. It's just a matter of convenience. For some metric is more convenient because it's what they are used to, but also the advantages of a decimal based system with few conversion factors. For others imperial is what they are used to and need to learn the conversion factors... many conversion factors... many, many conversion factors. But both are equally precise.

I know that, and it is not actually what I said.

If everyone used imperial units consistently (which could work for lengths, though some imperial units are different in different countries), they could be precise.

But they don't - and conversions back and forth will mean inaccuracies creep in and rounding errors can add up.

When accuracy is important, the world uses metric - except for a decreasing proportion of hold-outs in the USA. If your drawings, designs or measurements pass through the hands of imperial unit users in the USA, accuracy is likely to drop.

It is not just important that you use a precise scale - some imperial scales are as precise as metric. (An inch is formally defined as 25.4 mm.) But you need to use a /single/ scale - the same scale everyone else uses. Metric.

Uh, really???

"all need to be specified in a precise scale - metric"

Perhaps I'm starting to forget my English.

The precision of a unit has nothing to do with who is using it.

Ok, now you are on a different topic, unit conversions. But the fact remains that there is nothing about metric that is more precise than imperial units.

BS! The conversion is simple, 2.54 cm to the inch, exactly, or should I say, "precisely"?

Now you are doubling down on your bad bet! You claimed you weren't saying metric is more precise than imperial and now you are saying it is!

BTW, you need to understand precision. It's actually a term that is being misused by you. What exactly do you mean when you say "precision"? Precision has to do with repeatability of a measurement relating to effects of equipment in the real world. This has *nothing* to do with what scale you are using.

Please stop being silly about this. There's nothing inherently more "precise" about metric than imperial. I hope you also realize that every specification has a tolerance. This is clearly shown in drawings when they are dimensioned in both imperial and metric units.

Resistor power dissipation doesn't vary much between part sizes like

0603, 0805, 1206, provided you can heat sink the end caps. Sinking is limited by the pcb pads. If you do big copper pours, thermal crowding limits heat spreading. So the sideways parts are better, specifically for current shunts on big copper pours.

One can also parallel a few resistors and place a couple of strategic pickoff vias.

Base 12 is good because it has lots of dividers. To make learning multiplication tables easier, I would choose the digit values to go from -5 to +6 rather than from 0 to 11. That has lots of other advantages too. (Maybe from -6 to +6 is even better for symmetry, although then there would be numbers that could be written in multiple ways.)

Jeroen Belleman

Those everyday units, inches and tablespoons and cups and miles per hour, are familiar and no trouble at all. We do engineering math in SI units. I do sometimes compute things like degC/W per inch. Here are some measurements of the thermal conductivity of some coaxial cables:

Stainless hardline is much better but it's expensive and hard to get. SS barely conducts heat.

We never use Fahrenheit, except for room temp and cooking. My wife works in degrees F.

#10 wire is 0.1" diameter and 1 milliohm per foot. #20 is 10 mohms/foot. #30, 100 mohms. Easy to remember.

The instrument front panels are 19" wide. Any significant differences would look goofy.

1U is 1.75 inches.

We do PCB layouts mostly in inches. There is no precision lost. Trace widths measured in integer mils are convenient. "Make that one five."

US decimal points are just as good as metric decimal points. We can measure inches as accurately as you can measure centimeters.

Does anybody use centimeters? Seems like an orphan unit.

The real advantage of SI units is avoiding strange conversions, like between watts and horsepower and BTUs. Thermal calcs are a nightmare in imperial units.

Pressure in atm is pretty arbitrary.

On 2022-02-19 17:45, snipped-for-privacy@highlandsniptechnology.com wrote: [...]

We don't think of centimeters as a unit. The unit is the meter. Centi is just a prefix meaning 1/100th. Engineers tend to use mostly prefixes that are powers of 1000.

There are some weird exceptions. You'll see hPa because it happens to be close to 1mbar. You'll see daN because it happens to be near the downward force of a 1kg mass.

Jeroen Belleman

That's easy. It's about how far light travels in a nanosecond. I think that was how it came about, but they didn't measure the speed of light very well early on. Kinda like a meter being 1 million between the equator and the poles of the Earth. Close, but not all that close.

My skis are 180 cm. People tend to use cm for their heights. Better than hands.

Oh, that's a solved problem, though; sometime in the past (1935?) the US made a three-digit conversion from 'inch' to 'meter' that is definitive (it defines the inch in SI units, so Systeme Internationale applies). We here in US have to deal with 'statute mile' versus 'mile' as a result, but...that's not an international problem).

Conversions can be exact, but of course there's no getting around numeric-representation errors; diagonal of a square isn't rational, so we NEVER have "all" of the digits written down. That's not a standards problem, it's just... a problem.

The 'inch' was never international-standard; Denmark had a different inch. That's why it makes little sense to make a local inch the definition of a meter. France did the world a favor when (after a king redefined the 'bushel' measure to increase land rents) declared a new measure for world distribution.

Not just the USA; all NATO countries accept #6-32 screws as 'a' standard size, and BSP (British Standard Pipe) pipe threads are all over the world. Accuracy is available equally to all, and calculators can handle more digits than I've ever needed. Cube roots to ten digits was painful before personal computing, but I could do it with the right glowing-digits office machine. Slide rule, though, wasn't gonna work.

On Saturday, February 19, 2022 at 1:48:27 PM UTC-5, whit3rd wrote:Accuracy is

A friend was working for IBM on a military sonar problem. She had to calculate table values for a Sin lookup or something. So she used a calculator with I'm not sure how many digits. I thought it was an HP, but that has 15 digits which would seem to be enough. The point is they had problems with the functions using this table and discovered the values were not accurate enough! I'm surprised that even 10 digits weren't enough, but I guess there are issues of differences between large numbers that require very high internal accuracy in the calculations.

Which 'mile' do you mean? 'Nautical mile' perhaps?

Land-measure miles were originally done with (I think) the convention that 39.37 inches is exactly one meter. When the change became official that 25.4 mm is one inch, a meter became equal to 39.370079 inches, and that would have changed all land-measure boundaries. So, they made the land-markers correct, in a system with 'statute mile' instead of common 'mile'. The 'statute foot' is now the base for old chains, rods, acres, etc.

units foot surveyfoot * 0.999998 / 1.000002

I have two!