HP-42S Rocks!



#14

I haven't been able to put this little guy down for very long since I got it two weeks ago. This machine is just great! It fixes lots of things that were limiting and/or awkward of the 41C and adds some very nice extras of its own. In the first column are the full alphanumeric capability, easy access to functions, without typing and nearly ideal form factor. In the latter column are items like named variables, matricies with editor, complex numbers, built in solver and numeric integrator and graphics.

Now for the downside. It's such a pity that the 42S represents the high-water mark of keystroke programmable RPN calculators. The machine is full of promise for the future. The I/O is officially limited to printer output, though clever folks discovered you could dump raw memory, ROM and RAM, through the hidden memory editor, out the infrared port. It's still one way, though. The additional memory compared to the 41C is nice, but its tiny compared to other machines of the day. Once again,clever folks have designed a hardware hack to raise the memory to 32K. But then you have to fill it all up one keystroke at a time! The machine cries out for more RAM, and effective mass storage.

The CPU is slow. Nice new features like the alpha graphics are practically unusable for anything beyond small accents because of the speed. The system software isn't perfect either. The matrices are cool, but they are slow too. In an application I'm writing, I use alpha graphics to draw a screen. The graphics data is stored in a 5X7 matrix. The graphics are slow, but if I dump that data out to registers before I run the display update routine, it runs about twice as fast! So there's room for optimization there.

Some annoying limitations from the HP-41C are retained in the 42S. Registers are 56 bits, just like on the 41C, so they can only contain 6 alphanumeric characters. (6x8 = 48 bits + header byte). This make the alpha register size increase to 44 characters less helpful than it might otherwise be. The RPL machines showed the way to stacks and storage that could contain any type of data.

Taking all that into consideration, the 42S is still an awesome machine, all these many years later. HP has refused to release the 42S ROM into the public domain. I take a faint glimmer of hope from this that they might release a "43S" machine someday, with enhanced RAM and processor. Even if they did that in the same way they did the 33S, without changing the system software very much, it would still be a heck of a product. And perhaps some day something like OpenRPN will succeed to the point where it would be realistic to consider what an advanced keystroke programmable RPN machine based on current hardware technology would look like. When that happens, I hope the designers each have access to an HP-42S, to remind them of what the state of the art looked like at its zenith!

Regards,

Howard


#15

The 32KRAM upgrade makes it sing and dance... Thanks to some of our gurus, here. Open one jump, close another, change the RAM chip and there it is!

I successfully upgraded at least five (maybe six) units, one of them is my own unit. About keystroke programming, let's see another side of the upgrade: consider inverting a big complex matrix...

Best regards, and welcome to the HP42S fan club!

Cheers.

Luiz (Brazil)

Edited: 27 Dec 2005, 12:39 a.m.


#16

Thanks for the welcome, Luiz!

The 32K upgrade sounds interesting. Scanning the articles forum here, I don't see one relating to that upgrade. Is information about it available somewhere on the net?

Regards,
Howard


#17

Hi, Howard;

please, have a look at the posts in this thread and you'll find some info about the 32KRAM chips used to upgrade the HP42S.

If you have the skills and the necessary equipments to go ahead with the upgrading AND you still need advice, let us know. There are some pics around (I have at least two, need to search for) with detailled images.

Best regards.

Luiz (Brazil)

#18

If I may be so bold, MoHPC Article # 199 gives many of the essential details. Do ask questions if you have any.

And, I've bought parts by mail from futurlec.com (COMPONENTS tab, Integrated Circuits / Memory / RAMs). You want a 256K bit (32Kx8) low-power CMOS surface mount part. In the barely-remembered past, they sold me quantity 1 at reasonable price. Don't know if they're still so accomodating. But make sure you get the surface mount part with the proper dimensions.

Good Luck!

Edited: 6 Jan 2006, 4:45 p.m.


#19

Thanks for the reply, Paul and Luiz. I decided that soldering tiny surface mount parts was beyond my meager soldering skills. If it was a throwaway machine, I might risk it. But not on a 42S that has the 2X speed capability!


#20

You're welcome.

There's also the question of whether you'll ever enter more than 8K of programming, anyway? (And, should you get 8k filled up, will you then want to sacrifice it all to upgrade the calc's memory?)

The 32K is probably useful for large, computationally-generated arrays, perhaps? But the lack of speedy, reliable electronic I/O severely hampers the upgraded 42s as a programming platform.

I took the whole 32K upgrade issue as more of a technical challenge than anything else. Having accomplished it, I never used my 42S again. In fact, I love the 33s' programming model for the kind of problems I end up playing with on the calculator anyway. For anything more complicated than the 33s can easily handle, I'm off to Java on a computer.

Someone really oughtta implement a one-finger, hunt & peck robotic programming service for the 42s & 33s. It probably wouldn't pay any bills, but it would be a fun challenge and a marvelous contribution to the community!


#21

LOL!

What I'd like to see is a beowulf cluster of HP-41s. 8)

And I've nearly filled the memory of the 42S up with a 1500 line program to play Yahtzee, with bells and whistles. I also sacrificed it (but not before saving it with INPRT) in order to download the ROM so I could use it with EMU42. Your points are well taken, though. I'm not using the machine for real work, but as part of my quest to understand how the architectures of HP handheld calculators evolved over the years. I'm positive nobody would want to key my program into a real 42S, memory enhanced or otherwise. People might like to look at it under EMU42 or Free42, however. It does run in both environments.

Regards,
Howard

Edited: 6 Jan 2006, 11:27 p.m.

#22

Hi, Howard --

Very astute observations of yours, regarding the HP-42S. I agree with everything you said, save for one item.

I'm sure that I saw one for the first time (long with the HP-32S and/or HP-32SII) in 1991 at a H-P calculator promotion at a college bookstore. However, I was quite pleased with my HP-15C from 1983, and was a very tight-spender at the time, due to circumstances. (It's probably just as well that I didn't buy one, because I would have spent valuable time learning it.)

I rediscovered the HP-42S in 2002 when participating in this forum, and got a 1989 model that summer through eBay from a somewhat-disgruntled former H-P employee. (Note: the ROM version on the 1988-1990 flat-screen models is different from the later ROM versions, allowing the user to run a program that will change settings to double the processing speed, until the "EXIT" button is pressed.)

It is true that only six alphanumeric characters can be stored in the numbered storage registers or stack registers. However, a numbered register dimensioned for real (not complex) values is 8 bytes (64 bits), because that is the size of the binary-coded decimal floating-point numbers used on all Pioneer-series models.

I/O would have been a very nice thing to have. I wonder if the trim form factor and the expensive, limited-capacity 1.5V cells could have supported I/O adequately. My opinion is that I/O would have been a pre-requisite for H-P to install 32kB of RAM, for issues of practicality and not allowing the user to overextend himself. (Consider a large library of keystroked programs lost, due to lack of backup capability...)

There are ways to utilize large RAM for purposes other than storing programs or data -- for example, programs that create large matrices for calculation of sensitivity or statistics. However, no built-in HP-42S functions do these things.

Regards,

-- KS


#23

My 42S has a serial number starting with '29' indicating a 1989 manufacture date. What was that magic command to get 2X speed?

Thanks for the correction on the register size. I assumed that since a register could store only 48 bits of text, that the width was actually 56 bits like the 41C. Knowing there are actually 8 more bits makes the limitation even harder to swallow. (Although I'm not sure 7 or even 8 characters would be a huge improvement, practically.)

I agree that full-featured I/O would probably have seriously compromised the very nice form factor and the price. I also agree that 32K would be risky to maintain on battery backup alone. But part of what I meant by saying the machine was "full of promise" was the fact that those things could be provided today within the form factor and relatively cheaply. Two SD slots could be stacked in the thickness of a 42S, and all you would need would be one to provide huge quantities of storage, and the capability of loading programs and data from a PC.

Regarding the fact that none of the HP-42S built-ins provide features for large matrix applications, I'd expect a modern follow-on to the 42S to have a fairly extensive math library that would include things like that. The RPL machines are the benchmark there, once again. They seem to set a very high standard in breadth and quality for calculation support libraries, though I am by no means expert enough in that area to judge. But along with a modern processor would come a nice, large, modern, flat memory model. That in turn would allow very large quantities of ROM as well as RAM. What to fill the former with ought to be an interesting discussion.

I'm really hoping this will turn out, some day, not to have been merely an exercise in wishful thinking. If you look at what the mechanical engineering people are doing with small, relatively inexpensive prototyping factories, then we could be on the verge of an era of small-scale, low-cost manufacturing. Even without breakthroughs in that area, projects like Open RPN are trying to deliver the goods today. Here's wishing them all the success they can handle! 8)

Regards,
Howard


#24

HP-42S fast mode under program control

Unfortunately, it does not work on my HP-42S :-(

Regards, Juergen


#25

Thanks, Juergen!

Sorry to hear it doesn't work for you. It does work on my machine, so I'm grateful to you and to Karl for the tip.

Regards,
Howard

#26

I cannot agree more. I had a 42s throughout most of college. It handled everything I needed it to for my engineering degree.

The downside: Other people like them too, enough to steal them

Alas, I no longer have one (for over eight years now). But I still remember how much better it was than anything else or since. I also vote for HP to release a new model. Here's hoping they read this forum.


Possibly Related Threads...
Thread Author Replies Views Last Post
  42s questions and 42s vs 35s snaggs 13 1,076 09-19-2011, 02:44 AM
Last Post: snaggs

Forum Jump: