The Museum of HP Calculators

HP Forum Archive 20

[ Return to Index | Top of Index ]

A more useful standard display format
Message #1 Posted by Dieter on 1 Apr 2011, 9:24 a.m.

First time users of classic HP calculators may be slighty puzzled when the display shows 0,00 or 0,0000 as the device is turned on. I even remember the question why "such an expensive calculator" would only give a three-digit result, while a cheap 5-Euro-Casio... ;-)

Both the first HP (-35) and newer models (last 20 years) offer a display format ALL or STD where the whole singificant mantissa is shown. So 1/8 is displayed as 0,125 and 1/3 as 3,33333333333E-1. The WP34s project even uses this format as its default display mode.

However, for everyday use there's a major drawback in this implementation. Even moderately small values like 0,5153... or 0,03543... are displayed in scientific notation, which means that the user's eye has to move all the way across the display to its right end, read the exponent there and translate it back into a meaningful result. As far as I am concerned, I do not want to see "2,04135445874E-2" if a calculated price simply is "2 cents", or "3,35283587165E-1" if the result is 33,5% or even just "roughly one third". It's even worse on the HP-35s with its 14-digit-display hiding the complete exponent. The user has to scroll the display manually before he is able to see whether the result is 0,33 or 0,03 or something even less.

A simple FIX 2 or FIX 4 is not an option here since in this case the major advantages of the ALL format are lost: if the result is an integer, the display shows an integer (3 instead of 3,00 which could also stand for 2,99835 or 3,00208). Generally, useless trailing zeroes are omitted while on the other hand all significant digits are visible.

Okay, what's my point now? I would like to suggest a useful standard display format similar to what I remember from earlier non-HP calculators: Moderately small results should not be displayed in scientific notation, even if this means the last two mantissa digits are not visible. The advantage is a much more intuitive and "natural" display format that allows the user to capture the result faster, easier and without any "exponent translation error". My suggestion is: All numbers with absolute value >= 0,01 or 0,001 should be displayed like 0,012345... resp. 0,0012345... and only values below this limit should be displayed in scientific notation. Like this:

   current display           improved display
  ---------------------------------------------
   3,12354882604E-2           0,03123548826
   6,66666666667E-1           0,66666666667
  -1,41421356237E-3          -0,00141421356  
   3,14159265359E-6           3,14159265359E-6
For obvious reasons this cannot be implemented in current HP calculators. But it can be done on the WP34s. :-)

Dieter

      
Re: A more useful standard display format
Message #2 Posted by Timo on 1 Apr 2011, 9:56 a.m.,
in response to message #1 by Dieter

Hi Dieter,

a very interesting thought! Personally as an engineer, I'd prefer you 0,001 limit, since the "Mili"-Something units (what a terminus for an engineer... :o ) are found more frequently when doeing engineering calculations.

Personally, I almost exclusively work with the FIX4 or FIX6 setting since these modes display the digits I'm usually most interested in. However, I'd really like to see your proposal implemented, at least for beta testing purposes.

            
Re: A more useful standard display format
Message #3 Posted by Martin Pinckney on 1 Apr 2011, 11:01 a.m.,
in response to message #2 by Timo

I almost always use FIX format because the nature of my work seldom requires displaying very large or small numbers, and FIX is easier to read. I change the number of decimal digits displayed depending on the needs of the project at hand. Most often I use FIX 2 or 3; occasionally 4.

            
Re: A more useful standard display format
Message #4 Posted by Garth Wilson on 1 Apr 2011, 2:19 p.m.,
in response to message #2 by Timo

My 41 is rarely in a mode other than ENG 3. The exponent in ENG quickly translates to "giga," "mega," "kilo," units, "milli," "micro," "nano," etc. which we use all the time in engineering. SCI is a small waste of time in figuring it out. On a couple of programs on the 71, I have it replace those exponents with the letters representing the above before printing, like "kHz" (instead of "___E3 Hz"), "A" (instead of "___E-6 Amp"), etc., or just removing "E0". So what I would prefer, if the current situation were a problem at all, would be a mode that puts the exponent into letters, which take less display room than the E numbers with possible minus signs.

As for the size of the decimal point, the 71 uses a square of four dots instead of a single dot. It is very visible without looking like a huge pot hole.

                  
Re: A more useful standard display format
Message #5 Posted by Walter B on 1 Apr 2011, 6:43 p.m.,
in response to message #4 by Garth Wilson

ENG 3 is a very useful mode, I fullheartedly agree. With the mantissa right adjusted, you get all the important information within a small area. It would be nice to get E0 suppressed, but it's only this - nice, not really necessary IMHO.

In most real life calculations (except in finance, of course), 4 significant figures are sufficiently precise - if they are significant at all ;-)

                        
Re: A more useful standard display format
Message #6 Posted by Paul Dale on 2 Apr 2011, 1:22 a.m.,
in response to message #5 by Walter B

Quote:
In most real life calculations (except in finance, of course), 4 significant figures are sufficiently precise - if they are significant at all ;-)

A bit of an over generalisation here I suspect. My day job requires 7+ digits all the time. Look up ECEF coordinates. To get centimeter accuracy, 9 digits are required. Likewise, for a field that is 10km long, 4 digits just aren't close to sufficient.

- Pauli

                              
Re: A more useful standard display format
Message #7 Posted by Walter B on 2 Apr 2011, 3:29 a.m.,
in response to message #6 by Paul Dale

Quote:
A bit of an over generalisation here I suspect.
For sure :-) I know many people inevitably *need* a high computing precision built in - but many, many more believe in and bother with lots of insignificant figures :-( Personally, I've met many of the latter set.
                              
Re: A more useful standard display format
Message #8 Posted by Garth Wilson on 2 Apr 2011, 2:35 p.m.,
in response to message #6 by Paul Dale

He did say most, and I'm sure that's true.

                  
Re: A more useful standard display format
Message #9 Posted by Eric Smith on 1 Apr 2011, 9:08 p.m.,
in response to message #4 by Garth Wilson

I want a FIX mode that over-/under-flows to ENG instead of SCI. It was possible to do that on the HP-41 with synthetic programming, but it should be a standard mode.

                        
Re: A more useful standard display format
Message #10 Posted by Paul Dale on 1 Apr 2011, 9:31 p.m.,
in response to message #9 by Eric Smith

Quote:
I want a FIX mode that over-/under-flows to ENG instead of SCI. It was possible to do that on the HP-41 with synthetic programming, but it should be a standard mode.

The wp34s now has this ability :-)

For FIX and ALL modes when gaining an exponent.

- Pauli

Edit: added the commands effected.

Edited: 2 Apr 2011, 12:26 a.m.

      
Re: A more useful standard display format
Message #11 Posted by Manolo Sobrino on 1 Apr 2011, 10:05 a.m.,
in response to message #1 by Dieter

And that's why some classic "EOS" Casios are great. Fx-5000f did exactly like that, it switched to scientific mode when results were too small, otherwise it remained in standard notation up to the third decimal digit (maybe fourth, mine is not here to check). More recent ones (e.g fx-6300G, fx-5500LA -I'd like to try fx-4000P and fx-4500PA, don't like newer ones too much- switch to scientific just when there's no mantissa to show, pain in the neck...) Throw in the nice ENG button and I'd get one.

            
Re: A more useful standard display format
Message #12 Posted by designnut on 1 Apr 2011, 10:56 a.m.,
in response to message #11 by Manolo Sobrino

I think displays might use a more prominent symbol for the decimal point. It tends to be obscured by the numbers. I suggest a ddouble height symbol in place of a single dot. I get a little lost when we have a long string after the decimal with no thousands delineators. couldn't some symbol be used as a marker to aid this display.

                  
Re: A more useful standard display format
Message #13 Posted by Marcus von Cube, Germany on 1 Apr 2011, 12:12 p.m.,
in response to message #12 by designnut

We Europeans don't have the visibility problems you mention. :)

The format mentioned by Dieter is used by many Casio calculators. It's called NORM 1 and switches to scientific notation for 0.001 but shows 0.01 as is.

                        
Re: A more useful standard display format
Message #14 Posted by Manolo Sobrino on 1 Apr 2011, 12:28 p.m.,
in response to message #13 by Marcus von Cube, Germany

You're absolutely right about Casio, I hadn't even noticed (what a shame), thank you. By selecting NORM mode once you enter into NORM 1 mode: 10^-2(0.01)>|x|, by selecting NORM mode twice you enter into NORM 2 mode: 10^-9(0.000000001)>|x|. Apparently FX-1000F/5000F,6500G,7500G just have a single NORM mode that cancels FIX or SCI modes and works like NORM 1.

Edited: 1 Apr 2011, 3:28 p.m.

                              
Re: A more useful standard display format
Message #15 Posted by Dieter on 2 Apr 2011, 12:13 p.m.,
in response to message #14 by Manolo Sobrino

Of course a HP or even WP calculator has to top this, so we will have to introduce a general NORM n mode, switching to scientific notation not before the value is less than 1E-n. So NORM 3 would display 0,0123 and 0,00123 while smaller values are shown as 1,23E-4 or 1,23E-16. ;-)

Dieter

      
Re: A more useful standard display format
Message #16 Posted by Nigel J Dowrick on 1 Apr 2011, 5:07 p.m.,
in response to message #1 by Dieter

I too find the "fixed decimal" format annoying; I also feel that ALL modes or their equivalent give me too much information. When would I ever want 10 significant figures? (I'm a physics teacher, not an accountant.) Scientific or Engineering mode is good for limiting significant figures, but I don't like to see powers of 10 for numbers that aren't too big or too small.

When I wrote my own calculator program (a RPN calculator for the Casio FX-9860G series) I hard-wired in a limit of 6 significant figures. Numbers are displayed without trailing zeroes after the decimal point; standard form is used if a number is greater than 1e9 or smaller than 1e-4.

So 2.4 is 2.4 (not 2.40000); pi is 3.14159; pi times 1e8 is 314159000; pi times 1e-4 is 0.000314159. (Sixteen significant figures are stored internally.)

I really like this system, but I suppose it's a personal thing. I've never missed the undisplayed significant figures. Does anyone else feel that six is enough? (except when dealing with sums of money greater than 10000, of course! Not often a problem for me.)

Nigel (UK)

            
Another useful display format
Message #17 Posted by Dieter on 2 Apr 2011, 11:54 a.m.,
in response to message #16 by Nigel J Dowrick

Quote:
I too find the "fixed decimal" format annoying; I also feel that ALL modes or their equivalent give me too much information. When would I ever want 10 significant figures?
I think this is a good point. Especially since I was about to propose the following additional display format. ;-)

There is something I always missed on other calculators - a display format showing a certain number of significant digits. Yes, this can be done in SCI mode: SCI n simply shows n+1 significant digits. But, you guessed it, in scientific notation. My preferred format would look like this:

   display format: SIG 6

value display ----------------------- Pi 3,14159 e^Pi 23,1407 10^Pi 1385,46 10^(Pi^2) 19333,7 e^-Pi 0,0432139 10^-Pi 7,21784E-4 or 0,000721784 3^25 8,47289E+11 or 847289000000

Yes, it's roughly the same as SCI 5 RND ALL. In other words: all values are shown the way they were displayed on a (in this case) 6-digit calculator. So finally SIG 12 is the same as ALL. What do you think about this?

Dieter

                  
Re: Another useful display format
Message #18 Posted by Paul Guertin on 2 Apr 2011, 7:33 p.m.,
in response to message #17 by Dieter

Quote:
There is something I always missed on other calculators - a display format showing a certain number of significant digits.

I agree that this would be a valuable improvement.

Paul Guertin

                  
Re: Another useful display format
Message #19 Posted by Ken Shaw on 4 Apr 2011, 2:58 p.m.,
in response to message #17 by Dieter

Agreed. I have the same issues with the difficulty of displaying significant figures in Excel.

On HP calculators, I typically go with FIX 4, which gives a reasonable balance between digits to the left and right of the radix. It's not ideal, though.

      
Re: A more useful standard display format
Message #20 Posted by David Hayden on 2 Apr 2011, 10:35 a.m.,
in response to message #1 by Dieter

For a calculator that can display any character in any position of the display, Dieter's proposal not only makes sense but it can be done with no loss of information when the cutoff is 0.01. For example, .00123 requires exactly the same number of characters to display in either format:

1.23E-3
0.00123
adding more significant digits adds the same number of characters to either display format.


[ Return to Index | Top of Index ]

Go back to the main exhibit hall