HP Forums

Gone back to Uni to do a Maths major. I went to the bookshop and examined the Casio calculator on the booklist and thought, what a piece of crap!

So I went home and rumaged around for my old HP 32sII calculator. Couldn't find it, but did find my 49g+ which I'd mistakingly bought thinking it would be better than my 32sII.

Ffter a week was still struggling to use the bloody thing, and remembered why I hated it all those years ago. Big, FUGLY and stupid.

Back on the internet, time to just get a calculator for "calculating". I found this forum, and started reading and reading.. apparently the new calculators are crap.. the old ones are better, so then..

32sII was crap, the 32s was better
No!! the 42s is the best ever made, I mock your 32's
42s? What about the I/O please, give me my 41
Are you kidding me? It all stopped in 1979 with the 15C

After spending a week on ebay searching and loosing on some very pricey auctions for 2nd hand caculators, I thought, stuff it, I'll buy the 35S. At least it will let HP know I'm interested while I wait for a NOS 42s unopened in box to turn up for $50.

Well, my 35s arrived. Its fantastic! I don't have 10 years under my belt with a particular calculator, so some small change is fine with me.

I find the 35s very easy to use. I like the double stack, its small, nice buttons, easier to find things. Way more functions on keys and not hidden in menus (like the 49g+).

To top it off, it looks like its from the set of Battle Star Galactica, ORIGINAL VERSION.

Whats not to like?

So, if your not a die hard 35 year HP user like some people hear, don't worry, you'll love it and in 35 years you'll probably be complaining about the discontinuation of the 35s.

Daniel.

Quote:

---

Whats not to like?

---

This.

Hi Daniel. It misses keystrokes for me. I don't believe it's anything mechanical going wrong but rather it's a software problem. By pushing a variety of keys extremely fast I can see the "Busy" indicator displayed and when it appears the keystrokes don't register reliably. If I slow down it seems to perform better.

Good luck with your math major.

Regards,

John

I too like my 35s which I bought a couple of years ago as a replacement for the dreadful 33s I had endured for about 5 years. It is definitely much better in some respects, most noticeably the key feel, larger Enter key and slightly better display. I have two 35s', one which I use at work and one at home and for most things it has reasonably utility. There are however some quite major annoyances in the method of operation not to mention bugs. Like you, I searched for a reasonably priced older model to try and after a long wait eventually found one at a reasonable price in the shape of an early 11C which, while lacking some of the features and speed of the 35s, is just much more pleasant and logical to use and program. It also feels much more durable. The 11C is now my regular calculator at home (I wouldn't risk it at work as I have lost a few items there unfortunately) and I am enjoying writing simple but useful programs for it. So much so that I also now like the idea of a 41C eventually for the card expansion and storage. I expect a good one will come along at some point at the right price - like most collectables, it is just a case of waiting for the right one. So while I would concur with you that the 35s is usable as a day-to-day calculator, I can see the point of seeking out the older models, not just as collection pieces but to use as working calculators. They come from a time when HP cared about their products and sought to give their users the best experience they can. The newer products unfortunately tend to feel what they are - the product of a third party who has developed and assembled a cheap product which emulates older, higher quality models and is trading on the back of an established brand name. That seems to be the way of the world now sadly.

Try the 34S :-)

Not that I had much real use for a calculator doing honours mathematics at university.

- Pauli

Quote:

---

This.

---

I applaud your boundary testing, but I wonder how relevant those are for the majority.

Quote:

---

I applaud your boundary testing, but I wonder how relevant those are for the majority.

---

If you, OTOH, are a professional and depend on good results, the 35s is not for you.

I think the logical argument is subjective. Im not saying your opinion is invalid, just that it might not be shared by everyone.

I find it better than the 32sIi with the exception of live comma's during input.

Put it this way;

1. How does the 35S compare to the Casio or TI counterparts?
2. Do you think its more than a few tweaks away from perfect?

Daniel

> Program entry bug for large programs. Create a 999 step program and then try to
> enter a new LBL. You're not allowed. Delete one step, create the LBL and put the
> step back and all is fine.

Lets take this for example, how many times a day do you think the average user will encounter this?

Daniel

Quote:

---

Lets take this for example, how many times a day do you think the average user will encounter this?

---

Quote:

---

32sII was crap

---

There went your credibility.

Quote:

---

There went your credibility.

---

Right. I think that the point he was trying to make is that each generation is considered worse than the last by some. -- 35S is crap compared to the 32Sii and before that the 32Sii was denigrated in comparison to the 32S, etc.

Bugs 1, 9, 15 and 16 are all potentially serious. Some more so than others. None of these are "boundary" conditions.

Programs that lock up the device? (16)

Weird crashes and lock up requiring battery removal? (1 -- although the specifics of these have never been isolated).

Plain wrong results when not near any singularities or ill conditioning (15).

Incorrect display (9).

These are not boundary testing, they are a sure sign of poor implementation. Bugs 4 and 5 also fall squarely into this category. Didn't anyone do even a shallow test of these two?

Bug 15 is the clincher for me. The answer isn't even remotely correct. It is completely bogus. I cannot trust any calculator that cannot evaluate expressions properly -- that is what a calculator is primarily for after all.

Many of the others are pretty esoteric I'll agree but the sheer number of them and that the more serious ones above are present is indicative to me and others of many underlying yet undiscovered problems.

I really wanted to like the 35s when it was released. It looks nice and it has a decent keyboard. Then the bugs started being discovered :-(

- Pauli

OK, perhaps I misread him. Sorry.

I haven't noticed many complaints about the 32sii vis a vis the 32s. I've got both and I really like them. I know the sii is a bit slower at program execution, but not in a major way. I like the simplicity of the keyboard of the 32s, and I like the alpha prompt capability and algebraic solver and fractions of the sii.

Edited: 10 Aug 2011, 11:37 a.m.

I think the 35S is vastly inferior to the Casio FX-115ES.
The Casio can do these things out of the box that the 35S cannot:
vector cross product,
matrix math, as in determinants, inverse, etc.,
solve a quadratic and cubic equation,
straight forward rectangular to polar conversion,
..and maybe more too, those are just off the top of my head

You can get most of that with programming the 35S, but it is very hard. You cannot just use the programs in the manual because many of those programs use the same labels. The 35S does not allow you to reuse labels, so you need to rewrite the programs or find alternates.
Then if you get it programmed correctly you have to pray you never run the batteries dry because you would lose everything.

Edited: 10 Aug 2011, 11:35 a.m.

It is a red flag for me whenever I see someone use the word "maths".

Quote:

---

I haven't noticed many complaints about the 32sii vis a vis the 32s.

---

The original poster may use it ... at his own risk ;-).

Quote:

---

I haven't noticed many complaints about the 32sii vis a vis the 32s.

---

Neither have I. I just *think* that is what the OP was trying to say.

I am not lucky enough to have a 32S (maybe one day) but I have no issues with my 32Sii other than it isn't a 42Sii <g>.

Cheers

Paul, I may be wrong but it seems that the rest of the world uses "maths" as we in the US use "math." I'm not sure why this is so, but I've seen it many times.

Quote:

---

...how many times a day do you think the average user will encounter this?

---

HP abandoned support of the incompetently designed HP 35S immediately after introduction, as shown by the massive list of firmware errors that have been ignored in the five years since. But...you are correct...the HP 35S sure does looks pretty.

Performance-wise, the HP 35S is a fine product for those who aren't concerned about reliability, accuracy, or usability of results...people who are indifferent students, professionals whose work does not use a scientific calculator, or people who are otherwise not invested in any meaningful or philosophical manner to what the machine produces.

If that's you, then you've got something that will bring you much satisfaction. Enjoy!!

This is the one thing that annoys the p*ss out of me.

Why no I/O capabilities???

Quote:

---

I think the 35S is vastly inferior to the Casio FX-115ES.

---

The Casio fx-115ES (a.k.a. fx-991ES) is one of the most remarkable handheld machines I've ever come across in the past 39 years...and it now sells for about $13 at some of the major US retail stores. It's the first non-RPN/RPL machine that I found worth learning to use comfortably.

But of course, it's not (very) programmable. It has those built-in functions you mention, plus numerical differentiation and integration (using Gauss-Kronrod quadrature). Plus, 40 constants and 40 unit conversions. It's just a very nice, very capable, very low cost but durable unit that I've carried around and used daily for more than three years!

IMHO, its biggest downside is the auto-shutdown that takes place after about three minutes of non-use. Any results that were shown in the display will be gone when the machine is turned back on.

I'd take it along before I'd take an HP 35S (four times the price). I also greatly prefer it to TI handhelds, including the TI-89 Titanium HW4. I see only two attractive calculators in the current HP product line: HP 50G and HP 30b.

Thanks for pointing that out, I had no idea.
I wanted more info and found some interesting conversation on the topic here: http://www.dailywritingtips.com/math-or-maths/

Quote:

---

Why no I/O capabilities???

---

A major market -- quite possibly the biggest single market -- for the HP-33S or 35S is for use on NCEES exams. These are the professional licensing exams required for engineers (particularly civils) and surveyors in the US. If you look up the 33S or 35S at Amazon.com, for example, you will see that the "Customers Who Bought This Item Also Bought" list includes numerous NCEES study manuals.

Today, the vast majority of people who buy advanced programmable calculators choose graphing calculators. But graphing calculators are not permitted on NCEES exams -- so this is a niche (perhaps the last remaining niche) for traditional scientific calcs only. And a 33S or 35S is the most powerful allowed option for these exams. In fact, some vendors still sell 33S/35S exam software (perhaps the last remaining form of commercial calculator software).

NCEES disallows calculators with I/O capabilities. If the 35S had them, it would be banned from this market, just as graphing calcs are.

Edited: 10 Aug 2011, 2:08 p.m.

Slightly OT, but not too far:

The late production HP-32S (not II) used the same recessed LCD as the 32SII , which means much better readability due to larger "digits" and non-glare LCD cover, and the much nicer relatively uncluttered keyboard of the 32S, of course;-)

Ray

Mike, you should consider the re-purposed 30b AKA 34S.

I also have used my HP 35s since I got mine in 2008. Its been a blast for me personally. I'm certain that for more demanding engineering area's of use this may be inadequate in terms of accuracy and what have you, but then there are people like myself who want a *quality* calculator that they can buy brand spanking new, and go through a couple years of College along with it and apply their own scuffs, scratches, and bumps on their HP 35s. Each of which I could tell a story about years later. I'm going to be starting my second year here at Purdue, and I have had a lot of fun working with this thing. I love the looks. I love the feel (vs calcs. of today). And I love programing it and discovering new ways to do problems. My personal HP 35s have been drenched in rain countless times while waiting for the city bus, and still works perfectly. I have also not replaced the batteries yet in the 3 years of steady use and ownership. So all in all I am happy with the HP 35s. It was either this or getting the TI-30XA's! (cant use multi-line calcs, but some profs. haven't noticed what i use. . . yet)

Gven its been along time since I've done maths, Ive been revising *everything* from Year 8 maths onwards, doing about 50 problems a day to cath up. Havnt got a wrong answer yet.

Ill let you know when find I find one.

The massive list is filled with irrelevant boundary testing.

Quote:

---

HP abandoned support of the incompetently designed HP 35S immediately after introduction, as shown by the massive list of firmware errors that have been ignored in the five years since. But...you are correct...the HP 35S sure does looks pretty.

Performance-wise, the HP 35S is a fine product for those who aren't concerned about reliability, accuracy, or usability of results...people who are indifferent students, professionals whose work does not use a scientific calculator, or people who are otherwise not invested in any meaningful or philosophical manner to what the machine produces.

If that's you, then you've got something that will bring you much satisfaction. Enjoy!!

---

http://homepage.mac.com/nwjh/HP-35S

Has most of the things you listed.

Also,

Page 4-10 of the manual shows rectangular to polar conversion
page 17-10 of the manual shows vector cross product

Do you have a 35s or just relaying what youve hard others say?

I was relaying my experience reading the various opinions on these forums and elsewhere. One guy says the 32sIi is great, then I read its crap compared to the 35s, another says its a poor cousin to the 42s, then some compain at the 42s etc and so on

Quote:

---

There went your credibility.

---

Theres a bunch of guys in HP who make calculators. im sure Carly did give them a beating as a non-core business unit in her new HP.

However, when these guys finally get approval to take a scientific calculator to market, and the old voices get listened to i terms of look and feel etc, then if you want a better and improved 2nd iteration, you better support whats left of the old guys in the calculator division and buy one.

Buying 20 year old calculators on ebay doesnt turn up on the calculator divisions balance sheet, and lack of interest in RPN becomes a self fullfilling prophecy.

Quote:

---

This.

---

6 dont get it, works either way
9, shows correct value

Just did fifteen and pressed run 20 times, always same answer

-466.9270

16. Im still new to RPN programming, how do I find SF 10?

As for the rest, youve written a loop without an exit conditiom. Isnt this just bad prgoramming? I once crashed a big As/400 with a bad loop I wrote, did use 200 gig of virtual memory first (this was alot 10 years ago).

The point is that inexpensive calcs like the Casio FX-115ES feature capabilities such as vector cross product as intrinsic operations, whereas they have to be programmed into the HP-35s, which is an inexcusable omission for a calc sold at its price. Also, given the limited 1-character label system in the HP-35s, the cross product program alone consumes the 3 letters R, E and C, so they are no longer available for other programs. Of course, HP decided to discontinue the great HP-42s, which allows variable and label names up to 7 characters long and has the vector cross product as a built in function along with complete matrix capabilities, even with complex numbers. Now, admittedly, the HP-35s inherited all these limitations from the HP-32SII, however, it's really just a cheaper version in a pretty modern looking package that only has emotional value IMO.

Other than display contrast and no memory I/O, the only real complaint about the HP 42S by knowledgeable people has been the lack of any subsequent HP RPN scientific that even approaches the capability of the 1988 HP 42S since that was discontinued in 1995.

Many of us had hoped that the HP 35S, whose "35" designation is said to honor the original HP-35, would motivate HP to finally produce a new RPN machine that was similar to or better than the 20-year old HP 42S. But HP grossly failed the opportunity. (HP doesn't even have enough attention to detail to get arrows on the yellow and blue shift keys pointing the correct direction...yellow functions are UP, not UP AND LEFT, and blue functions are DOWN, not UP AND RIGHT. Sheesh...how hard would that have been?)

Sadly, the HP 35S is a worthy successor to nothing...well, except maybe to the HP 6S (if you can live without the solar cell). :-)

Edited: 10 Aug 2011, 11:12 p.m.

Quote:

---

This.

---

On my 35s

89.99999999 (is that close enough to 90 to make the bug happen?). I get;

1.74532925199 E-10

On the 42s on my iPhone I get, wait for it;

1.74532925199 E-10

Clearly I'm missing something. I can take a photo if you want.

My serial is CNA 03701109

Daniel.

?????

I am also having trouble recreating this one. 89.99999999 COS matches answer on 42S and 48SX. Ditto for the TAN bug.

EDIT:

Never mind. I see the problems.
For COS I see a problem with the following:

89.99

89.999

89.9999

89.99999 (truncated?)

89.999999 (truncated?)

89.9999999 (truncated?)

89.99999999 (correct?)

How very odd...

Mind you, I have to question whether the error is really significant given that the error at ir's worst appears to be less that 1E-14. It is a bug however. The errors are more significant for TAN. Reason I didn't see it initially is that for 89.99999999 TAN both the 35 and the 42 give the same answer.

Cheers

Edited: 11 Aug 2011, 12:12 a.m.

Quote:

---

The point is that inexpensive calcs like the Casio FX-115ES feature capabilities such as vector cross product as intrinsic operations, whereas they have to be programmed into the HP-35s, which is an inexcusable omission for a calc sold at its price.

---

Ok, but the Casio isn't programmable, isn't RPN and isn't built as well. I'm sure your right, but you'll also get more features in a Toyota compared to a BMW for the money, or a bigger engine in a Ford.

I still don't think it detracts from the HP 35s. Which is very cheap at $59 compared to what they used to cost in the past. Personally, I'd rather they increase the price and further increase quality (i.e. include paper manual). However, this would further reduce value for money.

Quote:

---

Also, given the limited 1-character label system in the HP-35s, the cross product program alone consumes the 3 letters R, E and C, so they are no longer available for other programs.

---

You could re-write it to use 1 variable. Some people break down their programs into routines for re-use, but I'm sure you already know that.

Quote:

---

Of course, HP decided to discontinue the great HP-42s, which allows variable and label names up to 7 characters long and has the vector cross product as a built in function along with complete matrix capabilities, even with complex numbers. Now, admittedly, the HP-35s inherited all these limitations from the HP-32SII, however, it's really just a cheaper version in a pretty modern looking package that only has emotional value IMO.

---

But you'd agree the HP 35s is better than the HP 32S, the 32Sii, has more memory, more variables. It also has a two line display. Its also 1/3 to 1/4 of the price of a 32sII on ebay!

My point is its a good calculator! Is it the last best calculator that HP ever need to make? No, but from what I read in these forums, you'd think it was terrible!

If they sell alot of these and bring out a mkII, it really wouldn't take much to turn it into a 42s killer.. if you had multi-label programs, and a SD card reader, it would be pretty close right?

Quote:

---

You could re-write it to use 1 variable. Some people break down their programs into routines for re-use, but I'm sure you already know that.

---

I believe that Michael is referring to the labels needed to call 3 programs. Not to the memory variables.

Quote:

---

Many of us had hoped that the HP 35S, whose "35" designation is said to honor the original HP-35, would motivate HP to finally produce a new RPN machine that was similar to or better than the 20-year old HP 42S.

---

Quote:

---

But HP grossly failed the opportunity. (HP doesn't even have enough attention to detail to get arrows on the yellow and blue shift keys pointing the correct direction...yellow functions are UP, not UP AND LEFT, and blue functions are DOWN, not UP AND RIGHT. Sheesh...how hard would that have been?)

---

The HP 41C had sloped keys like the 35s and it has BLUE on the bottom of the key, same as the 35s.

Its easier to read than the 32sII was, thanks to the sloped keys. I see it as an improvement, and is consistent with previous products.

Quote:

---

Sadly, the HP 35S is a worthy successor to nothing...well, except maybe to the HP 6S (if you can live without the solar cell). :-)

---

Other than your hopes that it would replace the 42s not being met (which given its a 30 series, what ever gave you that idea?), why isn't it?

If you ranked all HP RPN scientifics made, it would have to be in the top four, some people don't mind the lack of IO and love the 42s, some people want horizontal format, and want a 15c, but after that, the next best is the 35s.

For $59, it probably doesn't deserve the kind of derision your directing at it.

Daniel.

Quote:

---

I believe that Michael is referring to the labels needed to call 3 programs. Not to the memory variables.

---

Yes, my point was the program listen on 17-11 of the user guide for Vector cross product uses 3 labels, but with the GOTO feature in the 35s, it doesn't need to. You could re-write it to use just 1 label.

Sure, your still limited to 26 labels, and doesn't replace the 42s, but its not the end of the world, and better than the 32s or 32sII, which don't seem to have the same level of dersion heaped upon them.

Daniel.

I don't quite see how buying an unsatisfactory product brings an improved version that much closer. From the sound of it, there hasn't been any move by the company thus far to correct the software bugs that have been discovered by end users.

I agree, is a bit sloppy that they're loosing nth degree accuracy (for no good reason), not sure its the end of the world, given we should all know our exact values for 90 degrees!

Interestingly, have a look at the wildy different accuracy's of trig functions on calculators;

http://www.rskey.org/~mwsebastian/miscprj/models.htm

In order of accuracy for arcsin (arccos (arctan (tan (cos (sin (9) ) ) ) ) )

HP 35S, 8.99999986001

HP 42S, 8.99999864267

HP 32S, 8.99999864267

HP 41C, 9.000417403

HP 15C, 9.000417403

Funnily, its more accurate than the 42S etc for other trig values. Its obviously a different algorithm, but maybe not a bug??

Edited: 11 Aug 2011, 12:48 a.m.

Quote:

---

http://homepage.mac.com/nwjh/HP-35S

Has most of the things you listed.

Also,

Page 4-10 of the manual shows rectangular to polar conversion
page 17-10 of the manual shows vector cross product

Do you have a 35s or just relaying what youve hard others say?

---

Let me try to explain in detail this time.

That page you linked to has the quadratic formula on it.
He has no program for vector cross product, or matrix operations, or the rectangular to polar conversion. So your statement "Has most of the things you listed." is incorrect. He has one thing we need, the quadratic program. For that program, you use one label, Q, and yet still can't do a cubic like the Casio. You could try the program in the manual too if you like.

The rectangular to polar in the manual is a bit of a hack and has its shortcomings.
People here had a better solution. If you ask HP about this they send you a PDF with the solution that the guys here came up with. No kidding, and that's how I had to get it. Label used for this program: Z

Vector Cross product. Yes, go ahead and use the manual. Labels R, E, and C get used up.

If you want matrix operations, you can go here and type in all 400+ lines: http://www.stefanv.com/calculators/hp35s_matrix_multitool.html
He did a great job, but over 400 lines, yikes. Good luck.
Label used: M

That's a lot of work to get a 35S to do the same things as the Casio, and the Casio is under $20 at Walmart, Office Max, etc. so it's cheap and easily replaceable as well.

If you do get the 35S programmed with all the above then you have to be nervous about losing battery power, because if you do, it's all gone.
And, since HP decided to put the rubber feet on the battery door of the 35S, it is possible that you can knock the door off and the batteries could get dislodged. I know that is it a long shot, but stranger things have happened.
The 33s did not have this problem since it had metal plates to hold the batteries in place, plus the door had no feet on it. You could lose the battery door of a 33s and it wouldn't even matter. That was a major ergonomic step backwards for the 35S in my opinion.

Anyway, you seem to like your 35S so enjoy. Good luck with all the programming.

Edited: 11 Aug 2011, 1:43 a.m.

Quote:

---

I agree, is a bit sloppy that they're loosing nth degree accuracy (for no good reason), not sure its the end of the world, given we should all know our exact values for 90 degrees!

---

You're doing maths. You will do most everything in radians. Asking for people to know exact multiples of Pi / 2 is a very different thing.

Quickly, which of these numbers isn't a near multiple of Pi/2?

 26.7035375555

-36.1283155163

  7.85398163397

 17.2877595947

  9.42477796077

Since your calculator cannot be relied on to return accurate results, you have to do this by eye.

As I've mentioned, I like the esthetics of the 35s and have written what must be one of the largest programs for it. However, a calculator must instill faith in its results and the 35s lets me down here.

For almost any other HP calculator I have faith that the answer I get is as close to right as is reasonable. I have confidence that a program I write will operate as written.

- Pauli

Quote:

---

If you ranked all HP RPN scientifics made, it would have to be in the top four, some people don't mind the lack of IO and love the 42s, some people want horizontal format, and want a 15c, but after that, the next best is the 35s.

---

This entirely depends on what criteria you use to rank them. I certainly wouldn't rate it in my top four based on any criteria I really care about. I suspect very few here would. 41c, 41cv, 41cx, 15c, 42s, 32s, 32sii, 11c, 34c, 32s, 67. All are superb calculators and all for good reason. Toss the 34S and 41CL into the mix as well (they are kind of HP RPN calculators).

For accuracy, 15c, 11c, 10c, 32s, 32sii, 42s would be my top six. William Kahan designed the algorithms for the Voyagers and did a splendid job of it. Most subsequent HPs have used these.

For esthetics, the list is very variable. Many people only like red LEDs, other hate dot matrix. It would be impossible to make a list that everyone would agree on but for some people it would make the top four for others the bottom.

Light weight? It would go okay there.

In terms of most known bugs, it scores very well indeed :-)

Battery life? The three scientific Voyagers would have to be top of the list followed by the 41 series. 20+ years one on set is not unheard of for a Voyager.

Ones I reach for to do casual calculations? Definitely not even mentioned on that list.

In the class of HP RPN scientific dot matrix calculators with over 16kb of RAM it is number 1 or 2.

Speed? It goes okay here some of the time.

Still, if it is what you like use it. You are not going to convince anyone here that it compares well with the classics...

- Pauli

Quote:

---

Interestingly, have a look at the wildy different accuracy's of trig functions on calculators;

---

...

Quote:

---

HP 35S, 8.99999986001

HP 42S, 8.99999864267

HP 32S, 8.99999864267

HP 41C, 9.000417403

HP 15C, 9.000417403

---

I think you'll find this is demonstrating the exact opposite of what you hope :-)

Calculating these operations with intermediate results correctly rounded to 10 and 12 significant digits:

Operation	10 digits	12 digits

9		9.000000000	9.00000000000

SIN		0.1564344650	0.156434465040

COS		0.9999962727	0.999996272743

TAN		0.01745499985	0.0174549998555

ATAN		0.9999962724	0.999996272744

ACOS		0.1564416604	0.156434441642

ASIN		9.000417403	8.99999864267

I see the values in the final row in the table you presented above but for the older HPs not the 35s. This indicates that the older HPs are performing their calculations with correct rounding in the last digit for all these operations. You cannot ask for better accuracy than that.

The 35S on the other hand certainly isn't correctly rounding its results. That the final result appears "better" is a pure fluke of randomness not because it is better designed or implemented -- it isn't.

- Pauli

Edited: 11 Aug 2011, 2:38 a.m.

Sorry I cant do nice formatting on this forum like you did, but 35s

9
0.15643446504
0.999996272743
0.0174549998555
0.999996272743
0.156434462627
8.99999986001

Whats wrong that? The first Atan is more accurate than the 42s, and it ends up with a more accurate result.

How can you say its not doing it correctly?

Can we really realistically expect anyone to make a red LED calculator. Heck, why not nixie tubes. Not sure this is a valid criteria.

I like the old machines too, but for a non-collector, HP newbie, should I get a Casio or TI or HP 35s?

I think you can say its a good calculator.

Daniel

I'm using pre tags to format this nicely.

Here are your results again slightly annotated:

9

0.15643446504     SIN

0.999996272743    COS

0.0174549998555   TAN

0.999996272743    ATAN

0.156434462627    ACOS

8.99999986001     ASIN

Quote:

---

Whats wrong that? The first Atan is more accurate than the 42s, and it ends up with a more accurate result.

How can you say its not doing it correctly?

---

ATAN( 0.0174549998555 ) is to thirty significant digits 0.999996272743534515028695045928. I've made the first 12 significant digits bold. Look at the thirteenth digit, it is a 5. I don't know about you, but I was taught to round things larger than 5 up.

In this case the correctly rounded result is 0.999996272744 which is what the older devices are getting. This is the correct answer for any device producing results with 12 decimal digits. The 35s is wrong here. It isn't rounding its results correctly. Yet another reason to not trust its arithmetic. This is cascading in later calculations and in this case just happens to produce what appears to be a more "accurate" result. It isn't, the 42s is producing the ideal perfect result for this calculation for a device using 12 decimal digit numbers.

Don't believe me? I'd suggest some further reading from William Kahan's pages. He designed the algorithms the older HP devices are using and he also writes quite well and has plenty of examples where seemingly silly insignificant errors like this cascade out of proportion and completely ruin all the later results.

- Pauli

Quote:

---

ATAN( 0.0174549998555 ) is to thirty significant digits 0.999996272743534515028695045928. I've made the first 12 significant digits bold. Look at the thirteenth digit, it is a 5. I don't know about you, but I was taught to round things larger than 5 up.

---

Quote:

---

I don't, but rounding half to even would also produce a result different from the 35s output in this case.

---

Rounding half even is for the exactly halfway case only. That's why I wrote larger than 5 up :-)

Exactly 5 is a special case. i.e. if the result were 0.99999627274350000000000000000000 then round half even would come in to play. Not that it matters in this case.

- Pauli

Sorry Pauli, I managed to misread what you wrote. You're right of course.

Quote:

---

ATAN( 0.0174549998555 ) is to thirty significant digits 0.999996272743534515028695045928. I've made the first 12 significant digits bold. Look at the thirteenth digit, it is a 5. I don't know about you, but I was taught to round things larger than 5 up.

---

But its not wrong, it is the right result, the original ended with a 3, and the inverse does of tan comes back. Have you thought they might actually be storing some variables in a higher resolution?

How can you say its more accurate for the Atan result to be different from what the original was! Thats madness.

Quote:

---

Don't believe me? I'd suggest some further reading from William Kahan's pages. He designed the algorithms the older HP devices are using and he also writes quite well and has plenty of examples where seemingly silly insignificant errors like this cascade out of proportion and completely ruin all the later results.

- Pauli

---

Thanks, I will have a read.

Im just happy there is a 35s, otherwise there would be no future. Im probably a bit younger than some here and have another 30 years to cover. What do you think my chances of getting a NOS 42s in 2041 will be?

Its not perfect but there is nothing else close available new. If someone starts building brespoke low volume calculators at exhorbitant prices Id buy one. At the moment there is no other alternative.

I think the nit picking in this forum exagerates the delta between the 35s and nirvana.

Ps. Just had a read, basically Matlab is inaccurate on anything less than 68k based Apple Mac. I trust you will dutifully dispose of your new computer and buy a vintage Mac.

Edited: 11 Aug 2011, 6:34 a.m.

Quote:

---

Thats madness.

---

Not madness, just floating point arithmetic. Forget just about everything you know about normal arithmetic, it doesn't apply anymore.

A couple of examples to confuse:

• a + (b + c) is not always equal to (a + b) + c.
  
• With binary arithmetic, x * 0.01 is not the same as x / 100.
  

Have a look at the WP34S, it is more akin to what the 35s could have been. I'm biased of course, but I've tried to make the results as good as possible.

- Pauli

I understand floating point, 19 years as a programmer.

1. Does the 35s store itermediatory values in higher precision
2. Why does it get more accurate answers?

Where can I buy a 34s? What about the bug reports posted at the top of the forum?

As I said, make a bespoke calculator and Ill buy it. (people criticise the printed 35s keys and your talking stickers?). There are some double standards here.

This is the future?

Quote:

---

I understand floating point, 19 years as a programmer.

---

The former doesn't necessary follow from the latter :-) See What Every Computer Scientist Should Know About Floating-Point Arithmetic. I'd be surprised if just about any programmer didn't learn something.

Quote:

---

Where can I buy a 34s? What about the bug reports posted at the top of the forum?

---

The 34S is a repurposed 30b. It is available ready flashed. e.g. http://www.thecalculatorstore.com/ but there might be others.

Yes, there have been bugs and no doubt there will be more. They have always addressed quickly. We've also had no numeric issues for quite a while now and not many of those ever. e.g. the forensic test gives the proper correctly rounded result for a sixteen decimal digit device.

Quote:

---

As I said, make a bespoke calculator and Ill buy it. (people criticise the printed 35s keys and your talking stickers?). There are some double standards here.

---

The stickers are an economy measure. Commit to buying a couple of thousand devices and we can get a printed keyboard ;-) I'm not willing to commit to that kind of quantity.

- Pauli

Nope. There is a *much* nicer overlay now. That is also based on a 20b which has a poor keyboard. The 30b is much better in this regard.

- Pauli

Why dont you offer the OS to HP license free? Where can I read more about the UI etc. Im sure I can learn more but years ago I did units on scientific computing where we had to evaluate algorithms for significance. Mainly C and fortran.

If HP released the 34s would the 42s crowd be happy?

1000 units is not that much. If you can deliver youd raise that in pre-orders.

Daniel.

Ps. Sorry for the criticism of the stickers, but its not a serious alternative.

I am removing the earlier picture from my public dropbox folder. This is a much better representation of the beauty of the 34s.

Quote:

---

Why dont you offer the OS to HP license free?

---

We already did. Weren't interested.

Quote:

---

Where can I read more about the UI etc.

---

The manual is on the sourceforge site both separately and in the zip archive -- there is a windows gui emulator there to play with too. The leading edge is in subversion.

Quote:

---

If HP released the 34s would the 42s crowd be happy?

---

Probably not, it isn't a complete super set of the 42s :-) I think a lot of people would be happy with it though.

Quote:

---

1000 units is not that much. If you can deliver youd raise that in pre-orders.

---

I wrote thousands :-) I'm not willing to commit to purchasing that many in the hopes that it sells.

Quote:

---

Ps. Sorry for the criticism of the stickers, but its not a serious alternative.

---

The stickers are very good and quite usable. Mine aren't showing any sign of wear after quite a bit of use. Also no sign of coming off. If they ever do wear, they can be replaced.

- Pauli

Thanks Gene. That is much better.

- Pauli

Yes, you are correct. I wasn't looking at the program and focused on your use of the word "variable", sorry.

And yes, I made that exact point about label usage in a recent discussion of the 35S. It is far superior to the 32S, 32Sii and 33S in regard to labels (and some other things too).

Quote:

---

I applaud your boundary testing, but I wonder how relevant those are for the majority.

---

I think Ill stick with the "completely useless" 35s. You guys were pretty quick to cover up and get rid of that other photo.

I guess when I fail all my exams and assignments from all the bugs Ill learn my lesson *cough*

I think the attitude of the OP is that we should all be grateful that HP even sells any RPN calcs anymore, regardless of how faulty they are because there are no alternatives. Fortunately, there are a few afficionados who have provided alternatives outside the HP corporate circle. Meanwhile, I'll continue to use and enjoy my many vintage HP and even non-HP RPN calcs that are still fully operational due to their robust construction. IMO the only currently produced HP scientific calc that is worth a dang is the HP-50g, and it wouldn't bother me one bit if they discontinued production of all the other models, including the HP-35s. Were it not for the various PE and NCEES exams, I'm not sure the HP-35s would have a significant market, given that for the same price one can buy a decent graphing calc, which is what most students and professionals prefer. I retired 2 1/2 years ago from my company, where only the older engineers such as myself used RPN calcs, and the company was supplying the junior engineers with TI graphing calcs. As to what I used when I took my EIT exam, it was a slide rule and a CRC book of math tables.

Which? And how?

Especially given that at least one third are not bugs. The list is also by someone who has a bias as the creator of his own calculator software (which is ironically buggy).

Daniel

Quote:

---

16. Im still new to RPN programming, how do I find SF 10?

As for the rest, youve written a loop without an exit conditiom. Isnt this just bad prgoramming?

---

Been eyeballing the posts and want to throw my 2 cents worth in. Nobody is really wrong here. For the purists, the 35S doesn't stack up to some of the vintage hardware. But for those who use the 35S for everyday calculations, it's a great performer. The keyboard has a better feel than any other brand on the market. The bug list? To be honest, my real-world, daily "engineering" calculations consist of basic algebra, logs and trigs. No boundary extremes, only a few short programs that I wrote and regularly use. For that purpose, it truly is a nice, daily driver calculator. I don't care for the feel or methodology of the TI or Casio calcs. Would I love to have all the bugs fixed and the keyboard layout re-done? You bet. But I can also wait without grousing all the time.

Quote:

---

The list is also by someone who has a bias as the creator of his own calculator software (which is ironically buggy).

---

Not fair. First, I believe that the list was a compilation. Second, this list came out well before the WP34S. And, third, bugs in the WP34S actually get fixed!

Cheers,

Marwan

Quote:

---

Not fair. First, I believe that the list was a compilation. Second, this list came out well before the WP34S. And, third, bugs in the WP34S actually get fixed!

---

Seconded!! I too have a 35s and I honestly like it, and for the way I use it I have never encountered the bugs; but it's true that the developers of the 34s fix bugs almost immediately. I myself have reported a few and I think that no company could give such a wonderful "customer support". They even added a few feature I requested, mere hours after I requested them!! And bear in mind that the whole 34s project has been brought on by 3-4 people in their spare time, for free, and they don't have a huge company backing them, which makes their achievement all the more incredible IMO.
That said, I still like my 35s, and I don't want to bash it, but it would have been only fair (and standard practice for a serious company) to release a fixed rom after the bugs were discovered... after all, they have had the time, and they're a business, and they need to "fidelize" customers...

Cristian

Quote:

---

[...] You guys were pretty quick to cover up and get rid of that other photo.[...]

---

Hmmm... So what, if they did? I didn't even see what the other photo was, but the second picture is what a 34s looks like, if you do it properly. So if the previous photo was of some home-made, hand-cut, flimsy early prototype cover, it's only normal that it looked uglier. The 34s as it is, is a wonderful calculator, and with the professionally-cut vinyl stickers it looks much better than many other commercial calcs. You should at least try the emulator before bashing it. As I said elsewhere, I too have a 35s and I like it, but for everyday use I use the 34s. Of course, I wish it had the two-line display of the 35s, and its nice body shape...

I refuse to do any large-scale "permanent" programming on any non-IO machine. I just think it is crazy.

I was seriously behind the curve, living happily with my 11c all the way up to 1995 and then until 2001 with a 32sii. Then I discovered I/O. Why waste any time on non I/O for anything other than small ad-hoc programming?

Edited: 11 Aug 2011, 12:31 p.m. after one or more responses were posted

A certain person named Lorenz found out about the importance ordinary correct rounding can have on nonlinear systems...:-)

I agree 100%. That's why the limited memory of the HP-32SII doesn't bother me one bit, and the massive memory of the HP-35s is irrelevant. All my serious programming is contained in my HP-48SX and HP-50g.

I switched to a 35s for a while. Before that, I switched to a 33s for a while. Going back to the 32sii was like coming home :-)

Actually, I use all these devices less and less, to a trivial extent now. And my favorite isn't even RPN any longer!

I bought my children 33s machines. They are much better than the 35s. The keyboard "looks funny" but the direct feature set and non menu function set is way better. And you can decompose polar to rectangular. In other words, it is a *useful* calculator that doesn't cost a lot. Whereas the older HPs which are better, cost a lot.

Edited: 11 Aug 2011, 12:06 p.m.

The market speaks. It is Ti and Casio, with the HP being a niche.

That is a sure sign that HP blew the opportunity. I agree with Mike Morrow on that point.

New products ultimately need to be better than old products and need to be superior to competing new products. The Casio comparison is apt.

Edited: 11 Aug 2011, 12:25 p.m.

Quote:

---

. I don't care for the feel or methodology of the TI or Casio calcs.

---

That is an important point. The Methodology. I know what you mean. Sharps, Casios, and perhaps to a lesser extent TI methodology drive me nuts. (Except the Sharp BASIC machines, which were totally fantastic but that is another story...)

Edited: 11 Aug 2011, 12:29 p.m.

The first photo which you linked to in my public dropbox account ...

(which by the way is great...email me to get a referral so I can get even more space to store stuff online ...free)

... was the very first attempt at an overlay from a .jpg file I printed and taped over the 20b.

About 1-2 weeks later, Eric Rechlin had created the overlay system in use now which is marvelous.

If you were trying to get / use a 34s today, it would look like the picture you see now in this thread, not the one I did about 3 days after the firmware was running on the calculator.

So, please, no accusations of bias, which are unfounded.

I got rid of my HP-33s, because I didn't like the buttons, the decimal point was invisible (early S/N, later ones fixed that), and it ate batteries. The HP-35s is much better in this regard, however, I still use my HP-32SII a lot more. If I do in fact receive the HP-15C LE that I've ordered, I may put the HP-35s up for sale as well.

Edited: 11 Aug 2011, 12:49 p.m.

I am one of those that actually likes the 35S. But no, sorry, not in my top 4. I think that Pauli makes some good points that no matter how you choose your criteria for rating this machine there are not many areas in which it is top 4.

That does not make it a *bad* machine in my opinion but neither is it an all-time great by any measure.

Better than the 33S? Yes (IMHO). Better than the 32Sii? In some ways--worse in others. Better than the 42S? Not even close. Better than the 41 series? Even less close. Better than the 15C? It has some advantages over that machine (actual mnemonics in programs) but many, MANY, disadvantages. Better than the 34C? Not in my book. The 55, 65, 67? Again, not in my book.

I like the 35S and use it now and again but it will never replace my 41 or 42 (never mind my RPL based machines) and with the 34S available it actually slipped one place (still waiting for the keyboard overlay to become available again).

Feel free to use whatever calculator you prefer. We - being older members of this forum - can only recommend. And FYI, some of us (not me) are real hardcore experts in this matter.

I think, however, that you tend to compare apples and oranges:

• The HP 35S is put on the market by a huge company. HP was famous for its calculator quality decades ago already, so they should really know how. And they came out with a pretty buggy 35S - falling short of their own quality standards as incorporated in earlier calculators.
  
• The WP 34S is the work of a small gang (two heads most of the time, three since March, with occasional support by a few more). It's a hobbyist project as you can read here (3.5MB Owner's Manual).
  

Just FYI,

Walter

Fair enough. Given all of the custom roms being written, Im suprised no one has done a usb i/o hack for the 35s. The box itself is nice. If they could also work out how to implement multi-labels the job would be done.

The 35S lacks any possibility to update / extend the functionality once the thing is built because the ROM is just that, read only. So there isn't a chance for an outside party to enhance or bug-fix the device. Only HP or their contractors have the option to create new ROM contents and only for newly produced machines.

Quote:

---

If you were trying to get / use a 34s today, it would look like the picture you see now in this thread

---

Unfortunately, it would look an awful lot like a stock 20b or 30b, since the overlays haven't been available for almost two months.

Justin,

Please contact Eric Rechlin via the museum mail and ask him about his progress in this matter. He doesn't respond to everybody but you can try ;-)

HTH,
Walter

I emailed him a few days ago. He told me that they should be available in a couple of weeks or so.

Cheers,

-Marwan

Quote:

---

The HP 41C had sloped keys like the 35s and it has BLUE on the bottom of the key, same as the 35s.

---

OK...let's try this again.

Look at the yellow shift key. See the arrow going UP and to the LEFT? That should indicate that the yellow function label associated with any key is...yes...ABOVE and to the LEFT of the associated key. But it's not. It's directly ABOVE the key.

Look at the blue shift key. See the arrow going UP and to the RIGHT? That should indicate that the blue function label associated with any key is...yes...ABOVE and to the RIGHT of the associated key. But it's not. It's BELOW and slightly LEFT of the central face of the key.

Clear?

When so little attention even to small and easily corrected cosmetic detail has been exercised by the manufacterer, it's a certainty that much more significant problems will be found elsewhere. In that respect, the HP 35S more than fully meets all expectations!

Quote:

---

Other than your hopes that it would replace the 42s not being met (which given its a 30 series, what ever gave you that idea?), why isn't it?

---

Sigh! The HP 35S is NOT part of any "30 series" that you just attempted to fabricate from whole cloth. The 30-series machines are the Spice models of the late 1970s and early 1980s, such as the HP-34C. There have also been other HP machines that are "thirty-something", such as the HP-35, HP 38G, HP 39GS, and HP 30b, to list some. Do you imagine these to be part of your 30-series too? In fact, the HP 32S and 32Sii are part of the Pioneer series. The HP 33S is not part of any other series, nor is the HP 35S. Some study before posting could possibly aid you if a factual presentation is a goal.

Quote:

---

If you ranked all HP RPN scientifics made, it would have to be in the top four...

---

I'll assume that what you intended to say was actually "If all models of HP RPN scientifics ever made were ranked, the HP 35S would have to be in the top four..."

That, of course, is patent nonsense and an indicator of unfamiliarity with the subject. The errors of the HP 35S design, uncorrected after almost five years of production, earn it well the status as possibly the worst design that HP has ever tried to pawn off.

Quote:

---

For $59, it probably doesn't deserve the kind of derision your directing at it.

---

In an age where a capable laptop or notebook PC can be had for less than $300, the persistently defective HP 35S is a very poor value indeed and in fact.

Throughout, your argument seems to be that as long as most of the time a correct result is produced, the product is satisfactory. In that view, if the HP 35S added 2.000 plus 2.000 and got 4.001, that would be acceptible because only one case out of an infinite number of possibilities was incorrect. Unlike programmers, most in the scientific and (especially) engineering professions must have significantly higher and better standards.

Those whose standards are, er, "relaxed" may certainly relish their HP 35S hot dog. But don't expect knowledgeable members of this forum to bestow an imprimatur to the selection, nor accord the praise and approval that you are obviously seeking. It's not going to happen.

Edited: 13 Aug 2011, 11:17 a.m.

Please note that the Free42 or 42S running on the iPhone (or iPod Touch) is not an exact emulation of the HP 42S. It's functionally very similar, but it doesn't run the HP 42S ROM image or internal code.

Quote:

---

He told me that they should be available in a couple of weeks or so.

---

Thanks for the update, Marwan. I emailed him about a month ago through his website and didn't get a response. I didn't want to bother him again.

On the other hand, using a "blanknut" 34s will probably help once I do get an overlay. I've memorized a lot of the layout just by referring to the diagram in the manual and picturing it on the 30b. I've gotten where I don't really even "see" the 30b's keyboard anymore. I shouldn't have too many problems finding functions once the keyboard is actually labeled!

LOL -- Precision is in the eye of the beholder?

Still, better than a "couple of months or so" :-).

Quote:

---

Please note that the Free42 or 42S running on the iPhone (or iPod Touch) is not an exact emulation of the HP 42S.

---

Does your critique apply to just the iOS version or does it apply to Free42 on any platform?

Are you aware of any difference in calculation results between Free42 (decimal) and the HP 42S? (Free42 iOS is only decimal math.)

Quote:

---

I had one and sold it. Did you try to see what labels are used by these programs? Do you not understand the amount of work involved to get a 35S to be at the same level of capabilities as the Casio?
...
That's a lot of work to get a 35S to do the same things as the Casio, and the Casio is under $20 at Walmart, Office Max, etc. so it's cheap and easily replaceable as well.

---

How do I use the Casio with RPN? Seems like that is even more work. Or if your criteria is cheap then that $20 is easily beatable.

In 1985 I needed something with more oomph than the TI-30 I'd been using since 1976, and a programmable was strongly recommended. After comparing what was available for under $200, the choice was either a TI with LEDs and I/O, or the HP 15C. The 15C had no I/O, but seemed to be higher quality, but used RPN. The 15C was a newer design, and more expensive than the TI. I decided on the 15C and never regretted that decision.

After using RPN for 25 years anything else is just a toy, not a tool. I wish Casio, Sharp, etc. would produce RPN calculators comparable to their current products. But they don't. And as such any comparison I see with an RPN calculator has only one possible outcome -- RPN or nothing.

My next calculator tool was a 28S (gift from my wife, and she got a 48SX!). Now I mostly just collect them... A toy would do all I need from a calculator, but why buy a toy that causes me to lose my place and/or my results when I use it?

I have several old HPs now, and a new 12c (25th platinum) and a 35S. The 35S seems fine. It has never caused me to lose my place in a calculation or lost my results forcing me to rekey. I cannot say the same for the TI-84's my boys are required to have for school.

Probably someday I'll pick up a 30b or maybe two, and maybe one of the faster 12c's. They just aren't quite special enough yet for me to lay out the cash. Now if my HP stock would get back to $65 or so...

Quote:

---

Theres a bunch of guys in HP who make calculators. im sure Carly did give them a beating as a non-core business unit in her new HP.
... you better support whats left of the old guys in the calculator division and buy one.

---

I'd love to be wrong, and somebody please correct me if you _know_ otherwise, but I wouldn't describe the team as being "a bunch" and I'm pretty sure there are no old calculator guys doing calculators at HP.

In other words, the last products the old calculator guys did were the pioneer line, the first of the RPL and graphing models, the palmtops (95lx, 100lx, 200lx) and some of the early laptops (omnibook 300, remember the funky little popout mouse?). Those products were designed in corvallis. The calculator responsiblity was relocated to singapore (or somewhere asia-pacific) without personnel, and everybody in corvallis went on to other things.

That's when calculators mostly languished and died at HP. Responsibility moved from asia to australia, and then most recently to a kind of distributed operation again with all new folks.

I agree with supporting the calculator folks at HP. They are doing great work in a challenging environment.

Just don't be confused into thinking the current folks have inherited anything from the folks that did the old machines that everyone here is always raving about. The old guys are gone and 99% of what they did was lost during the shuffle starting in the early to mid-1990's. Those old calculator guys don't even recognize HP as being the same "Hewlett-Packard Company" that they used to work for, because it isn't -- neither figuratively nor technically (the old Hewlett-Packard Company of California split apart and the part that kept the name was acquired by HP of Delaware and the intellectual property was transferred to yet another company, Hewlett-Packard Development Company L.P. of Texas, with HP Delaware allowed a perpetual license under condition that any new IP would be owned by HP Texas).

I would recommend taking that 30b and flashing it to a 34s.

You'll enjoy that machine.

It's not a critique, just something I believe may be convenient to clarify.

I understand that any details about Free42 internals (such as precision, algorithms, etc.) applies to all versions, not just iOS versions.

I understand that Free42 Decimal works with more precision than the physical HP-42S calculator. You can look at the Free42 website or contact Thomas Okken for detailed information.

Best regards,

(Please disregard any idiomatic mistake)

Instead of using six nines after the decimal try seven. 89.9999999
I do have a 33s with a 2007 date code and it should have the same trig results as the 35s. Note these trig inaccuracies started with the 33s.

33s: 572,960,832.397
Free42: 572,957,795.131

The cos and tan bugs exist and you should not assume they show up at the same angle. Try a few things and you will see.