EEVblog #1244 - Mailbag

[EEVblog]

More Mailbag!



SPOILERS:
Mailbag SPAM!
Arizona
If it smells like chicken, you're holding it wrong.
2:57 https://www.tympan.org Open source hardware audio hearing development system.
11:20 Circuit Scribe conductive ink learning kit https://www.circuitscribe.com/
24:20 Sony RM-440 video editing unit teardown
30:42 Pulse Sensor https://www.pulsesensor.com
31:36 DIY Mouse Kit https://github.com/biomurph/Mouse
32:26 DIY OpenHak open source Fitness tracker https://www.openhak.com
36:34 Faber Castell TR3 scientific calculator and slide rule
42:22 Thermocouple Raspberry Pi Hat http://www.scientificdatasystems.co.uk
46:35 Casio FX-82 Calculator


Ohms Per Square Sheet Resistance video:

Thermocouple Tutorial Video:

[johnlsenchak]

I  don't think  "Fundamental  Friday's " is boring, I wish  you would  do more of them 

[Fungus]

Quite a good slide rule.

Does the middle, slidey part flip over? It looks like it's upside down.

[ggchab]

I  don't think  "Fundamental  Friday's " is boring, I wish  you would  do more of them 

I also like Fundamental Friday's 

[Ibex]

RE: Tympan

Dave,

Click on the three-bar menu icon (hate 'em! ) at the top of the tympan.org page and there are links to some information pages and their forum. Tympan is an open source hearing aid project. From a quick glance at the forum it looks like people are trying to do some interesting things with it.

[Kleinstein]

Getting 7 input channels for the thermo-couples is  likely because there is one channel used for internal purpose like the cold junction or zero correction.

When testing the conductive ink with sharp probes, there are two contributions: one is the thick film part and the other is the spreading resistance around the small contract. To really check the painted resistance it would likely take a 4 wire measurement.

With the calculator it looks like the one digit that sometimes got to bright probably got a problem (e.g. shorted out common pin). This may also effect the keyboard.  There is a visible solder bridge / tin splatter on the board.
I don't think the calculator is genuine Faber Castell, more like an OEM solution. The LEDs look a little like the HP bubble ones.  Some 20-40 mA from 4 coin cells in series sounds like a fail, unless rechargeable. The slight rule is a backup when the batteries run out  .

[Fungus]

In the west we had slide rule/calculator combos, in the east they had these:

[amspire]

I started Electrical Engineering at NSW University in 1974 and we all had to learn slide rules. Calculators were not allowed in exams as something like the HP35a cost about 1/5 the price of a VW Beetle. Allowing calculators would have been unfair on the majority of students. We had to take in our slide rules and log tables. A lot of exam questions were designed so the answer was something like 2 Amps exactly. If you were getting some weird irrational number, you had obviously done something wrong.

In 1975, good affordable scientific calculators like the HP21c became available, and calculators were allowed in exams. I never saw one student touch a slide rule again. I never saw a lecturer use a slide rule again.

This is the same year Faber Castell released it's slide-rule calculator.

Not good timing.

I still have my Japanese Sun Hemi No.130W slide rule in near pristine condition.

[Fungus]

In 1975, good affordable scientific calculators like the HP21c became available... I never saw one student touch a slide rule again. I never saw a lecturer use a slide rule again.

Yep. Slide rules were the way to calculate for over 300 years but they died in six months or so when cheap pocket calculators became available.

This is the same year Faber Castell released it's slide-rule calculator.

Not good timing.

Faber-Castell were the only company really still developing the slide rule at that point. The Faber-Castell 2/83N is the pinnacle of slide rule design, it's a huge beast of a rule that literally goes up to eleven (lesser rules stop at 10).



The one on the back of the calculator in the video has a similar set of scales, it's obviously from the same era.


Fun fact: You could still buy new "Rechenschieber" on the Faber-Castell web site until a couple of years ago. I guess they just liked them.   

[Kleinstein]

AFAIR the transition from slide rule to pocket calculator was not that fast. For some tasks the slide rules where still use for quite some time, as the first generation calculators did not have the trigonometric functions and older peoples where just used to it. I remember my father sometimes still using the slide rule despite of having a calculator (still needed mains power though).

With high power consumption and small batteries the operating time without mains power would be rather short (e.g. maybe 1 hour, possibly less). It really depends on the calculations and precision needed - low resolution multiply and divide are easy with the slide ruler, others (e.g. adding) are better done with the abacus.

[Fungus]

AFAIR the transition from slide rule to pocket calculator was not that fast.

There was probably still lots of people using them for a while, sure, but as far as sales/manufacturing was concerned it was game over.

[thm_w]

RE: Tympan

Dave,

Click on the three-bar menu icon (hate 'em! ) at the top of the tympan.org page and there are links to some information pages and their forum. Tympan is an open source hearing aid project. From a quick glance at the forum it looks like people are trying to do some interesting things with it.

There are no basic pages like "how to use it" though.
I don't understand the mindset of someone who took the time and money to ship the product, but didn't scribble or handwrite some quick instructions (plug in usb, go to github url, etc.).

[ozcar]

In 1975, good affordable scientific calculators like the HP21c became available... I never saw one student touch a slide rule again. I never saw a lecturer use a slide rule again.

Yep. Slide rules were the way to calculate for over 300 years but they died in six months or so when cheap pocket calculators became available.

This is the same year Faber Castell released it's slide-rule calculator.

Not good timing.

Faber-Castell were the only company really still developing the slide rule at that point. The Faber-Castell 2/83N is the pinnacle of slide rule design, it's a huge beast of a rule that literally goes up to eleven (lesser rules stop at 10).

My last year at university was 1970. As students, our options were:

Slide rule. I still have a mint condition Faber Castell identical to the original one I had, and a few others, including a circular one. Only used to amaze young folk these days.

The then best-seller paperback by Godfrey & Siddons. Not just log and antilog tables, but also things like square roots, reciprocals, and trig functions. I've not touched one of those for well over forty years, but I see lots of old ones for sale - some published as late as 1985.

Fortran code, punched into cards using IBM 029 keypunch machines, or entered via 2741 terminals (more-or-less an IBM golfball typewriter with a computer interface). Not the sort of thing us students had at home, of course. Today, just for giggles, I can run the old IBM System 360 Fortran on my PC (http://www.jaymoseley.com/hercules/starthere.htm).

For one course, we used Facit mechanical calculators. Those could add, subtract, multiply and (sort-of) divide. I also encountered a nifty motor-driven mechanical calculator made by Diehl. I don't know what such machines cost, but they were way beyond my means.

I was out of university before I encountered HP 9100 calculators in the early '70s. By then those, which were the first calculators ever made by HP, had been available for a few years. However those were big heavy desktop machines that cost close to $5000. How much was a new VW beetle back then? You probably could have got two of them for that sort of price.

[Fungus]

The then best-seller paperback by Godfrey & Siddons. Not just log and antilog tables, but also things like square roots, reciprocals, and trig functions. I've not touched one of those for well over forty years, but I see lots of old ones for sale - some published as late as 1985.

I've got a few log tables, too, including one to seven decimal places (eight interpolated) compiled around 1750. I really can't imagine the amount of work involved in calculating them, it's nearly 600 pages that look like this:



Those two pages show sine and tangent of 1 degree, with minutes from 12 to 17 horizontally and seconds going downwards. On the next page is sine and tangent of 1 degree with minutes from 18 to 23... 

It's not just the calculations either, the amount of work needed to typeset/proofread those also defies belief. How many little lead numbers are used in that book?

They go for about $10 if you want one: https://www.amazon.com/Logarithms-Baron-Von-Vega/dp/B00NYBBHMC

For one course, we used Facit mechanical calculators. Those could add, subtract, multiply and (sort-of) divide. I also encountered a nifty motor-driven mechanical calculator made by Diehl. I don't know what such machines cost, but they were way beyond my means.

The one to get is the Curta: https://en.wikipedia.org/wiki/Curta

I was out of university before I encountered HP 9100 calculators in the early '70s. By then those, which were the first calculators ever made by HP, had been available for a few years. However those were big heavy desktop machines that cost close to $5000. How much was a new VW beetle back then? You probably could have got two of them for that sort of price.

Yes, but by 1974 the microprocessor had arrived and a scientific calculator cost under $100 dollars, eg. Casio fx-10. I remember being at school in the late 70s and just about everybody had some sort of Casio or Sharp scientific calculator, even LCD was common. They were under $20 by then, IIRC.

[bsfeechannel]

Slide rule crash course for Dave.

The slide rule is based on the fact that any positive real number can be written as a base raised to some exponent,  y = b^x. John Napier realized that when you multiply these two numbers you add their exponents.

So y₁ · y₂ = b^{(x₁ + x₂)}.

When you slide the rule, you're adding their exponents linearly, but the rule is graduated instead with their logs so that you do not have to rely on a table of logs and exponents.

log_b (y₁ · y₂) = x₁ + x₂

That way you turn a complicated multiplication into simple additions and subtractions of lengths on the rules.

No batteries, no hassle.

Piece of cake.

[rjp]

RE: Tympan

Dave,

Click on the three-bar menu icon (hate 'em! ) at the top of the tympan.org page and there are links to some information pages and their forum. Tympan is an open source hearing aid project. From a quick glance at the forum it looks like people are trying to do some interesting things with it.

Its just a copy and paste of a well established open source audio project called Teensy, they dont appear to have done anything new at all, except put a $300 pricetag  on an  existing board.

https://www.pjrc.com/store/teensy36.html
https://www.pjrc.com/store/teensy3_audio.html

the drag and drop gui and low level audio libs.

https://www.pjrc.com/teensy/td_libs_Audio.html

Combining the 2 boards and adding bluetooth is hard to justify $200 odd extra dollars over the original from the guy  who put in the hard work.

[Johnboy]

The Faber-Castell rules were some of the best. When considering the plastic rules, there were few made with the quality and features of the 2/83N.

Regarding the TR-3, the rule attached to the reverse includes select scale functions the electronic calculator cannot perform. That was the whole idea. Very few slide rules perform simple addition and subtraction of integers. It was assumed you would perform simple addition/subtraction with a pencil in hand; preferably one made by Faber-Castell.  So whatever the electronic calculator couldn't give you, you flipped the thing over and set up the next operation with the rule, then flipped back to the keys to punch in the result, etc.

I was lucky enough to score a 2/83N a couple of years before the link Fungus quoted blinked out of existence. My daily workhorse is actually the smaller pocket model of the same rule, the 62/83N. Like Dave's new acquisition, it has the log-log scales (in Dave's case, all six on the same side of the rule, and all eight in the case of the 2/83N and 62/83N). The real downside is that because the scales are so tiny at the pocket size, the precision suffers a lot. The tradeoff was that you didn't have to carry around a log table, as long as you only needed a few significant digits.

[Fungus]

That way you turn a complicated multiplication into simple additions and subtractions of lengths on the rules.

Now explain the uses of the P scale... 

[Fungus]

The Faber-Castell rules were some of the best. When considering the plastic rules, there were few made with the quality and features of the 2/83N.

Yep, not just the design of the scales, the clarity of the printing, etc.

My daily workhorse is actually the smaller pocket model of the same rule, the 62/83N.

That's my favorite slide rule, too, but not the very last version, the one before that with the metal+screws end pieces. These versions aren't as advanced but the construction means they slide a lot better.

(from my collection...)

The real downside is that because the scales are so tiny at the pocket size, the precision suffers a lot.

I wouldn't say "a lot".

Technically speaking it has twice the error so you go from about 0.1% accuracy (10" rule) to 0.2% accuracy (5" rule).

[ozcar]

Old joke was that if you asked an engineer "what is two times two?" they were likely to give the answer as 3.99. Or perhaps "3.99 but let's call it 4".

[Johnboy]

(from my collection...)
(Attachment Link)

The real downside is that because the scales are so tiny at the pocket size, the precision suffers a lot.

I wouldn't say "a lot".

Technically speaking it has twice the error so you go from about 0.1% accuracy (10" rule) to 0.2% accuracy (5" rule).

I was referring to the difference between looking it up in a log table and finding it on the slide rule, not the difference between the 10" and 5" rules. I should have tucked the word "especially" in there, because there is no comparison between those tables and a slide rule, any slide rule, even a concentric one.

I agree that even a considerable augmentation of scale length adds at most one significant digit of difference for most calculations (although I like to squint and imagine the rule can do more). To show how surprising this can be, I'll derail this thread ever so slightly further to post a couple from my own collection, although they are not in beautiful condition like yours.

[Fungus]

Old joke was that if you asked an engineer "what is two times two?" they were likely to give the answer as 3.99. Or perhaps "3.99 but let's call it 4".

Only for small values of 2.

[Fungus]

I was referring to the difference between looking it up in a log table and finding it on the slide rule, not the difference between the 10" and 5" rules.

Ah, OK.... those 7 digit (8 if you interpolate) log tables are definitely more accurate than a slide rule, if a little slower to use.

I should have tucked the word "especially" in there, because there is no comparison between those tables and a slide rule, any slide rule, even a concentric one.

The 10" rule is about the sweet spot for slide rules in terms of accuracy.

I've got 20" rules but the extra bit of precision simply doesn't feel worth it for all the cumbersome extra sliding. Those things can be 40" long if the slider is all the way out.

I've got a Gilson Atlas which is a 50 foot slide rule wound into a spiral on an 8" disc. It sounds like an awesome idea but it's a pain in the ass to use because it's an awkward two-step process to find out which ring the answer is on. A four digit log table gives the same accuracy in a lot less time. Even five digit log tables are probably faster than the Atlas and will get you an extra decimal digit.

I agree that even a considerable augmentation of scale length adds at most one significant digit of difference for most calculations (although I like to squint and imagine the rule can do more).

Double the length is like adding one binary digit of precision. To get an extra decimal digit you have to make the rule ten times longer.

To show how surprising this can be, I'll derail this thread ever so slightly further to post a couple from my own collection, although they are not in beautiful condition like yours.

You're not doing too badly. You've got a Faber-Castell 62/83 and a 20-inch K&E Log-log Duplex Decitrig - the big hole in my collection. 

(As a Heinlein fan, it hurts double not to have one of those K&Es... )

[bsfeechannel]

Now explain the uses of the P scale... 

Glad you asked.

According to the manual of my 52/82 die pythagoreische Teilung P stellt die Funktion y= √(1 - (0,1 x)²) dar; sie arbeitet mit D(= x) zusammen. Die Teilung ist gegenläufig, daher rot eingefärbt.

Die P-Skala bietet den Vorteil, daß man für große Sinus- und kleine Cosinus-winkel genauere Werte erhält. Während man auf der Skala D z. B. für sin 67° das Ergebnis nur mit 0,92 ablesen kann, findet man auf P, unter cos 67° = sin (90 - 97) den Wert genauer mit 0,9204.

Beispiele:
sin 72,3° = 0,9526; cos 12,3° = 0,9771
sin 81,2° = 0,98821; cos 6,3° = 0,99397

Zu jedem sin-Wert auf der Skala D über dem Winkel auf S findet man den cos-Wert auf der P-Skala und umgekehrt. Man hat immer beide Werte sin α und cos α und kann ohne den Winkel abzulesen gleich vom sin zum cos übergehen.

Beispiel: sin 77° = 0,134 entspricht dem cos-Wert 0,991

Für Zahlen nahe unter 1, 100 usw. findet die P-Skala mit großer Genauigkeit Verwendung.

Beispiel: √(0,925) = √(1-0,075) = √(1-(0,274)²) = 0,9618

Der Radikand wird von der nächsthöheren Zehnerpotenz subtrahiert. Die Differenz wird mit dem Läuferstrich auf der A-Skala eingestellt und die gesuchte Wurzel auf P abgelesen, diese wird noch mit der Wurzel der Zehnerpotenz, von der wir subtrahierten, multipliziert.

TG;DR: It can be used to calculate the sine of large angles or cosine of small angles with better precision. Once you know the sine of an angle on the D scale, you can obtain its cosine on the P scale. It's also useful for calculating the square root of numbers close to 1, 100, and so forth.

[tocsa120ls]

That calculator/slide rule thing goes for more than €250 on eBay  O_o