The Museum of HP Calculators

HP Forum Archive 20

[ Return to Index | Top of Index ]

Which Calculators Should I Keep?
Message #1 Posted by Dave Shaffer (Arizona) on 3 Mar 2011, 11:16 p.m.

I recently visited my parents - my Dad is 92 and my Mom will be 90 tomorrow.

We had a nice visit, but I come away appalled at the amount of "stuff" they have, nearly filling their house. There is scarcely room to sit on their couch, the dining room table is unusable, the floor is covered with books and magazines, and the "guest bedroom" is overwhelmed with my Dad's art equipment and materials. There are even some HP calculators in there, too: at least a 32S and a 97S in pretty good condition, and maybe even an original 35 somewhere. (My Dad is a retired mechanical engineer, which explains the calculators.)

Now that I am home, amidst my own stuff, I am verging on being appalled at how much of it I have, too. Therefore, I am contemplating a de-acquisitioning mode.

This brings up the question of which HP calculators I should keep. Although getting rid of a few would not really help the volume-of-junk problem, they would be among the easiest and most lucrative items to auction off. I could use the money gained thereby to buy another camera lens or two (hmmm, this may not be converging!).

What do I have? Well, the list includes a 35, an 11C, a 41C and a 41CX, a 32S, a 42S, a 48GX, a 33S, and a 35s. The 35, 11C, and 41CX have to stay for sentimental reasons. The 33S sits by the bed, for those last-minute-of-the-day brainstorms, and the 35s is my current everyday calc and lives on my desktop. I don't really have any need for the 32S, the 42S, or the 48GX (and, I can borrow a 32S from my wife if necessary). All are in very good condition, except the 35 and 11C, which work fine, but show signs of lots of use.

What are your thoughts on getting rid of some of these? Will I regret it later? I doubt it. I am retired and unlikely to ever need them for any kind of work. Unlike some here, I don’t feel any particular compulsion to have one of every calc that HP has ever made.

I know there are many here who would like to have some of these, so you may soon get your chance.

      
Re: Which Calculators Should I Keep?
Message #2 Posted by Namir on 3 Mar 2011, 11:21 p.m.,
in response to message #1 by Dave Shaffer (Arizona)

Dave,

My son Joey whom you met will be in Flagstaff next week for a visit. Maybe he should stop by and you can give him the machines that you think you can live without. Hang on to the 41CX because this model is a beauty. The rest ... give it to Joey!

:-)

Namir

PS: Things happen for a reason ... and Joey's visit to Flagstaff has a reason too! <grin>

      
Re: Which Calculators Should I Keep?
Message #3 Posted by Michael de Estrada on 4 Mar 2011, 12:21 a.m.,
in response to message #1 by Dave Shaffer (Arizona)

Well, obviously you are not a collector, just a practical user. I too am a retired engineer, however, I am also a collector of some of the older more vintage HP models as well as many non-HP calcs, so I don't have a desire to de-acquire, except in certain cases of excessive redundancy. I recently sold two of my HP-45 sets, but still have 8, including 7 complete sets, so I can't say that I've really de-acquired that much. I am also a pack rat like your parents, but I don't regard that as any great sin. One of my spare bedrooms is called the DOS Room, because it's home to 6 bulky DOS PC's that have no practical purpose, and I usually fire up periodically to play old DOS based games. I guess if I were in your position I'd get rid of some of the newer models like the HP-33s, which is functionally identical to the HP-35s. Also, and HP-41C is redundant if you have an HP-41CX, which is far more capable. As far as regretting it later, well, I think you'll always be able to re-acquire, buy maybe the cost will be higher.

      
Re: Which Calculators Should I Keep?
Message #4 Posted by John Robinson on 4 Mar 2011, 1:32 a.m.,
in response to message #1 by Dave Shaffer (Arizona)

Dave,

I think you have answered your own question :-) Keep the 35, 11C, 41CX for sentimental/collector reasons, and the 33s/35s for daily use. Ditch the rest I say !!

I had a cull a couple of years ago, I ditched all non-cardreader LED calcs, all 71b stuff I had, and even most of the HP-IL peripherals (a PIL box is a good replacement). I certainly have not regretted any of that !!

What that left me with are several 65/67 units (just for fun), numerous 41c/cv/cx units (for sentimental reasons and daily use, and cool new stuff like PIL box and Clonix) and voyagers 11, 15, 16 and even the 12 (because they are just so darn rugged), although being an engineer, I am not sure why I have the 12 as well. I do still have a 32s and a 42s, but I have no idea why !! And lastly, a 35s, which I keep at work, just in case it gets stolen, I'd rather that get stolen than a 41 or a voyager.

Cheers, John

Edited: 4 Mar 2011, 1:33 a.m.

            
Re: Which Calculators Should I Keep?
Message #5 Posted by Walter B on 4 Mar 2011, 4:33 a.m.,
in response to message #4 by John Robinson

Dave,

Quote:
I think you have answered your own question :-) Keep the 35, 11C, 41CX for sentimental/collector reasons, and the 33s/35s for daily use. Ditch the rest I say !
I concur, though that won't be my personal selection (I'd ditch the 33s and 35s and keep the 32S and 42s - but its your choice). And - as you mentioned yourself - it isn't really a big step towards a clean home if balanced by some additional lenses ;-)
                  
Re: Which Calculators Should I Keep?
Message #6 Posted by Martin Pinckney on 4 Mar 2011, 10:44 a.m.,
in response to message #5 by Walter B

Quote:
I'd ditch the 33s and 35s and keep the 32S and 42s -
Second the motion.
                        
Re: Which Calculators Should I Keep?
Message #7 Posted by Xavier A. (Brazil) on 4 Mar 2011, 7:56 p.m.,
in response to message #6 by Martin Pinckney

I do agree.

      
Re: Which Calculators Should I Keep?
Message #8 Posted by Bram on 4 Mar 2011, 3:52 a.m.,
in response to message #1 by Dave Shaffer (Arizona)

If you haven’t already done so, I’d suggest to let your father tell the stories that accompany these particularly calculators. I guess you’ll get the answer you’re looking for (at the risk of keeping them all, I have to add)

      
Re: Which Calculators Should I Keep?
Message #9 Posted by Maximilian Hohmann on 4 Mar 2011, 4:59 a.m.,
in response to message #1 by Dave Shaffer (Arizona)

Hello!

If ever I am faced with this kind of decision (it will come one day...) I will get rid of everything with an LCD and most non-HP calculators with very few exceptions. In your place, I would only keep the 35.

Regards, max

      
Re: Which Calculators Should I Keep?
Message #10 Posted by bill platt on 4 Mar 2011, 12:47 p.m.,
in response to message #1 by Dave Shaffer (Arizona)

Hi Dave,

I think you answered your own question.

32S, the 42S, or the 48GX . I am starting to think that I was stupid not to ditch my 48 series machines back in the 49G+ days, when the 48gx was worth a lot more...now that the 50G exists, the value has fallen a lot on all except the survey cards (which are worth far more than the calculator).

Conservatively, that is $70, $175, $90. Of course the 42S might fetch a lot more given the right accessories (box etc) description and timing.

Clutter is our natural state of being. Our ancestors figured out long ago that barring circumstances requiring movement (e.g. Toureg)we hunted and gathered....

Best regards,

Bill

      
Re: Which Calculators Should I Keep?
Message #11 Posted by Egan Ford on 4 Mar 2011, 3:39 p.m.,
in response to message #1 by Dave Shaffer (Arizona)

Quote:
Now that I am home, amidst my own stuff, I am verging on being appalled at how much of it I have, too. Therefore, I am contemplating a de-acquisitioning mode.
I share your thinking. I'm only 43, my wife and I (and kid off to college) share a large home with large rooms more optimized for stuff than people; and we have a lot of stuff. Too much stuff. Fortunately, it is mostly organized and it doesn't appear that we have a lot of stuff. But at our young age, we could double our stuff.

We started downsizing last summer. It's a slow, ongoing process, to reduce.

I've started asking myself if I have extracted enough joy out of my calculator collection. And I think I have. I do not program them anymore for fun, or write much about them. It is a lot of work and I have other interests too. And it does take up space. If I lost them all to fire or theft, would I replace them? No, except one, a 15C. My original 15C is still going strong. It would be the one that I could not part with. With some fresh batteries and a few good textbooks I can rebuild civilization after it collapses at the end of 2012. :-)

      
Re: Which Calculators Should I Keep?
Message #12 Posted by John B. Smitherman on 4 Mar 2011, 7:30 p.m.,
in response to message #1 by Dave Shaffer (Arizona)

Dave, interesting post that's got me thinking about my own situation.

If you decide to keep the 33s or 35s then you keep either the 32s or 42s so that you can use the latter to check the results of the former. :-(

Good luck,

John

            
Re: Which Calculators Should I Keep?
Message #13 Posted by Gerson W. Barbosa on 4 Mar 2011, 9:03 p.m.,
in response to message #12 by John B. Smitherman

Quote:
If you decide to keep the 33s or 35s then you keep either the 32s or 42s so that you can use the latter to check the results of the former. :-(

Sad, but true. Some years ago, when checking the HP-12C Platinum with my trigonometric functions program upon the 33s, I found discrepancies in the tangent function near 90 degrees. I thought my program was flawed... until I decided to test it against the good old HP-32SII...

http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/archv016.cgi?read=103065#103065 (messages #8 through #28)

Later Karl Schneider made an analysis of this bug and presented other findings:

http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/archv016.cgi?read=103989

Sadly the HP-35s has inherited the bugs and they are still there after all these years.

Regards,

Gerson.

                  
Re: Which Calculators Should I Keep?
Message #14 Posted by Michael de Estrada on 4 Mar 2011, 10:37 p.m.,
in response to message #13 by Gerson W. Barbosa

I calculated Tan (89.999999999) on both my HP 35s and HP 32SII and got the same result of 57,295,779,513.1 even though they do differ with fewer decimal digits. For example Tan (89.999999):

HP 35s = 57,295,787.7856

HP 32SII = 57,295,779.5131

HP 15C = 57,295,779.51

HP 41CV = 57,295,779.51

HP 50g = 57,295,779.5131

HP 32E = 57,295,779.51

Curious.

Edited: 4 Mar 2011, 10:50 p.m.

                        
89,999999 TAN
Message #15 Posted by Walter B on 5 Mar 2011, 5:50 a.m.,
in response to message #14 by Michael de Estrada

And the HP 42S and 20b return 57.295.779,5131 as well.

Edited: 5 Mar 2011, 5:52 a.m.

                              
Re: 89,999999 TAN
Message #16 Posted by Gerson W. Barbosa on 5 Mar 2011, 9:07 a.m.,
in response to message #15 by Walter B

So does WP-34s :-)

57,295,787.7856                                                 [hp 33s]
57,295,779.5131                                                 [HP-42S]
57,295,779.51308232           (57,295,779.5131 on the display)  [WP-34s]
57,295,779.51308231529914091  (57,295,779.5131 on the display)  [Free42 Decimal 1.4.62]
57,295,779.513082315059033981499673                             [Windows calculator]
57,295,779.5130823150590339814996731291072821506647830506       [WolframAlpha, input line: tan(89.999999 degrees)]
                                    
Re: 89,999999 TAN
Message #17 Posted by Michael de Estrada on 5 Mar 2011, 12:00 p.m.,
in response to message #16 by Gerson W. Barbosa

Also:

HP 27 - Tan(89.999999) = 57,295,779.51

HP 29C - Tan(89.999999) = 57,295,779.51

However:

HP 25 - Tan(89.999999) = 58,363,500.16

HP 21 - Tan(89.999999) = 58,363,500.16

So the Woodstocks used different algorithms in different models.

Then:

HP 67 - Tan(89.999999) = 57,295,779.51

HP 91 - Tan(89.999999) = 57,295,779.51

So fairly early on HP had proper algorithms.

                                          
Re: 89,999999 TAN
Message #18 Posted by Eric Smith on 5 Mar 2011, 3:13 p.m.,
in response to message #17 by Michael de Estrada

HP-21/22/25/25C used algorithms closer to the classic series. HP-27/19C/29C/67/97 used improved algorithms; part of that was carrying intermediate results to 13 digits. See the HP Journal article "The New Accuracy: Making 2^3=8" by Dennis Harms, November 1976.

                                                
Re: 89,999999 TAN
Message #19 Posted by Michael de Estrada on 5 Mar 2011, 5:48 p.m.,
in response to message #18 by Eric Smith

Well, the HP-22 does not have trig functions, but I assume you are referring to other functions such as logarithms. Anyways, the problem here is not so much with the digits used, but with the algorithm used when a function approaches a discontinuity.

                                                      
Re: 89,999999 TAN
Message #20 Posted by Eric Smith on 5 Mar 2011, 6:55 p.m.,
in response to message #19 by Michael de Estrada

I'm pretty sure the problem here IS a result of not enough digits used in the intermediate calculations; that's exactly what you need more of as you approach the discontinuity.

                                          
Re: 89,999999 TAN
Message #21 Posted by Ronald Williams on 9 Mar 2011, 5:52 p.m.,
in response to message #17 by Michael de Estrada

Also a 1973 HP35 tan(89.999999) = 58,363,500.16 tan(89.99999) = 57,3015.3057 tan(89.9999) = 5,708,970.39 tan(89.999) = 57,296.55162 tan(89.99) = 5,729.565869 tan(89.9) = 572.9572655

                                                
Re: 89,999999 TAN
Message #22 Posted by Marcus von Cube, Germany on 10 Mar 2011, 8:59 a.m.,
in response to message #21 by Ronald Williams

Such a table looks better surrounded by [pre ] tags.

Also a 1973 HP35 

tan(89.999999) = 58,363,500.16
tan(89.99999)  = 57,3015.3057
tan(89.9999)   = 5,708,970.39
tan(89.999)    = 57,296.55162  
tan(89.99)     = 5,729.565869 
tan(89.9)      = 572.9572655
                                    
Re: 89,999999 TAN
Message #23 Posted by Marcus von Cube, Germany on 5 Mar 2011, 12:22 p.m.,
in response to message #16 by Gerson W. Barbosa

The best I could get so far is Derive. I can't tell if the digits are all correct but they agree to what Alpha has returned:

If I set the precision to 100 digits and enter tan(89.999999°) I get:

5.729577951308231505903398149967312910728215066478305068762520082189827157958993840731226709890277654·10^7

(Underlining by me. In fact I have to use the "approximate" button, otherwise, just the expression is returned.)

As a comparison, Alpha returns: 57,295,779.5130823150590339814996731291072821506647830506

Quite good are the double precision Sharp BASIC computers. My PC-1280 returns 12 significant digits after the decimal point which are all correct:

57295779.513082315059

Most of my Casios and TIs return 4 to 5 correct digits after the decimal point. The worst I could find is the ClassPad 300 with OS 3: 57295754.96 is way off (even the integer part has only 6 correct digits!)

Edited: 5 Mar 2011, 12:52 p.m.

                                          
Re: 89,999999 TAN
Message #24 Posted by Gerson W. Barbosa on 5 Mar 2011, 1:05 p.m.,
in response to message #23 by Marcus von Cube, Germany

Ask WolframAlpha for more digits twice to get 

5.72957795130823150590339814996731291072821506647830506876252008218982715795899384073122670989027765471197239176239827
5578625656376921104805563273741214034175570805889095630942752778232728701844110860750074115361494293916630014... × 10^7

The underlined digits agree with what you got with Derive. I think I'll stop here lest WolframAlpha crashes :-)

The HP-35 result matches the ones reported by Michael for the HP-21 and HP-25.

                                          
Re: 89,999999 TAN
Message #25 Posted by Paul Berger (Canada) on 5 Mar 2011, 1:26 p.m.,
in response to message #23 by Marcus von Cube, Germany

Worst I got was my TI59 57290501.51 200LX gives 57295779.51308232

                                                
Re: 89,999999 TAN
Message #26 Posted by Maximilian Hohmann on 5 Mar 2011, 1:31 p.m.,
in response to message #25 by Paul Berger (Canada)

Hello!

Quote:
Worst I got was my TI59 57290501.51

Hard to believe that a Ti59 helped me all the way through school, university, doctorate and flying training. And I'm still alive :-)

Honestly, I never understood what this hunt for obscure „bugs“ (who ever needed the tangent of 89,999999 in real life anyway?) was all about. If I wanted to send out a space probe to rendez-vous with a comet 6 or 7 years after launch, I would hardly use a pocket calculator to do the maths, would I? And this is really the only application I can imagine where this kind of precisuion is required.

Regards, max

                                                      
Re: 89,999999 TAN
Message #27 Posted by Paul Berger (Canada) on 5 Mar 2011, 2:38 p.m.,
in response to message #26 by Maximilian Hohmann

I used a Commodore PR100 in school, I still have it 30+ years later but sadly it appears to be deceased, I have managed to bring it back to life a few time but now it resists my efforts.

                                                            
Re: 89,999999 TAN
Message #28 Posted by Thomas Radtke on 6 Mar 2011, 2:28 p.m.,
in response to message #27 by Paul Berger (Canada)

PR100 => 4.88 * 10^7 SR4190R => 57259375,1

                                                      
Re: 89,999999 TAN
Message #29 Posted by Gerson W. Barbosa on 5 Mar 2011, 2:56 p.m.,
in response to message #26 by Maximilian Hohmann

Quote:
Honestly, I never understood what this hunt for obscure „bugs“ (who ever needed the tangent of 89,999999 in real life anyway?) was all about.

Actually, this bug was never hunted for. It simply popped out, not in a lifelike situation, I recognize, but when using the 33s to test a program.

Quote:
If I wanted to send out a space probe to rendez-vous with a comet 6 or 7 years after launch, I would hardly use a pocket calculator to do the maths, would I?

C++ won't do any better, at least the version I've been using:

#include <stdio.h>
#include <stdlib.h>
#include <math.h>    

int main(int argc, char *argv[])
{
  double x;
  x = tan(89.999999*M_PI/180);
  printf(" x =  %.9f\n\n",x);
  system("PAUSE");	
  return 0;
}

returns

 x =  57295779.814456776

Pressione qualquer tecla para continuar. . .

Also, Excel 2007 [=TAN(89.999999*PI()/180)] returns 57295779.8143678. Perhaps because 89.999999 has not an exact representation in binary float point. If greater accuracy is important, I would suggest TurboBCD and these functions, which give 17-digit accuracy (57295779.5130823149 for this example).

Regards,

Gerson.

                                                      
Re: 89,999999 TAN
Message #30 Posted by Eric Smith on 5 Mar 2011, 3:19 p.m.,
in response to message #26 by Maximilian Hohmann

If a transcendental function on a calculator only gives five significant digits correct for some arguments, what confidence will you have that it gives any better results for other arguments?

What this shows is that the TI-59 didn't use a good algorithm for range reduction on trig functions. In particular, they apparently used a poor approximation of pi.

Range reduction is very tricky to do well. There's an entire chapter of Muller's "Elementary Functions: Algorithms and Implementation" devoted to an overview of range reduction algorithms.

There is a strong case to be made that when it is known that the range reduction cannot be done accurately for arguments outside a particular range, the function should return an error instead of a value with few or no significant digits. This is obviously appropriate for sin/cos/tan for arguments of magnitude >10^10 on a ten-digit calculator, because you effectively have no significant digits. It is less obvious where the cutoff should be for tangents of angles near 90 degrees.

                                                      
Re: 89,999999 TAN
Message #31 Posted by Dave Shaffer (Arizona) on 6 Mar 2011, 11:58 p.m.,
in response to message #26 by Maximilian Hohmann

Quote:
And this is really the only application I can imagine where this kind of precisuion is required.

The NASA program I used to help support is now shooting for sub-mm accuracy for whole Earth geodesy. For an Earth radius of around 6370 km, that requires better than parts in 1e10 accuracy and precision. I don't know what computers or compilers they use now, but in the "olden" days (i.e. 15 years ago), it was all quadruple-precision Fortran (on HP minicomputers!). There are oodles of trig calculations involved, so once in a while one might wind up needing a tangent in the vicinity of 89.99999.... (although, of course, it's all done in radians).

PS - thanks to all for the various suggestions on which calcs to use/keep/sell. I am firmer in my intent to unload my 42S and 48GX. I will play with the 32S (it's actually a 32SII) for a while, to see how it compares with the 33S and 35s. I know, that may be blasphemy to imply that the two newer ones are better, but I certainly am more used to them.

                                                            
Re: 89,999999 TAN
Message #32 Posted by Michael de Estrada on 7 Mar 2011, 12:46 a.m.,
in response to message #31 by Dave Shaffer (Arizona)

Once you start using your HP-32SII, you'll get rid of the HP-33s and HP-35s. It's much more solid with much better keys and display. It has basically the same capabilities, except for the 1-line display, which I actually prefer. The number digits are much bigger and bolder, and I really don't need to see the y-stack value all the time when I can just hit x<>y as needed to view two-variable results. And its not buggy to boot.

                                                                  
Re: 89,999999 TAN
Message #33 Posted by Martin Pinckney on 7 Mar 2011, 10:08 a.m.,
in response to message #32 by Michael de Estrada

Quote:
The number digits are much bigger and bolder...
This is why I hang on to my 32sii, 22s, and 14b. Because someday my eyes will be too weak for my 42s, 27s, 17b.
                                                                        
Most of the modern HP scientific calcs are lousy
Message #34 Posted by Michael de Estrada on 7 Mar 2011, 10:59 a.m.,
in response to message #33 by Martin Pinckney

The current line of HP scientific calcs is very unimpressive in comparison with the older now discontinued models. In fact, the only current model that I use with regularity is my HP-50g, which has very good keys and display and is very solid feeling. Also, it is based on the old RPL line of calcs that were designed by HP. Sure, the HP-48gII is also a good calc, but it is obsoleted by the HP-50g, so I'd keep it if I already owned one, but wouldn't buy one today. Otherwise, I would not keep or purchase any of the other models currently offered. In fact, I have removed the batteries from my HP-33s and will stop using my HP-35s when its batteries expire, although I hardly ever use it anyways, preferring my HP-15c and HP-32SII instead. Being an RPNer, I have no interest in the algebraic entry models anyway, which comprise half of the current scientific line.

Edited: 7 Mar 2011, 11:00 a.m.

                                                                        
Re: 89,999999 TAN
Message #35 Posted by Walter B on 7 Mar 2011, 11:11 a.m.,
in response to message #33 by Martin Pinckney

Martin,

Quote:
Quote:
The number digits are much bigger and bolder...
This is why I hang on to my 32sii, 22s, and 14b. Because someday my eyes will be too weak for my 42s, 27s, 17b.
Because someday your fingers will be too vibrant for any of your today's favourites, don't hesitate to put an HP-91 in your stock d;-)
                                                            
Re: 89,999999 TAN
Message #36 Posted by Marcus von Cube, Germany on 7 Mar 2011, 1:48 a.m.,
in response to message #31 by Dave Shaffer (Arizona)

Dave, having read about eBay fees recently and taking into account the fellowship of the 42s here, I'd suggest you offer your surplus HP stuff to the forum members first (in the classified section.)

                                                            
Re: 89,999999 TAN (Sub Millimeter?)
Message #37 Posted by Stuart Sprott on 7 Mar 2011, 5:32 a.m.,
in response to message #31 by Dave Shaffer (Arizona)

Dave,

Are sure about the sub millimeter precision for Earth Geodesy?

I am a Surveyor and I find that hard to believe. With contential drift and earth tides involved it is hard to see the accuracy getting much better than +- 10 cm.

                                                                  
Re: 89,999999 TAN (Sub Millimeter?)
Message #38 Posted by robert rozee on 7 Mar 2011, 6:49 a.m.,
in response to message #37 by Stuart Sprott

when i worked for trimble navigation, our "holy grail" was sub-centimetre accuracy, which is only an order of 10 different. achieving sub-millimetre would certainly throw up an interesting set of new issues - you might end up having to factor in current moon location and a load of continental drift data into the mapping info!

                                                                  
Re: 89,999999 TAN (Sub Millimeter?)
Message #39 Posted by Dave Shaffer (Arizona) on 7 Mar 2011, 1:06 p.m.,
in response to message #37 by Stuart Sprott

Stuart,

Oh Yes, I am serious!

See this . Look at Bill's first bullet.

Station positions are now routinely measured at sub-cm levels in single 24-hours experiments, and continental drift (properly called "plate tectonics") is measured for all of our geodetic sites, at levels better than 0.1mm/year for data series with 10 years or more.

This kind if accuracy requires and takes into account all sorts of things that you don't normally think about:

plate tectonics - actually, measuring this is one of the goals

earth orientation (the earth wobbles badly as it rotates, partially due to excitation from earthquakes),

changes in the rotation rate of the earth (the earth slows down and speeds up at the microsecond level every day; we can sometimes tell ahead of time that an El Nino is coming: the warming Paficic Ocean waters expand, changing the moment of inertia of the earth and therefore, due to conservation of angular momemtum, the rotation rate of the earth - just like a figure skater changes her spin rate when she pulls in her arms) There is also a seasonal rate effect: there is more land mass in the northern hemisphere - when it snows in the northern winter, the snow accumulates, changing the net distibution of water (moment of inertia), and the rotation rate changes systematically,

ocean loading - when the tide comes in, the weight of the water depresses the land for sites "near" (100s of km!) the ocean,

tidal motion - there are solid earth tides (as you note), just like the water tides. These are, I think, on the order of tens of centimeters, twice a day. By measuring them, you get some idea of the bulk response of the earth to gravity (of the Sun and Moon) - the Love coefficients.

There are all sorts of results that you get, too, besides the obvious ones (like tectonic rates): we are measuring the rebound of the earth in Scandinavia after the last Ice Age - the weight of the ice depressed the underlying rock, which is now slowly (! 1 mm per year-ish) springing back to its original shape. We get a fairly good handle (from the earth orientation effects) on the shape and differential rotation of the earth's inner core (it is NOT spinning at the same rate as the outer earth!).

Do a google/yahoo search on "VLBI (or satellite or GPS) geodesy" for more. (VLBI = Very Long Baseline (radio) Interferometry, which is done with hydrogen maser time standards, whose mid-term (a few minutes) time stability is good to parts in 1e14. The relative time of arrival of radio signals at multiple radio telescopes spread out all over the earth is easily determined at the picosecond level. Light (and radio waves) travel 0.3 mm in one picosecond.

                                                                        
Re: 89,999999 TAN (Sub Millimeter?)
Message #40 Posted by Stuart Sprott on 8 Mar 2011, 6:19 a.m.,
in response to message #39 by Dave Shaffer (Arizona)

Dave,

All I can say is that is astounding. Thanks for the info.

                                                            
Re: 89,999999 TAN
Message #41 Posted by Mike Morrow on 7 Mar 2011, 12:06 p.m.,
in response to message #31 by Dave Shaffer (Arizona)

Quote:
...I am firmer in my intent to unload my 42S and 48GX. I will play with the 32S (it's actually a 32SII)...

Keep in mind that the 42S and 48GX in your list of potential disposables have built-in IR printer output and BEEP/TONE functions. I find that a calculator's ability to make noise (audio or printer feed) can be a very useful user input prompt that I really miss when running a program on grossly less capable machines like the HP-11C, 32SII, 33S, or 35S. The 23-year-old HP 42S is almost universally acknowledged as still the best RPN calculator that has ever been made by anyone, anywhere, anytime.

If you had in fact owned a HP 32S, rather than the very common HP 32SII, that would have been valuable to collectors as a somewhat hard-to-find machine.

                                    
Re: 89,999999 TAN
Message #42 Posted by Thomas Okken on 5 Mar 2011, 5:06 p.m.,
in response to message #16 by Gerson W. Barbosa

Hmm, poor accuracy from Free42 there. It looks like it loses precision in COS close to 90°. I'll have to look into that.

                                          
Re: 89,999999 TAN
Message #43 Posted by Gerson W. Barbosa on 5 Mar 2011, 6:05 p.m.,
in response to message #42 by Thomas Okken

It was not my intention to point this out. IMHO, five correct digits beyond what the display can show are more than enough.

Regards,

Gerson.

                                                
Re: 89,999999 TAN
Message #44 Posted by Thomas Okken on 6 Mar 2011, 5:49 p.m.,
in response to message #43 by Gerson W. Barbosa

Gerson, I appreciate that, but it looks like a design oversight in the BCD20 COS routine; the trigs and logs in that code works using Taylor series approximations with fixed maximum numbers of terms, but in the case of COS approaching zero, that maximum number of terms is insufficient. The code should either calculate additional terms, or, preferably, use trigonometric equalities to stay away from those difficult arguments.

- Thomas

                                                      
Re: 89,999999 TAN
Message #45 Posted by Gerson W. Barbosa on 6 Mar 2011, 6:40 p.m.,
in response to message #44 by Thomas Okken

Quote:
The code should either calculate additional terms, or, preferably, use trigonometric equalities to stay away from those difficult arguments.

You're right, the latter might be a better option. Once I had a problem with acos(x). I was advised to use another trigonometric identity, which eventually fixed the problem:

http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/archv016.cgi?read=103573

Regards,

Gerson.

                                          
Re: 89,999999 TAN
Message #46 Posted by Paul Dale on 5 Mar 2011, 6:23 p.m.,
in response to message #42 by Thomas Okken

Quote:
Hmm, poor accuracy from Free42 there. It looks like it loses precision in COS close to 90°. I'll have to look into that.

Just pull in the decNumber based code from wp34s :-)

- Pauli

                                                
Re: 89,999999 TAN
Message #47 Posted by Thomas Okken on 6 Mar 2011, 5:42 p.m.,
in response to message #46 by Paul Dale

Hi Pauli,

Re: decNumber: where can I find that?
I'm not in a rush to replace the BCD20 code (it performs well and seems pretty mature by now; fixing COS would probably be a lot less work than ripping all the floating point code out of Free42 and replacing it) but still, I would like to take a look at it. Are you distributing it under a GPLv2-compatible license?

Thanks,

- Thomas

                                                      
Re: 89,999999 TAN
Message #48 Posted by Paul Dale on 7 Mar 2011, 1:18 a.m.,
in response to message #47 by Thomas Okken

The decNumber library is available from Mike Cowlishaw's page. This only includes basic arithmetic and a few important transcendental functions.

My additional code is on the WP34s Sourceforge page -- in the subversion repository. Look at decn.{c,h} and complex.{c,h} initially.

There are C++ bindings I believe but I've used only the C ones. Also, gcc is starting to support decimal floating point types natively.

- Pauli

                        
Re: Which Calculators Should I Keep?
Message #49 Posted by Martin Pinckney on 5 Mar 2011, 10:27 a.m.,
in response to message #14 by Michael de Estrada

Quote:
Curious.
Perhaps Chinese calculators are infected with Radon.
                              
Re: Which Calculators Should I Keep?
Message #50 Posted by Michael de Estrada on 5 Mar 2011, 11:34 a.m.,
in response to message #49 by Martin Pinckney

Only expensive ones made for HP:

Sharp EL-531W - Tan (89.999999) = 57,295,779.51

      
Re: Which Calculators Should I Keep?
Message #51 Posted by Ren on 8 Mar 2011, 10:16 p.m.,
in response to message #1 by Dave Shaffer (Arizona)

Your dad may have a slide rule or two as well.

Find them a good home...

Ren

dona nobis pacem

(soft spot in his heart [head?] for a device he never had to use)

            
Re: Which Calculators Should I Keep?
Message #52 Posted by Dave Shaffer (Arizona) on 8 Mar 2011, 10:52 p.m.,
in response to message #51 by Ren

I still have a (very nice!) slide rule: K&E deci-log-trig or something like that. When I was an undergraduate (1964-1968), slide rules were all there was!

                  
Re: Which Calculators Should I Keep?
Message #53 Posted by Martin Pinckney on 8 Mar 2011, 11:47 p.m.,
in response to message #52 by Dave Shaffer (Arizona)

Quote:
...slide rules were all there was!
Pencil and paper...
                        
Re: Which Calculators Should I Keep?
Message #54 Posted by Walter B on 9 Mar 2011, 1:46 a.m.,
in response to message #53 by Martin Pinckney

... and tables.

                              
Re: Which Calculators Should I Keep?
Message #55 Posted by Martin Pinckney on 9 Mar 2011, 2:47 p.m.,
in response to message #54 by Walter B

Of course! An integral part of doing any scientific or business calculations that involved almost anything other than unary functions.

                  
Re: Which Calculators Should I Keep?
Message #56 Posted by Palmer O. Hanson, Jr. on 9 Mar 2011, 9:33 p.m.,
in response to message #52 by Dave Shaffer (Arizona)

Quote:
Slide rules were all there was!
As an undergraduate in the late 1940's we used slide rules if two to three digits were enough. We used books of log tables if we wanted five or six digits.

As a graduate student in 1959-1960 I had access to Fridens, and if I could justify it (unlikely for a lowly graduate student) to a RemRand 1103.

When I started working on inertial navigation systems in 1960 we did much of our work with Fridens. For transcendental functions we used tables or series expansions.

In 1968 I learned how to use the Honeywell Computer Network (HCN, a timeshare network) based on a sixteen bit DDP-516 which offered double precision for everything if I wanted it. Input and output was with a teletype. My slide rule and my log tables went into my desk drawer and I rarely used a Friden anymore.

By 1970 or so I had access to a Sigma 5 but rarely used it as I could only used batch processing with IBM card input.

[ Return to Index | Top of Index ]

Go back to the main exhibit hall