Standard Resistor Values: Why not a true geometric series?

I got a call last month from a guy at United Airlines who tests jet engines and APUs. He needed some 1000:1 current transformer burden resistors so he could read APU power output, using one of our VME boards. We have a bunch of Dale 1 ohm, 3 watt, 5% ww resistors in stock, so I measured them, 4-wire, at the standard 3/8 inch lead length. They were all within 0.3%. So they must wind them to the exact resistance somehow.

We delivered the resistors by hand and got the tour. In one of his "shops" he had a 757 with both engines pulled out. He has five full-size engine test cells, with *big* exhaust fans, load cells, the whole bit. And he has several smaller, bedroom-size, cells for testing APUs. Impressive. Aircraft MRO is an enormous and practically invisible business.

John

Reply to
John Larkin
Loading thread data ...

exact

explain

You appear to be suffering from memory loss and possibly other disabilities since I already covered all this in a previous post. The dumb geometric series approach is not useful in the most common scenario of the application engineer requiring an E-series value closest to the result of theoretical calculation, and this only necessary for 1% or better tolerance. That is where the viewpoint of successive geometric means results a simple algorithm for the calculation, which I presented in that same post. Any idiot who can't recall 5% or more is not an electronics engineer. The 920 is an erroneous arithmetic mean and not a geometric mean.

Reply to
Fred Bloggs

On Oct 17, 8:59 pm, The Phantom wrote: ...

Yeah, it frustrates me that my memory isn't quite good enough to pull out this piece of trivia. I do remember reading long ago about it, in a place I trusted. The history behind some of the developments in electronics can be quite interesting.

Cheers, Tom

Reply to
Tom Bruhns

exact

explain

So you say. I asked you to tell me when you covered it, and you don't seem to be able to answer the question. You just repeated your assertion that you did cover it. If you can't give me a link to a google archive of the post where you covered it, I'll assume you're the one suffering from memory loss.

Were you a member of the committee that made these decisions about what values to include in each E series? Can you provide a copy of the minutes?

You don't know that for a fact; you're just guessing, like the rest of us.

Reply to
The Phantom

Grow up, you little illiterate retard.

Reply to
ChairmanOfTheBored

Now heat them up and watch that specified derating curve fall right in.

You should see the units at MCAS Miramar.

Reply to
ChairmanOfTheBored

it.

proportional

You're STILL a goddamned illiterate retard. Hell, you're worse than Rosie O'Retard.

Reply to
ChairmanOfTheBored

Exactly. Not worth their salt.

Reply to
ChairmanOfTheBored

It is because it is the easiest way to design things. Instead of looking at the set of values from a mathematical progression POV, look at them from a design practicality POV.

Reply to
ChairmanOfTheBored

it.

proportional

Oh! Are you back to demonstrate my point again? Hey, great! Did you understand this post or did you need us to dumb it down for you?

Reply to
=?ISO-8859-1?Q?=AA=BA=AA_rrock

The max APU phase current is around 300 amps, so the 1000:1 CT output will be small, so the power dissipation will be low.

For extra credit: how low?

At UA they were also testing the engines from the C17, which they have contracts for. Once you build a facility this big, you've got to keep it busy.

A test cell is roughly 40x40x80 feet. A big cradle holds the engine up, feeds it fuel and control connections, and measures thrust and stuff. One end of the room is the air intake, with huge sound baffles to keep the roar from leaking out. They only do ambient testing. The engine manufacturer's development cells can simulate altitude and temperature, but one of those babies is over a city block in size.

Some people, like GE, just strap the engine underneath a 747 and fly it, rather than trying to simulate temp and altitude on the ground.

Hey, google for a lot of cool stuff.

formatting link

formatting link

John

Reply to
John Larkin

Even without calculation, I can guarantee it to be less than the 5 Watt rating of the device. Considering proper design methodology and practice, one could halve that figure and call that a design target.

So I would *figure* it to be somewhere very close to that. Hell, my educated guesstimates are probably better than your "look at the datasheet formulas and run some numbers" crap. You likely invariably forget the basics. You probably want to run it at 5 Watts. You logic being "it is a 5 Watt part..."

I made the flight simulators for the C17.

The craft has 52 on board computers, and 23 of them are considered mission critical. We built 46 (two sets) of test racks that all had two slide cradles each, and an open/short peg board for each of the 23 computers, including the HUD modules. They were delivered to McD (now Boeing) in Long Beach. The goal was to have two units in place on a rack, and use the peg board to simulate any open or short possible, to test the redundant system kick in capacity of each sub-system.

But it does anyway. The test cells at Miramar get full thrust, with afterburner runs, and one can hear it 15 miles away in Santee, Ca.

I used to work for a major (one of the best) chamber manufacturers Cincinnati Sub Zero. We made a 150 foot long chamber once to test missiles with at 150F below.

Take it up to Alaska, where the simulation work load to achieve such a low temp would be a lot less.

Some of the pulse jet videos were cool.

Reply to
ChairmanOfTheBored

The max power dissipation is about a tenth of a watt. It probably wouldn't be warm to the touch. That's my "run some numbers crap."

John

Reply to
John Larkin

I may have found some "For example, some years ago, the Radio-Television Manufacturers Association found it desirable to standardize the values of resistors. The ASA Preferred Numbers Standard was considered, but judged not to suit the manufacturing conditions and the buying practices of the resistor field at the moment, whereas a special series of numbers suited better. The special series was adopted and, since it was an official RTMA list, it has been utilized by later RTMA committees for other applications than resistors, although adopted originally because of seeming advantages for resistors. Ironically, the original advantages have largely disappeared through changes in resistor manufacturing conditions. But the irregular standard remains..."

This same author wrote a rather l "Choice of series is influenced by the fact that these units are sold with different standard tolerances, namely five, ten and twenty per cent, and there is a desire to have every unit manufactured, regardless of what its value may be, fall into some standard size and tolerance."

When he refers to "manufacturing conditions...at the moment" (prior to

1951, and definitely prior to 1936) he means that the manufacture of carbon composition resistors was not very precise; values of finished product had a large variance. But note that by 1951 he says "...the original advantages have largely disappeared through changes in resistor manufacturing conditions", meaning that by then they had achieved more precision in manufacturing and the values of the product weren't all over the place.

But in the early days, by having standard series values such that their tolerance bands overlapped, they could sell (almost) every resistor they manufactured, because it would fall into the tolerance band of some standard value.

So why did they choose for the E12 series (10,12,15,18,22,27,33,39,47,56,68,82) instead of (10,12,15,18,22,26,32,38,46,56,68,83), the latter deriving exactly from a

12th root of 10 method with proper rounding? Possibly because the first series has an underlap (zone of inaccessibility) between 12 and 15 of .3, and an underlap between 26 and 32 of .2, for a total of .4. The second series has the same underlap between 12 and 15, but an underlap of .1 between 22 and 27, and none between 26 and 32, for a total of .5. A smaller total range of inaccessibility meant fewer resistors were rejected as not being within the tolerance range of one of the standard values.

Given a desired tolerance, t (expressed not as a percent, but as a decimal fraction, such as .05 for 5%), to find the ratio by which a given value must be multiplied to get the next ideal value in a series, such that the tolerance zones just touch, calculate r = (1+t)/(1-t). For example, for a series with 10% tolerance, the multiplier should be 1.10/.90, or

1.222222+. The first few exact values in such a series would be 10, 12.2222, 14.938, etc.

Given a ratio, r, to determine how many members would be in a decade series of values generated by that ratio, compute 1 / LOG10(r). The just described series had a ratio of r = 1.10/.90, or 1.222222+ and 1 / LOG10(r) gives 11.47, somewhat less than 12. So, if an E12 series had been generated in such a way that the tolerance band for the exact values of the resistors was exactly 10%, we would end up with 11.47 (or course, we actually need an integer here) values in the series. The actual series has

12 values, which means there is more overlap of 10% tolerance bands. This is just what the manufacturers of the early 20th century wanted, because it meant they could sell more of the resistors they made. So, I think they started with the mathematically exact values derived from a root of 10 method and then "tweaked" the individual values to get a smaller total inaccessible range.

Another calculation that can be informative is to calculate the "just touching" tolerance bands for a given generating ratio, r. For the actual E12 series with 12 members, the ratio, r, is 10^(1/12), the twelfth root of

10, about 1.21153. To calculate what the "just touching" tolerance band would be for this series (considering exact, not rounded, values for the series), calculate (r-1)/(r+1). In this case we get .0956+, or 9.56%. So the actual E12 series would have "just touching" tolerance bands a little less than 10%, which means more overlap of 10% tolerance bands around those values. Which concurrently means fewer "inaccessible regions", which means being able to sell more of the manufacturer's production of high variance resistors.

The actual E12 series has less total inaccessible range than the "mathematically exact with proper rounding" series would have. So, that could provide the rationale for its selection.

But, the situation is reversed in the E24 series. The actual E24 series has MORE total inaccessible range than the mathematically exact series (assuming I didn't make a mistake in the calculations). So, what was the rationale for the selection of this series? They plainly wanted the 10% values to be grandfathered in, and this possibly constrained their choices for the other, intermediate, values. I haven't investigated this possibility yet.

It's clear from those two old papers that the "desire to sell all their production" had a dominant influence on the choice of values in the E6, E12 and E24 series, which were standardized in the early 20th century.

The 1936 paper I quoted from has a reference to an even earlier 1926 Proceedings of the IRE paper by Hazeltine. I'll look that up when I get a chance.

Reply to
The Phantom

This is the exact reason.

Reply to
ChairmanOfTheBored

ChairmanOfTheBored snipped-for-privacy@crackasmile.org posted to sci.electronics.design:

What a stupid twit. I did not even need a pencil and paper to figure out that it is about 90 mW.

Reply to
JosephKK

John Larkin snipped-for-privacy@highNOTlandTHIStechnologyPART.com posted to sci.electronics.design:

You should have let him stew for a couple of days.

Reply to
JosephKK

Hey! Give the dolt a break, would ya? He still hasn't gotten all the parts in since they were back-ordered. You didn't actually expect him to figure it out with out measuring it, did you? He said he wasn't going to calculate it and guaranteed it to be less than 5 watts, didn't he? At least let him plug it in first and do us all a big favor!

Reply to
=?ISO-8859-1?Q?=AA=BA=AA_rrock

Good. I encourage all my guys to do stuff like this in their heads. It gets pretty easy with practice. It also impresses the hell out of customers.

It's not so much as "calculating" but more like analog computing, or quickie interpolation between nearby, simple points. If you figure, in a case like this, that power dissipation is a few per cent of the resistor's rating, the math doesn't need to be close.

John

Reply to
John Larkin

I've got nine inches of twit to go up in your ass with, little girl.

At least you finally found the brains to capitalize the word "I".

Good going, asswipe.

Reply to
ChairmanOfTheBored

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.