the biggest flaw of the 35s: lack of i/o



#2

With 32K of memory, things would be really bad if, say, the batteries fell off and I had to rekey everything.

How much can a mini-MMC slot or whatever it is that mobile phones use these days cost? Probably less than the silly zip case!

Still, I'm very happy that there exists again a usable RPN calculator.

/ji


#3

Can't help with the micro-SD addition, but give CalcPro a call (800-677-7001) and order the 33s slip case (about $11 + shipping); it's much more convenient than the zipper job.

Edited: 2 Aug 2007, 2:32 p.m.

#4

I expect the rationale for not including a memory card slot includes many factors. Technical reasons include:

1) Not enough ROM space in the microcontroller to implement the FAT file system. (They could use the next larger version of the GeneralPlus chip, but it would cost more.)

2) Not enough RAM in the microcontroller to implement the FAT file system. A typical implementation of FAT, even if only one file will be open at a time, requires at least two sector's worth of buffer space (1024 bytes). The GeneralPlus chips don't have much RAM, even in the variants with lots of ROM. Off-chip RAM could be used, but it would take away from the RAM available to the user.

3) A few more I/O pins would be needed, and might not be available.

4) A regulated power supply would be required. The 33s and 35s run directly from the unregulated battery voltage, but the operating voltage range of the battery is not within spec for any of the SD and MMC variants.

5) Battery life would suffer unless components to switch power to the memory card were added, e.g., a PFET. Just using the enable input of a boost regulator isn't good enough, because it will still pass unregulated battery voltage through when the regulator is disabled.

6) Additional ESD protection components would probably be required.

7) The memory card socket itself isn't exactly inexpensive (possibly over $1.)

Non-technical reasons:

8) They already make other calculators with memory expansion and I/O. The 35s is intended to be a low-end programmable, not high-end.

9) If it had expansion capability, it probably wouldn't be allowed on tests.

Adding a memory card slot might increase the BOM cost by as much as $5, which might translate to a $20 increase at the retail level.


#5

and?

#6

Several people have explained reasons why this is probably not in the cards.

1 ) Testing. One of the biggest markets for the 33s (and presumably the 35s) is the testing market. Adding I/O to this level machine will promptly put it on the Not Allowed list, substantially lowering sales. For every one of us who would buy a machine because it had I/O, I'd expect 10 lost sales because of it not being on a test list.

2 ) Cost of the parts. Yes, it might only cost $1, $2, $3 (have no idea) to put the I/O into the physical machine, but multiply that by the number of units made for HP's incremental cost then ask yourself how many incremental sales would occur to offset the cost (see item 1 in this list). That's a very tough sell to whomever reviews these decisions.

3 ) Architectural legacy. This is a machine based off the HP32S line (HP32s, HP32s2, HP33s). That line does not have an OS designed in any way for I/O. Makes it much more involved for the limited resources to design such a machine from scratch. So, what other projects might HP have in the works that they now have to delay to add I/O to the 35s OS and lower sales of the unit at the same time?

I just don't see this happening.

If we're going to get any type of 42s style machine with I/O, it will take a lot of doing and convincing.

Note that this does not make me happy, but I believe reality is usually important to understand. :-)


#7

The SPLB31A (http://w3.sunplus.com/ShowFeature.asp?body=SPLB31A) has a UART. A serial port is possible and may be easier than SD to implement. Without a case mod Bluetooth would be a possibility. If implemented as a serial provider only (i.e. PC -> 35s only), then this removes the possibility of 35s <-> 35s for those concerned about cheating. Store the BT PIN is a register, connect with PC, dump/restore state.

#8

Lack of I/O isn't a flaw at all.

This is the 1st calculator (rather than handheld computer) that HP has designed since the 27S/42s that has essentially *unlimited* memory within its paradigm of operation. That paradigm being programming and solving power for Ad-hoc programs.

I/O makes for a completely different paradigm.

In the 32sii, you had so little memory that equations would quickly use up the memory, so you'd have to erase ad-hocs or useful regular routines in order to do anything.

In the 35s you don't have to worry about that.

In the 33s, the equations were limited to 256 characters. In the 35s , there is no limit.

I/O is available in other (RPL) machines. No need to have it in a calculator.

The biggest flaws are doubtless:

1. No easy decomposition/composition of complex numbers
2. difficult to use base system
3. cos near 90 degrees
4. Others?

I wouldn't call lack of I/O a flaw at all.


#9

"I/O is available in other (RPL) machines."

RPL based machines have too much "sweat equity" (learning curve) to attract users longing for I/O. TI and Casio have made this standard fare for their medium and high end calculators for some time. TI even has USB OTG on their TI 84 & 89.

The HP 50G attempt to provide I/O is quite simply inept. To blatantly advertise that port as "RS-232 capable" is unexcusable.

"No need to have it in a calculator"

This is news to all the HP 71 & HP 41 users who spent years writing programs and developing HP-IL portable systems for remote field work. These devices were referred to as calculators, but in many instances they were used as portable battery operated controllers. This capability greatly added to HP and RPN's widespread popularity and commercial success. The next generation replacement of the HP 41, the HP 48, was wholly unsuited to a controller task since it possesed only one serial port, which meant the controller could only communicate with one device at a time.

I think an RPN (not RPL) calculator with I/O and some controller capability would be widely appreciated by the community. I know I have been waiting for the last 20 years to find something more capable than my HP 41 and HP-IL. I have an HP 35s and I know for sure that it will not replace it.


Edited: 2 Aug 2007, 6:45 p.m.


#10

Quote:
This is news to all the HP 71 & HP 41 users who spent years writing programs and developing HP-IL portable systems for remote field work. These devices were referred to as calculators, but in many instances they were used as portable battery operated controllers. This capability greatly added to HP and RPN's widespread popularity and commercial success. The next generation replacement of the HP 41, the HP 48, was wholly unsuited to a controller task since it possesed only one serial port, which meant the controller could only communicate with one device at a time.

It may be wholly unsuited to a controller task in terms of connecting DIRECTLY to multiple devices, but it's certainly more than capable of connecting to a device that acts as an HP-IL controller module.

The serial interface has been a defacto interface to just about any peripheral on the planet for 20-30 years. I think most users are much happier that they could connect their HP 48s to their PCs and heavens knows what else than having the ability to connect to HP-IL.

If HP-IL connectivity was that important to the community, a serial <-> HP-IL adapter would have shown up filling the gap. The power, capacity, and capability of the 48g series dwarfs the 41 by pretty much all measures, so I'd think there'd be a lot of drive to help the platform so folks could upgrade from the 41.

I mean, heck, today if you wanted a serial interface for devices, I think most folks wouldn't mind access to I2C anyways. And now, instead, we have USB.

The USB handles the 99.99% of cases where people want I/O on the calculator: saving and loading data and files from host computers. For the other .01% of folks, having the, albeit "weak", serial port is a cheap solution for custom interfacing. Interface direct to the signals and strengths it provides, or amplify them to make it compatible with the RS-232 spec. Boosting the signal is not difficult and I think it seems clear the compromise of having the weakened serial port is better than not having any serial, save USB, at all.


#11

Quote:
If HP-IL connectivity was that important to the community, a serial <-> HP-IL adapter would have shown up filling the gap.

It's been available for over 30 years: 82164A.

What would be cool is a USB<>HP/IL interface.

Regards,
Howard


#12

Quote:
Quote:
If HP-IL connectivity was that important to the community, a serial <-> HP-IL adapter would have shown up filling the gap.
It's been available for over 30 years: 82164A.

I got the FSA164A which is almost identical but has two RS-232 ports on it and, for those who didn't mind the extra cost, could have up to 8 RS-232 ports on it.

To comment on several other posts above and below this one:
I also got the HP82169A HPIL-to-HPIB (IEEE488) interface converter. HPIL was basically a serial implementation of HPIB, and with the 82169A, the thousands, yes, thousands, of different models of lab equipment instrumentation available with HPIB interfaces appeared to an HPIL controller to be on the loop. The commands were the same.

As the entire engineering department of a tiny six-person company, I set up our first automated production testing using the HP-41cx as the controller, having a half dozen pieces of equipment connected, most of it rented, and running a 20-page program. At the time, I and my wife lived in a small apartment, and, although the test equipment at work was large and heavy and AC-powered, it was very practical for me to take just the 41 back and forth between home and work in my attache case and work on programs or other things with it, without having to have a bigger computer at home. It was also much quicker to get a test set-up going with the 41 than it was with a PC anyway.

As the company quickly grew, the 41 was replaced with an HP-9000-series 68000-based computer; but since the controller spent time waiting for filters to settle and instruments to respond, the bigger computer's speed was less than double that of the 41, the boot-up and load time were very long (unlike the instant-on 41), and it took longer to re-write the test program for the bigger computer than it had taken for the 41.

My lament about USB is that contrary to its name, it is not a bus, and one port can only go to one device unless you use external hubs. That's pretty mickey-mouse if you ask me. Sure it's a lot faster than HPIL, but the HPIL idea could have been taken to much greater speeds too if allowed to grow with technology.


#13

How many times did someone look at your test station, drop their jaw, form a puzzeled look on their face, and say "I didn't know a calculator could do that!".

Edited: 5 Aug 2007, 4:02 p.m.


#14

Quote:
How many times did someone look at your test station, drop their jaw, form a puzzeled look on their face, and say "I didn't know a calculator could do that!".
Back then (20+ years ago), basically everyone, including the sales engineers who would come from the instrumentation companies like HP (who were not very aware of what was going on in the calculator division), Wavetek, Bruel & Kjaer, Cytek, etc..

#15

Quote:

Back then (20+ years ago), basically everyone, including the sales engineers who would come from the instrumentation companies like HP (who were not very aware of what was going on in the calculator division), Wavetek, Bruel & Kjaer, Cytek, etc..


Yes, and I think this made some of the other divisions unhappy. I am convinced that in the case of HP-IL that little Corvalis group was a victim of its own success. I think they stepped out of the bounds of their "charter" and cut into the sales of desktops, which probably had a higher profit margin.

If HP wanted to, they could have easily integrated HP-IL on the HP 48G, but instead they "dumbed down" the I/O, stopped the sales of HP-IL peripherals and discontinued the battery operated instrument line.

The proliferation of those HP 41 controllers were like mushrooms in the desktop garden. I think the demise of HP-IL was deliberate and intentional.

Edited: 6 Aug 2007, 1:06 p.m.

#16

"If HP-IL connectivity was that important to the community, a serial <-> HP-IL adapter would have shown up filling the gap."

Not exactly true. The HP-IL story is kind of ugly. If you recall the HP 41 and HP-IL devices ceased prodution shortly after the introduction of the HP 48. I believe this was a consious and perhaps prudent decision by HP to move their user base over to the new paradigm product. Here is the caveat you may not realize. The HP-IL bus was proprietary. HP refused to support it ( stopped production), refused to license it (force customers to adopt our new product), and refused to put it in the public domain(at least at the time. I actually don't know the current status). If you were accustomed to the advantage of controlling up to 31 devices on a bus, HP's answer was - Tough Luck! Needless to say the only survivors of that market segment were the surveyers who only need to talk to a one device (a theodilite) at a time.

HP-IL would of course be inadequate in todays enviroment. It is however, unfortunate that the interface was abandoned. I am sure if it had been allowed to mature and improved it would still have many adherents today.

Edited: 3 Aug 2007, 1:29 p.m.


#17

Perhaps HP also saw at the time that the vast majority of the HP-IL market would be better served by the then emerging microcomputers and decided that it would be best to avoid going head to head with that foe.


#18

Quote:
Perhaps HP also saw at the time that the vast majority of the HP-IL market would be better served by the then emerging microcomputers and decided that it would be best to avoid going head to head with that foe.

Perhaps, I don't know. I would argue that it suffered a premature unnatural death rather than a natural death. I could find no evidence of lagging sales or excess inventory in the supply channels. I recall that, in fact, parts were back ordered most of the time. This was also about the same time decisions about operations moving to Singapore were probably being made, and I am guessing that HP-IL products and support did not make the transition cut. I suppose some tough decisions had to be made at that juncture.

Were microcomputers a threat? Certainly.

Was the market place better served? Well, served at least, but better served is perhaps in doubt.

Which begs the question "Does the market place know what it wants?"
Well no, not always. I don't think the market place anticipated or wanted things like the HP 35, or HP-IL, until they saw it. I think both ideas were so new and bold that neither one could have survived a focus group or marketing survey. If the published history is correct it was a simple matter of someone just saying " build it".

Sorry, please excuse the ramblings. I am just an old guy wishing that things were like they used to be.


#19

Quote:

"Does the market place know what it wants?"


The answer is of course a resounding NO! Markets only exist after ideas, not before them....

#20

Somewhere along about this time, HP developed the HP Interface Bus (HPIB) for controlling the broad line of HP scientific instruments (counters, spectrum analyzers, etc.). This was pushed on the community as a broader standard and became IEEE 488. It was PC compatible (you could buy interface cards, or talk to an HP interface box with your RS232 serial PC port - which all PCs had at this epoch), and you could also control many, many instruments because each had its own address (usually set with rocker switches on the instrument).

This was for serious control and data gathering, with data amounts and speeds way beyond what you would/could want to do with a '41 and HPIL. And the "product" was in your PC when you wanted to analyse it. I have no idea if there was any kind of HPIL<->HPIB interface hardware.


#21

Yes there was an HPIB to HPIL interface. I happen to have 2 of them. You could think of an HPIL bus as a slower cousin to HPIB, but with a ring topology. The commands went from the controller around the ring, and then back to the controller. I have used an HP-41 as an HPIB controller, but it is not appropriate in most situations. HPIB instruments usually require AC power. HP-IL was primarily focused on battery powered instruments.

If you are unfamiliar with the bus or more interested see

http://en.wikipedia.org/wiki/HP-IL

It was quite sophisticated and well supported with instuments at one time.

#22

But in truth, HP-IL really wasn't abandoned. The names have changed, the protocol has shifted, but the functionality remains and had gotten much, much, better.

I don't think anyone praised HP-IL on its protocol, but rather it's format and functionality. Like the ability to control several devices on a single bus.

If anything, today, they'd lament one of it's physical aspects: the loop. Having to loop the equipment would probably be off putting to some folks, much like we used to have to terminate ethernet cables or tap those thick nasty cables at specific intervals. Compared to todays RJ-45 jacks and twisted pair, noone is looking back and saying "them were the days".

But serial protocols flourish today, and are the dominant form of communication by a wide margin. And, whoo boy, are they fast. Offering all of what HP-IL offered and more.

So, HP-IL itself may be dead, but it's spirit lives on and shines bright.

Calculators are spectacularly challenged today. The amount of power available today is astounding, but it's the form factor of the calculator that keeps us coming back.

I mean, look at the iPhone. That thing has 700MB of software on it. 700MB! For a PHONE!

But it can't do what this 35s does, I/O or no I/O. It has no staying power, needing to be recharged often. And it's got a lousy form factor compared to the 35s. No doubt someone will come out with some kind of better calculator application for it, but no one will use it. Anyone who's serious about using a calculator for anything more than calculating a dinner check will get a real calculator. The 12C sells for a reason.

And you won't see the vertical applications for the iPhone like you do calculators, particularly HP calculators. Notably solution packs, or things like surveying, etc. Again, the platform doesn't work. Do you see workers taking the iPhone in to the field? Middle of a dusty plain, trying to make it function with gloves on? Nope. No way. Buttons are good. Calculators are strong, they're durable. They're also cheap.

Folks will be using their 35s 10 years from now. The iPhone will be dead and in the landfill by then (and, well, maybe the Smithsonian, but that's a different topic...).

Today there is a lot motivation that folks feel like they need to cram everything they can in to the "hand held computing device". I think the high end calculators of today really take it about as far as it can go. They'll never stop surprising me of course. They can always make them faster. Add more capacity, etc. The motivation is there because it's so easy. CPU horsepower is practically free today.

I recall when the Game Boy Advanced came out. They wanted to be able to play original Game Boy games on it. But the GBA was an ARM chip, and the GB was a Z80. Did they give up? No. Did they emulate a Z80? No. They bundled a Z80 core into the machine. What's a Z80, 10,000 transistors? Pah! Childs play today. Drag and Drop in a VLOG editor. And that's cheaper and faster than the FLASH space for a Z80 emulator.

Calculators tho, they have to achieve balance. I look at some of the Casios at the local Office Mart. Solar powered, immortal, large LCDs, a zillion functions, and -- wait for it -- and ONE memory register. ONE! Uno! M+, M-, MR, MC. WHAT??? You can perform quantum physics on this thing but only save one value? What are they thinking?

Well, for one, they're $10. For another, that's what their market does. They do calculating, not so much computing. Kids doing homework. Sold by the pallet with free pencils.

The 35s is all about balance. As someone else mentioned, it's got as much memory as it needs. It simply, as architected, can't use anymore. All of the machine limits hit at 32K. And that memory is designed more so for macros, really, than applications. Shortcuts folks incrementally develop for the machine as they use it. For folks who go "Boy this calculation is a bear, it would be nice to not have to do it again", and sure enough you really don't have too (well, once more as you type it back in). Keystroke programming is the classic "watch me" macro recording process.

But kids doing homework, they're not redoing calculations. They do "work" with a calculator as much as they do exercises. And most of them will praise the day that they don't have to do ANY calculation again, much less the SAME calcuation again. So, programming is much less important.

I do agree that it would be a shame to have a lot of work in the machine and lose it to a battery failure. It would have probably been nice to have a back up button battery. I imagine it has SOME lifespan without the batteries (I know my 15C is like that, to survice battery changes), but for safety, it would be nice to have some back up like that.

32K of RAM is a not really a lot to type back in, but it's a lot to lose.

I love my 48gx and 49g. Just incredible machines. I love the 15C, it's form factor is, bar none, the finest ever. It's handiness and perfect "thumbabiity". Unfortunately, the 15C form factor can't handle the bigger LCD, or take more buttons (even tho the 15C had 3 less than the 42S). So, they just can't seem to cram as much as is necessary (apparently) today in to a 15C form factor.

I think form factor of the 35s is pretty good, and it's lack of I/O is not an issue. It's a calculator in all its glory (though the infrared printer would have been nice).

Its hard today to keep a calculator a calculator. Designers can pretty much do anything they want with the days technology, but go too far and you end up with something that's not a calculator. It loses its focus and comprimises what a calculator is used for by the people who use them. And that's a crime against both the users and the calculator.

The 35s is a calculator, and boy does it seem like a really nice one.


#23

Nice post.

#24

Quote:
I wouldn't call lack of I/O a flaw at all.

Opinions differ, obviously. I don't think the lack of I/O is a flaw in the HP-35S per se. That's because the calculator is aimed at a specific market in which I/O is forbidden. But I pine for a modern, RPN calculator without the obvious compromises that constrain the HP-35S. It's no doubt a vain hope, but I'd like to see a flagship scientific in the niche the HP-41 used to occupy, at the top of the product line. In that machine, lack of I/O would definitely be a flaw.

Regards,
Howard


Possibly Related Threads...
Thread Author Replies Views Last Post
  HP Prime - User manual lack bluesun08 6 383 11-08-2013, 05:38 PM
Last Post: bluesun08
  HP PRIME Home settings var lack Damien 0 120 10-26-2013, 04:25 AM
Last Post: Damien
  [HP-Prime] Lack of user soft keys Andy B (Australia) 1 163 08-17-2013, 11:00 AM
Last Post: Tim Wessman
  The biggest collection update yet Keith Midson 2 266 10-31-2012, 12:16 PM
Last Post: Matt Agajanian
  Almost a Red-Dot. It seems only to lack the hole! Ignazio Cara (Italy) 9 324 02-22-2010, 07:30 PM
Last Post: tim m.
  A small cosmetic flaw Walter B 3 161 08-28-2007, 08:43 PM
Last Post: Trent Moseley
  HP 17BII+ Flaw - Revisited Iracildo Luis dos Santos 18 547 05-10-2006, 06:28 PM
Last Post: Chris Dean
  The arc tan flaw in early HP35 ROM is elucidated Jacques Laporte 2 144 04-10-2006, 08:39 AM
Last Post: Jacques Laporte
  HP 17BII+ Flaw Iracildo Luis dos Santos 6 225 04-04-2006, 06:37 AM
Last Post: Iracildo Luis dos Santos
  HP48SX Lack of Laplace transforms David Houston Thompson 11 333 10-03-2005, 05:36 PM
Last Post: I, Claudius

Forum Jump: