Musings and comments about our common interest
Old HP machines!
We’re now in the wake of the new major release of Mac operating system, Yosemite. There will be a number of new features but also correction for many small errors of the previous software. This is typical of today’s software: due to its own complexity, it is bug-ridden, and unsafe in many respects.
Compare that with the 12k * 10 bit of the assembly code of the HP15c. That code was crammed with fantastic mathematical routines that allowed that calculator to perform matrix operations, integrals and solve formulae. Most of the HP15c that have not been physically battered are still working. And talk about battery life. Around 8 pm I receive a 10% battery left message from my phone, while the original HP15c could last up to 15 years with the same battery pack. It could not be upgraded, but there was no need to: it was the ultimate pocket calculator, with everything a science practitioner would ever need, short of a true computer.
Wherever I look into the logic of the HP15c, it seems to me one of the best objects ever designed by mankind. It is the ultimate “less is more” machine. Apart from all keys having three written functions on it, some of them have more: the same key has completely different meanings depending on what you’re doing.
It had 10 digit precision, which was state of the art at the moment for a pocket calculator. Nowadays 14 and 16 digit are normal - however, when in reality you may need short accuracy? And it is not so important the number of digits, but the number of accurate digits. And the HP calculators of that era had absolutely first class algorithms. Most were create by Mr. William Kahan. They were created in machine code, so that it could run as fast as possible with the hp15c hardware.
It used a version of the nut processor (the same as the HP41c), running at an speed that appears ridiculous today: 244 kHz (yes, not MHz), and it took several minutes to perform a 8x8 matrix inversion.
The algorithms were so good that, when HP started the project to reissue the HP15c based on new, modern hardware, they decided to use exactly the same firmware, running on an emulation layer on an Atmel ARM-type processor. There have been several problems in adapting the old firmware to the new hardware - the pause function does not work well, and the low battery indicator does not function in time - at least for me.
But apart from that, it is as good as the original - just 100 times faster!
It is a pity that there are no more stocks of either one!
Leaving for holiday in a beach location. Still some work to do, but having been warned by “home management” that there are lots of things to pack from our children, and therefore I need to “travel light”.
Hearing “travel light” from my wife is at least surprising: she does not know what “travel light” means. My bags will will always be half of her’s - but let’s do the exercise anyway.
Which for the sake of this blog means which calculator(s) to carry with me. I have already decided not to take with me the HP41CL, although is the one I like best. It would be my solitary island choice - but not for a summer travel to a sun and beach destination. Too risky for its high value. So I have left it in Oslo.
I am taking with me the HP Prime. Now the question is whether I should take with me an HP15c - the HP42s suffers the same precautions as the HP41CL.
The main thing against the HP15c is that I will lose 1 hour with it at the beginning. Why? Because it lost all of the programs I wrote in it (235 steps of ugly code to fill in, with a lot of chances of getting some of it wrong). And I am used to my programs when working. I have it anyway on my Iphone.
And why the HP Prime and not the HP50g (with the promised comparison between both)? First, if we have to travel light, the Prime is thin and light. With its plastic cover, is probably the best protected calculator I ever had. What’s more, I can try to use it to get some of my children interested in maths (through calculators)
Then, why not both? That’s a question that I still need to answer to myself. After all, we’re only leaving next wednesday morning!
For the next month, I have decided to use the HP 42s for my daily financial work. I find it better than the wp34s, due to its 2 line screen, and the fact that you can have the menu on screen much more easily than on the wp34s. Basically, it boils down to the same preference of the HP17bII over the faster and more function rich HP30b: user interface wins every time, and solver is key.
The solver in the HP42s is more cumbersome than on the HP17bII: you need to write a program, with a number of standard lines to define the variables being solved; this takes longer than just writing the algebraic expression, like on the HP17bII. The solver on the HP50g is prepared to take formulae from libraries (good for students), but you can define your own formulae and have them used in the solver.
On a more general level, it is a real pleasure to use a pristine example of a Pioneer series calculator. The keyboard is pure bliss to use - one would like to get the telephone book to add, such is the keying pleasure. Then, the fact of having two lines on the screen (sadly missed on most other Pioneer series calculators) helps a lot in knowing where you are (how many times did you key x><y because you did not know what you had in the y register?)
For the coming weeks, I will be carrying the trusty HP15c in my jacket pocket, and the 42s on the portfolio. Let's see which one wins in the long term.
No, it is not for sale.
Last week I went with my family to New York, and we spent quite a lot of time in museums. I was really impressed by the Metropolitan - for me, miles better than even the Louvre or Prado in Madrid or Paris respectively (leaving aside for a while the Louvre building). But I guess that this is of no interest for a reader of this blog.
What might be of interest is visit to another interesting museum: the MoMA, as well in New York. Among masterpieces of modern art, there was a row of calculators:
It seems that the curator has a bent for Italian design, more specifically for Mario Bellini and the different models of the Divisumma series. This is not Datamath or HPmuseum.net, guys!
Here are the examples above one by one:
The first is an Olivetti 19. If you ask me, I have seen nicer models...
The one below says Underwood 280 - a printing model:
This is the Olivetti Divisumma 18, by Maario Bellini. Observe the seamless rubber keyboard - neat!
And here is the Olivetti Divisumma 28, as well by Mario Bellini, as well with a rubber seamless keyboard.
Notice anything particular about the above calculators? There is no screen! They were just printing machine.
Googling these models comes a little more information about them. They were produced around 1973 (when HP knew already how to produce an HP35!), and they are considered design icons. It is said that the 18 model is exceedingly difficult to find.
Additional investigation shwos that there are more units belonging to MoMA, just no on display. Here are some more pictures:
Paulus Van Leeuwen
A Braun, designed by Dietrich Lubs:
Another Braun, designed by none other than Dieter Rams (who designed a lot of Braun products, from appliances to loudspeakers. Sometime I will write on the Braun LE1, a Rams-ized version of the Quad ESL-57 and that is very sought after. Probably one of the finest speakers ever made)
Newer Olivettis (Logos series - I have seen quite a lot of one of them in my life):
....and lets stop here.
Frankly, I think that the HP12c or 15c would have a place here. Probably as well some of the 22-25 calculators as well, and the 45, 55 and 80.
I do not like any HP in green colour - from the 67 to the 48gx. The 48SX should belong as well to MoMA.
Scientific Calcs Collector’s Pack
In a previous blog page, I told you that many of our customers are calculator collectors as well. Many of them will not stop until completing their collection. And some of them have just decided to start theirs. These packs are for them.
Although HP calculators started by scientific models, they soon completed the line with financial models, distinguished by the Time Value of Money algorithm. Over time, HP has mastered this market, while during the 90’s they lost the scientific market to Texas and others. Since HP is so strong in financial calcs, we started these packs by the financials. You can see them here.
Now is the turn for the scientific. We have assembled the HP standard scientific models, and we have added as well the brand new kid in the block: the wp 34s, based on HP30b hardware.
The units included in this pack are:
There are a very limited number of units of 15c (150 in total, and HP told us they would not produce any more), and as well we have difficulties to locate HP30s over 70 units. So you need to make up your mind!
The true HP connoisseur will know that both the 10s, the Smartcalc300s and HP30s have not 100% HP pedigree; they cannot be used in RPN mode. I have a hard time using a non-RPN calculator, so these units will see limited use on my desk; but as a collector, I need to have them all.
If I had enough supply of classic units, I would include them; but as I seldom have more than one or of each (and often one of them belongs to MY collection), I cannot include them in the pack.
WP34s - a work in progress - Part I
The wp34s is a DIY attempt to replicate (and even improve upon) the HP42s, probably the best RPN calculator ever. Perhaps a little history is in order.
Hewlett-Packard created the programmable calculator market from scratch. The Hp65 first, and then the HP67, were the best calculators of their age, and they were the first to provide mass storage (if I remember correctly, "mass storage" means here a meagre 256 bytes. It is important to grasp the magnitude of this value: it is 1.000.000.000 times smaller than the hard disk of my Mac.
You really needed to be quite compact in your coding to be able to do something with that amount of memory. Well, in any case these were the best available handheld devices of their time and made their way to many spaceflights. There is even a website devoted to the HP calculator space history.
Then, the alphanumeric HP41c arrived.
It was purported as a handheld computer, and it was expandable - there was an enormous amount of peripherals and options, which made it able to do most handheld work of that time. Every one at that time wanted to have one - but not many could. On the other hand, the hp41c was barely pocketable; and the fact that many of its functions had to be addressed from the XEQ button made it less of a calculator. You really had to know the function catalog by heart to be able to take it to the maximum. Even though, and when you mastered, you had to type XEQ Alpha function keys Alpha - not very user-friendly.
In that respect, and in my humbler opinion, it was not as good as a pure calculator as its predecessors. As it was 100% customizable, you could reassign the hidden functions to the keys of your choice, and you even had removable keyboard overlays to show them - but it always left me with a feeling of DIY.
It also was limited in the number of functions and how they were implemented. Matrices and complex numbers come up to mind. Complex in particular were implemented in several HP and third party modules - but these were not compatible among themselves and in many cases led to just 2 stack levels available - not the right thing when you're calculating circuits!
Then the HP15c came. In many areas a step back from the HP41c (not expandable,not alphanumeric, programs stored as key codes, small memory area), it was a true handheld pure calculator. It had matrices and complex numbers done right, all functions were accessible from the front panel, and it could really fit in your shirt pocket.
Still, many of us perceived it as second rate compared with the HP41c. We were comparing real world (hp15c) with potential capabilities when expanded (hp41c). Now, with perspective, I see that the 15c is the better calculator (not the better computer, not the better lab device, but the better calculator)
Then, HP released the HP71c - a true BASIC computer in calculator size, and later the HP28c, which was the first RPL device, and the distant root of both the Pioneer and HP48 families. Still, the HP41c remained the workhorse of the HP calculator family. Adding complexity to devices made them more suitable for class use - not necessarily for professional use.
When confronted with the aging but still bestseller HP41c, HP thought that there should be a software-compatible calculator, based in modern hardware, but still appealing to the HP41c calculator base. It should still be RPN, as opposed to the new series of calculators; and it should offer a better way of reaching the additional functions that the XEQ method.
The answer to that need was the HP42s. It had nearly total software compatibility with the HP41c, but it also had a menu-driven interface for the additional functions, and it integrated complex mathematics and matrices in a seamless way - much better even than with the hp15c; in fact, in my opinion, better than any other calculator, past or future. It had also a way of using the solver function and menus in normal programming. It also allowed techniques for keeping your stack safe when you were using programs.
(It was an irony, though, that the little machine, the zenith of RPN, was itself programmed in SysRPL)
Later, the HP48 family appeared, but it was already a different animal: unlimited stack, different object types, a more complex programming environment (elegant, but too different from the current programming mainstream to be able to grab attention). It was expandable, but not in the same way as the HP41c was. In fact, that kind of expandability has not been reached later. The fact that Agilent was spun off later made impossible to add instrument connectivity.
So that left us without continuation to the top RPN line. Later HP RPN calculators (like the HP32 in their various guises, the HP33s and the HP35s) do not hold a candle to the HP42s)
When HP issued the HP12c running as an emulation in an ARM processor, there was hope for us that HP could release a follow up of the HP42s. However, the kind of screens that could be driven from this processor were not as complex as the dot-matrix one used in the pioneer series like the HP42s. I questioned HP about their likelihood to produce an HP42s, since both the screen and the keyboard live on the HP17bII+. It seems that this is a closed avenue, since the processor used in that series could not seriously run the HP42s firmware; and the whole enterprise would be a significant R&D commitment, not worth in the current scheme of things.
Simultaneously, HP had issued a number of calculators using the SRM processor, and a way to modify it through a serial connector on the back: the HP20b and 30b. These were financial models, and quite powerful at that: as they are programmed in "native" code, and not running an emulator, they are the fastest HP calculators around, even faster than the HP50g in RPL.
It was in this situation that Paul Dale and Walter Bonin set to produce a RPN calculator (to be continued in part II)
If you remember, some time ago we reviewed a project for a clone HP15c calculator made in Switzerland. While the project was extremely interesting, it had the worst possible timing: it coincided with the reissue of the HP15c by Hewlett Packard. With good sense, the creators backed off and re-thought the project, in order to give the user something more than the original did.
I am carrying the HP15c Limited Edition every day with me, and while its size is smaller than most calculators, it is a tight fit in my shirt's pocket, and it fits well in my jacket's; but sometimes I would like it to be smaller.
Enter the HP15cc:
First a pre-production sample
Then a more developed sample:
I look forward to get one. and you?
Today I was to participate as witness on a court case, regarding a termination of an individual from his job. The court case had high uncertainties for both parts, company and worker, so there was a strong interest in settling the issue.
The main discussion point was how to calculate the indemnity according to the Spanish method. Basically, you agree to an indemnity based on a number of days for seniority year or part thereof; topped by a number of salary months. The typical negotiation would be, "we offer you 30 days' salary for each seniority year, with a maximum of 16 months' salary", or "I want 45 days' salary for each seniority year, topped at 42 month's salary" (the legal limit when it is an unfair dismissal).
The problem was to discuss these items at the door of the court without a computer in sight. The fact that calculations should be done by lawyers did not help, either.
Please salute the HP15c Limited Edition.
Due to my job, and in these sad days of restructuring, I had the program in LBL 3. Calculating his seniority by entering the initial and termination dates on two registers, and using the other variables at play, we were able to check at least 20 different options, with the judge waiting for us to agree, until we found one that left both parties "less unsatisfied". They finally agreed.
It was clear for everyone there that it would have not been the case if there was no way to check the hypothesis on the fly, standing, at the court door.
Programmable calculators, 1 - computers 0.
There is room for calcs - sometimes.
Here are the rules:
Horses for courses...
I was thinking this when I packed my things for the summer holiday in Minorca. I had been working hard until Friday afternoon, but I left some things pending for the summer holidays (some financial analysis and the BP for the following year)
I needed to take one or more calculators with me. As many of you, I have a significant collection of calculators, so the choice is not evident at first sight. I have most of the typical calculators in the past or current line-ups - main missing item is 65-67. I hope to fill the gap sometime in the future.
Back to the packing chores. I wanted to learn more in depth the 50g, so there went a manual and the calculator. There are a number of mathematical tools that I wanted to get confident with. The 50g is a very powerful tool for a scholar - either student or professor, but I doubt that an engineer at work uses CAS or most functions at all - despite its value for a math lover. (Does a 50g really have a place in business? Yes, but you need to customise it heavily) Anyway, one for the bag. Just the doubt about taking with me a summer or winter HP50g. Let's be serious and take the winter one.
Then I took my trusty HP15c. Normally is in my jacket pocket. This time I took it with me because I wanted to create a library of programs to add to this -yours- website. You know, you need to be prepared for the rush! But this is not a calculator I'll take with me in shorts, to the beach!
But then, there was the job I still had to do. +-/*-kind of job, with a lot of A%, %T, %, but little more. Instantaneously, I knew I wanted an HP17bII, and no other calculator. Keyboard feel would not be a factor, but screen contrast would - then I'd take the silver HP17bII+.
I understand that this may sound like heresy. I know that most of our readers are engineers, and a calculator without all transcendent functions is a hampered machine; but when I wanted to do number crunching in the beach, this is the calculator I wanted to have by me.
Another possibility would have been the 19bII, but its bigger real estate utilisation (and its fragility when my son drops it to the floor) were some points against.
Your mileage may vary - and that's why I said "horses for courses"