Posts: 8
Threads: 1
Joined: Jan 1970
The HP9G seems like a nice little machine, so I checked into its specs at hp.com. Its "up to 24digit internal calculation accuracy" is definitely way beyond enough for me.
HP's info about the 9G's unit conversion capabilities is
here.
On page 2 of the above document is a table displaying the units used in the HP9G CONV menu.
The text underneath the table says that "some units are exact (e.g. feet, inch, gallon) while others are provided up to 13 digits of accuracy".
Looking at the table, I see that indeed several units (amongst which gallon) are off in their last (13th) digit. However, yard and pound are quite off the mark and only are specified with 8 to 9 digits accuracy.
Now, why can't HP get a yard as 0.9144 m (by definition) built into this calculator, instead offers us 0.9144000009144 m ?
And why can't HP get a pound as 0.45359237 kg (by definition) in it ? The 9G says "1 lb = 0.4535923772575 kg". Using Excel to figure out how many HP9G correct ounces are in an HP9G offthemark pound, it says "16.00000026"... That's a whopping 8 digits of accuracy, or 'back to the early seventies' !?
At hp.com I filled in an inquiry form about these HP9G unit conversion inaccuracies, then got an 'email undeliverable' message... That's why I'm posting here and hope to get an answer from a Hewlett Packard expert.
Well in due time I'll probably get one anyway, because it's a handy and quite powerful calculator that you don't have to lock up at night. Meanwhile, I will continue to use my good old [censored] fx4500P, and keep the correct conversion units in an Excel sheet.
Posts: 348
Threads: 74
Joined: Jun 2013
First you should realize that the 9g is an OEM product produced by someone other than HP and in addition was not designed by anyone at HP.
Second if 24 digits is too much, why is something off 1 digit in the 13th place too far off?
If you're needing to measure/convert things to that degree of precision then the 9g is a bad choice to begin with.
Posts: 8
Threads: 1
Joined: Jan 1970
Please do read my questions, I wasn't fussing about the units that have an incorrect 13th digit, rather the units that only have 8 or 9 correct digits, for instance inch is given correctly as 0.0254 m but the 9G yard is off, they'd better use 36 inches than the number that they provide for a yard.
And yes for most practical purposes my old slide rule is also precise enough. I'm not questioning the practical applicability of the 9G.
The point here is the difference between the accuracy claimed and the accuracy provided. If the exact international agreed on conversion unit has for instance 4 digits, that's it (the rest is all zeroes) why does this machine have these nonsensical 10th through 13th digits ?
Edited: 17 Feb 2006, 12:56 p.m.
Posts: 415
Threads: 19
Joined: Jan 2005
Obviously printing "HP" onto some piece of chinese &%$&%$ didn't improve the quality of the original product :)
That said, Sanyo offered 16 digits in their first pocket calculators, 35 years ago. Though it's tough to offer that on an entryline scifi calculator, it certainly would be possible. The mentioned "slips" in details wouldn't have happened 20/30 years ago at HP. But then this calc didn't cost you a fortune, so one probably has to accept some tradeoffs. I'm sure, for 5k$ of todays money you'd be able to buy a calc that has correct unit definitions ;)
Posts: 2,448
Threads: 90
Joined: Jul 2005
Herman,
I understand your frustration.
It is maddening beyond belief, when something so straightforward and factual should be wrong, and at the same time glossed over and glitzed up.
Posts: 8
Threads: 1
Joined: Jan 1970
Hm, I was starting to wonder if HP didn't replace their last engineer after he retired...
Is there a table somewhere where I can find which machines are really made by HP (if any) and of a good quality, and which models are OEM (Chinese @!#?@!) ?
PS just now received a second 'undeliverable' trying to contact HP...
Edited: 17 Feb 2006, 1:29 p.m.
Posts: 8
Threads: 1
Joined: Jan 1970
Well I'm mostly amused  check that table, and you will see that they just told someone "press 1, convert it and fill in the result in this table".
So now there are two surprising °F to °C and vice versa "conversion units" at the "Temperature" section.
Posts: 348
Threads: 74
Joined: Jun 2013
But at the same time, give HP credit for not hiding it. The information Herman found came from a document on HP's website.
Confusion reigns!
Posts: 8
Threads: 1
Joined: Jan 1970
That's definitely right  Casio for instance doesn't have the units for the fx991 listed anywhere.
Well thank you all for your responses and your moral support, this forum is very active and responsive, unlike HP...
Posts: 1,792
Threads: 62
Joined: Jan 2005
Herman (if you're still monitoring the thread) 
I poked around an HP30S, which probably uses similar firmware and computational methods (which do not utilize binarycoded decimal representations, to my understanding).
The poundkilogram conversions on the 30S seems OK. However, I found the same problems for the yardmeter conversions.
It seems that the root of the problem  strangely enough  is that one "yard" on these calculators is not exactly three feet!
The conversions between inches or feet to meters, centimeters, and millimeters all followed the internationallyaccepted definitions. However, conversions betwen yards and anything else weren't quite accurate.
One more thing: The HP30S does not distinguish between the international foot (0.3048 meters) and the survey foot (0.304800609601 meters). Thus, the conversions between acres (one acre is 43560 square survey feet) and any metric unit of area is incorrect. If the HP9G/9S also has area conversions, that same problem is probably there.
If you want a calculator with an excellent unit library and conversion utility, the HP28C or HP28S can be bought on eBay for well under US$100.
Regards,
 KS
Edited: 19 Feb 2006, 1:36 p.m.
Posts: 8
Threads: 1
Joined: Jan 1970
Yes, I was checking back.
Oh, I'll check my 30S too for the pound/ounces discrepancy of the 9g. The 30S indeed uses binary internal calculations, not BCD.
Someone once had a table with test results for a load of calculators, calculating something like arcsin(arccos(arctan(tan(cos(sin(9))))))= 9?. Results usually looked like 9.00023455 or 8.9987654. But the 30S was the first one to return 9 exactly thanks to its internal accuracy (and rounding off for the final result).
The HP9G has indeed more than 25 squares inches short on a square acre because it uses regular feet for its 'acre'.
And thanks for the tip Karl.
Edited: 20 Feb 2006, 11:48 a.m.
Posts: 3,283
Threads: 104
Joined: Jul 2005
The term 9 sin cos tan tan^{1} cos^{1} sin^{1} is known as the "Calculator forensics". A description can be found here: http://www.rskey.org/~mwsebastian/miscprj/forensics.htm.
Marcus
