HP 35s missing keystrokes

[r. pienne]

There was a thread back in early 2008 on the old forum where several people complained about missing keystrokes on their 35s: http://www.hpmuseum.org/cgi-sys/cgiwrap/...ead=131932 , but I have not been able to find any complaints after that, which suggests that the problem might have affected only early batches.

However, I now find that recently the C (clear) key on my 35s is getting seriously unreliable, and I often have to press it 2 or 3 times before it registers. The machine is less than 3 years old, and lightly used.

I'm wondering does anybody else have this problem with their 35s? Does anybody know of any fix for it?

Cheers.

R.

[jebem]

My 35S s/n CNA72800051 still works fine, but I use it just occasionally. However, it sometimes do not power ON at first try, but I do not see this as a real problem. Usually I just replace the batteries with fresh new ones and the issue goes away.

I think the "C/ON/OFF" key is a special function one, so the issue can be related to other causes other than a key mechanical failure.

Is the key issue happening on any operation, or is it happening to specific operations, like when cancelling a function or stopping a long running program, or Powering ON or OFF?
Other members have reported that when the "busy" annunciation is on, keys are missed as well.

I would make sure the batteries are fresh, as some issues can be triggered by poor quality batteries. If you have access to a multimeter, remove the cells, and make sure the reading values are identical on both cells and above 3 Volt.

Also, the self-test can help to check the keyboard (C+XEQ).
The calculator may lose the memory contents when running the self-test.
http://www.rskey.org/~mwsebastian/selfte....htm#hp35s

[r. pienne]

Thanks a lot, José.

The problem happens in every situation. The batteries are only about a month old, and measure 2.98 and 2.97 volts. I don't fancy losing 600 lines of programs just yet, so I'll skip the self-test unless it's needed.

After playing around with it a bit again, it seems that if the C key is pressed a bit harder it always works, so maybe it is a mechanical thing and I just need to get used to using a bit more force on that key. Ah well.

R.

[MarkHaysHarris777]

(04-26-2014 10:29 PM)r. pienne Wrote:  After playing around with it a bit again, it seems that if the C key is pressed a bit harder it always works, so maybe it is a mechanical thing and I just need to get used to using a bit more force on that key. Ah well.

The problem you are describing is a classic HP keyboard problem that goes back to the early key switches of the Wang 700, and others back in the late 60s to early 70s and beyond. The mechanical keys on the HP are patented (maybe its expired now), but they have two distinct problems that are difficult to eliminate. The first is what you are experiencing, and the reason 99.9% of all calculators manufactured today us silicone bubble button technology... the little mechanical metal contacts oxidize over time, more under some extreme conditions, and stop registering. The good news here is that the key boards may be disassembled and cleaned (lots of patience, and lots of time). The second problem these keys have is contact bouncing. Sometimes they register more than once (or not at all) because of contact bounce (electrical noise created because the switch 'makes' more than once very quickly even though only pressed once. Contact debouncing circuits must be employed for these kinds of input switches (HP calcs included) but sometimes corners get cut!

The HP35s is assembled with screws, so the keyboard can be removed and cleaned if you are very careful. Sometimes you can forgo that if you simply press the key more directly (not violently) and firmly. My 35s is only two months old (by serial number) and it is already starting to exhibit this behavior. Sooner or later I will need to disassemble it and clean the contacts.

If your environment is very dusty (very fine particulates in a manufacturing area, or some other venues with fine dust) or if you or those around you smoke, you will see problems quicker which will affect more keys sooner. Cigarette smoke is notorious for ruining the contacts of all kinds of electronic switches, and the keys of your HP are no exception. Keep it clean, keep it covered, and don't smoke around it. Be thankful that (unlike the fx115 or the 36x pro) the keyboard is not plastic welded shut. The 35s uses about 28 little screws (its a pain, but its doable).

Cheers

[Thomas Radtke]

(04-25-2014 03:34 PM)r. pienne Wrote:  There was a thread back in early 2008 on the old forum where several people complained about missing keystrokes on their 35s: http://www.hpmuseum.org/cgi-sys/cgiwrap/...ead=131932 , but I have not been able to find any complaints after that, which suggests that the problem might have affected only early batches.

Doubt it. The problem appears when typing faster than the keyboard matrix can be read. Maybe it has something to do with multiplexing IOs or similar.

You don't hear about it anymore as, after much enthusiasm in 2007, the 35s is mostly forgotten among professionals after all those bugs were discovered.

But yes, singles keys may give up the ghost at any time. My cursor down key, especially problematic since it isn't hinged in favour of this silly joypad arrangement, needs a firm press to become registered.

[MarkHaysHarris777]

(01-23-2015 04:48 PM)Thomas Radtke Wrote:  You don't hear about it anymore as, after much enthusiasm in 2007, the 35s is mostly forgotten among professionals after all those bugs were discovered.

Actually, most of those 'bugs' aren't bugs; the HP35s still has huge sales, and the unit is 'still' manufactured... mine was manufactured in late November of 2014. I have spent many joyful hours with it over the past couple of weeks and have found that it really isn't that buggy. I need to do a little more research, but I've found that 'most' of the bug-list is either a matter of extremes testing or misunderstanding about either the nature of the 8502 processor and its interrupt mechanism, or the nature of stored program organization in the unit (the rules governing recalculation of entry points and the limits of program execution) or a misunderstanding of the design goals of the 35s. The unit has a couple of minor glitches, one gotcha (related to the 8502 interrupt mechanism) and 1 or two minor annoyances which have been corrected by the WP34s architecture.

I am planning to write a response (for new users to the 35s) to the bug-list on this site; not in an argumentative way of course, nor to generate any flame-bait comments... nor anything like that... just to give a more fair and balanced assessment of the HP35s as a professional tool for working engineers. Also because more students are using the 35s than ever before for the FE exam. They deserve to have a balanced review that explains the minor glitches, warns about the gotcha, and gives the new operators a confidence in the excellence of this commercial product for real work.

I'm really excited about the product, frankly. The more I use it the more I like it; and you know what? I'm not running into any show-stopping bugs in the firmware. The unit works within its design constraints, does well 99.98 percent of what it was designed to accomplish (does it very well even) and is really the best engineered, best assembled, best 'usability' device I have seen in the last 20 years. (I'm not including my TI 89Titanium).

I mean seriously folks, if you read the bug-list from this forum you might come away thinking WTH? (let alone purchasing the unit for real work) I know, I thought that for a while. Then I started talking with folks who are actually using it for real work and discovered that the unit really doesn't have all those bugs, and that the bug-list is really more of an annoyance complaint to HP. When it comes right down to it, the HP35s works very well, is well made, will last for a long time, and will be vindicated in the end. Time will tell.

Cheers

[Thomas Radtke]

(01-24-2015 07:14 AM)MarkHaysHarris777 Wrote:  Then I started talking with folks who are actually using it for real work and discovered that the unit really doesn't have all those bugs, and that the bug-list is really more of an annoyance complaint to HP.

Maybe. It came when no one believed there would be an HP calculator with a large enter key again. The first pictures (the 35s as toy for Tims baby IIRC) revealed an old style keyboard and nice colors, and there was a video about the mechanical keyboard design. The 35s addressed everything users critizised about the 33s. You imagine how well it was received. Surely I ordered one as soon as possible. Then came one bug report after another. No complete set of composing/decomposing functionality for complex numbers, making e.g. R<>P conversions difficult, lockups, and all the other bugs were discovered. Of course the owners were greatly disapointed.

The 35s and its development team (the design was truely a great job!) deserved a bugfixed firmware and I bet TW et al. tried to get one but failed at some point.

[MarkHaysHarris777]

(01-24-2015 10:19 AM)Thomas Radtke Wrote:  I ordered one as soon as possible. Then came one bug report after another. No complete set of composing/decomposing functionality for complex numbers, making e.g. R<>P conversions difficult, lockups, and all the other bugs were discovered. Of course the owners were greatly disappointed.

Nice to meet you Thomas. The R<->P conversion thing is one of my pet peeves (even a hobby horse, so I'll spare you as much as I can). I collect R<->P conversion routines. It's fun. My interest is electrical engineering (amateur radio reactance circuit | tank circuit design). The HP35s easily converts and handles decomposition of coordinate systems (for use in the complex number plane) with just a few steps of programming which is (except for the overall concept) trivial in nature. So, its a little like purchasing a classic 1957 two-tone Chevy coupe only to find out that the guy who restored that gem forgot to wire the cigarette lighter. So you have to wire it yourself; even though, you yourself don't smoke. It just bugs you that it doesn't meet your expectations. hey, just wire the lighter and enjoy the ride!

Let me illustrate the programming I'm talking about. Sometime a little later (I'll key it in after I post this, to the general software library) I'll upload my favorite R<>P routine. It makes no claims... its just one example... takes up almost no space, is almost trivial, WORKS, and is easy. i use it from time to time.

But consider this (and I'm sure this has been discussed here at some point before my time):

3, 4, 5 right triangle in the complex plane...

[3] [i] [4] [ABS] There is your vector length (R)

[4] [i] [3] [ABS] ... there it is again

[4] [i] [3] [ARG] ... there is theta, assuming x=4 , y=3

[3] [i] [4] [ARG] ... there is the other theta assuming x=3 , y=4

And that's without any programming... but if you add those to an R<>P routine then whalla.

My point here is mainly this: We have a commercial tool that works, is easy to program, is fun to program, and can thereby be customized and tailored to suit individual preferences as well as needs, and can be relied upon to work for years to come: what's the problem?

Not everyone are capable programmers. Some of the concepts (trivial to some of us) are a royal pain in the arse to some others. I have met some wonderful aerospace engineers who can't program pouring water out of a boot even though the op codes are engraved on the heel... and I'm not picking on them either! Its just that the HP35s is designed with the idea that the programer knows what they are doing. If the programmer is clueless, can't RTFM, or simply doesn't know a tight loop from a post hole, well.... its not HPs fault, and its not a bug. That's all I'm trying to say.

Cheers,

[Thomas Radtke]

Hi Marcus,

(01-25-2015 08:31 AM)MarkHaysHarris777 Wrote:  My point here is mainly this: We have a commercial tool that works, is easy to program, is fun to program, and can thereby be customized and tailored to suit individual preferences as well as needs, and can be relied upon to work for years to come: what's the problem?

There's probably none if one approaches the 35s as you do. Still, when there were so many glitches to be discovered just by using it in the first few weeks, what do we have to expect in the long run? Can I trust it? Of course, no one knows, and that holds true for other calculators as well. Just recently we discovered new bugs in the 32SII and 15C firmware.

One problem certainly is that the 35s could be much better by addressing the obvious problems, and at least I have my quarrel with HP about why that never happened. But it's not that I'm not using it. Just the opposite, my 35s found its place at my office.

Anyway, your personal insights on the 35s are an interesting read. Always nice to meet someone liking this branch of calculators as much as I do.

BTW, the R<>P conversion programs Thomas mentioned in the other thread are a community effort, well tested and leave Z,T registers intact. I think they're hard-if not impossible-to improve any further :-).

[Jeff O.]

(01-24-2015 07:14 AM)MarkHaysHarris777 Wrote:  Actually, most of those 'bugs' aren't bugs; ...I've found that 'most' of the bug-list is ... misunderstanding about either the nature of the 8502 processor and its interrupt mechanism...

Mark,
I assume the above statement refers to the lockup problem exhibited by the 35s, item no. 15 on the bug list. You have also mentioned "tight loops" in a couple of your posts, again, I believe related to the lockup problem. I certainly know nothing of the nature of the 8502 processor and its interrupt mechanism, but in case I misunderstand, I wanted to point out that it does not seem that the tightness of the loop has any bearing on whether or not the calculator locks up. Soon after the 35s came out, users reported instances of the screen going blank during program execution, with no response to any key presses. The situation could only be remedied by pressing the reset button, which erased all memory. The problem was ultimately found to be caused by using an equation as a prompt to stop the program and wait for you to press R/S to proceed, without following the equation immediately by a PSE or STOP instruction. Actually following with a PSE instruction only pauses briefly to display the equation message. If you want to display an equation as a prompt, stop the program and resume with R/S after keying in input, then you must follow the equation with a R/S. But then you must press R/S twice to get it going again. In any case, the “tightness” of the loop following the equation did not matter, it could be 10 lines or 1000 lines, but if it did not terminate naturally, it stayed in the loop, blank screen, uninterruptible. If a PSE was included after the equation to only pause briefly to display the message, or if the equation was followed by STOP with execution resumed by pressing R/S twice, the exact same loop could be interrupted by R/S or C, and the normal “running” message would be displayed during execution. Like I say, I know nothing about the 8502 processor and its interrupt mechanism (and I don’t really think I should have to in order to keep the calculator from locking up, should I?), but based on the only difference being inclusion or not inclusion of a PSE or STOP instruction, do you think this is really related to something as deep as 8502 interrupts?

(01-24-2015 07:14 AM)MarkHaysHarris777 Wrote:  I am planning to write a response (for new users to the 35s) to the bug-list on this site; not in an argumentative way of course, nor to generate any flame-bait comments... nor anything like that... just to give a more fair and balanced assessment of the HP35s as a professional tool for working engineers...

A fresh look by a new user such as yourself sounds like a good idea. I look forward to your review.

(01-24-2015 07:14 AM)MarkHaysHarris777 Wrote:  I'm really excited about the product, frankly. The more I use it the more I like it;

I can tell, nice to see such excitement about a product released almost 8 years ago. I agree that the 35s is quite useable and the “bugs” or quirks won’t impact most users most of the time. Its most unforgivable sin was that it was so close to being so much better, were it not for the bugs that seemingly could have or should have been caught before the design was almost literally carved in stone. (Carved in silicon, I guess.)

[Craig Thomas]

While I own a 35s, I find I much prefer to use one of my 4 hp8's, or my newly purchased HP41CV (the ideal engineering calculator).

The implementation of various things in the 35s has always struck me as annoyingly cumbersome so I've never used it long enough to feel otherwise. And the battery life!!! Too short, and too expensive for new ones.

I was very lucky to recently get a $50 fully functional HP41CV to replace my long dead favorite machine. I'd buy another in a heartbeat but they aren't turning up that much here, north of Seattle, WA. Love that calculator!

[MarkHaysHarris777]

(01-26-2015 06:48 PM)Jeff O. Wrote:  Its most unforgivable sin was that it was so close to being so much better, were it not for the bugs that seemingly could have or should have been caught before the design was almost literally carved in stone. (Carved in silicon, I guess.)

I agree, to a point. The most unforgivable sin from HP in 2007 was that they chose to use an embedded 8502 core for their 'most powerful non graphing scientific calculator' instead of competing with state-of-the-art technology. The 8502 is the problem, frankly, with most of what gripes folks about the 35s. HP was trying to compete with the Casio and TI student markets (and just can't do it). My FX115ex Plus cost me a whole $13 (on-line). You can purchase the TI 36x Pro for $24 retail and about $17 on-line. HP simply cannot compete in that market; hence, the embedded 8502 from Kinpo Taiwan. That was the sin. But we DO have a relatively inexpensive programmable scientific which is for the most part pretty fabulous and its allowed on the U.S. exams... why keep beating it up because we're mifft at HP, rather than helping students get to know the 35s? And another thing, there might then be an opportunity to introduce them to the WP34s... trust me... getting folks to move from the 35s to the 34s is going to be way better than trying to get them to move to the 34s from the 36x Pro!

We need to clarify the difference between 'cause' and correlation'. Correlation is not causation. The cause of the lock-ups is NOT placing an equation in a loop as a 'pause' for input/ the cause is that the machine entered an infinite loop (and a tight loop to-boot) without access to its software interrupt for scanning the keyboard, and its easily fixed by placing a PSE opcode immediately after the equation prompt (or [R/S] ). Here is another point: Once you have a work-around for a severe bug, its no longer a severe bug... its now an annoyance, or maybe better a 'rule' for new normal operation. Once the community understands the step for recreating a bug (and folks avoid it) what's the problem? I don't use equations to prompt for user input (without a PSE or STOP) and I don't allow my programs to enter an infinite loop (very careful about that) and my machine has never locked up... not even once. So, granted, I have only been playing with this particular processor for a few hundred hours... so I may not have pressed every function finding button and\or scenario yet... but close.

True story---> my son is an engineering student as ISU this year/ NROTC studying CE (and having a blast). The engineering department handed out calculators (yeah, I know the parents paid for them) and guess which one they handed out? Do you think it was an HP anything? Nope. TI 36x Pro. <=== the students are not permitted to use their graphing calcs (83 84 89 50g &c) in most of their classes (for the same reason that NCEES does not allow them either!) So, his calc collection stays in the dorm room and he hauls the 36x Pro around (no RPN, RPL, and NOT programmable!) And he's studying engineering ! What is wrong with that picture?

Again, all I'm really saying is that now that HP knows we're mifft, well, we can rewrite the bug-list to reflect a little more balanced approach to the unit. Yes, its got a couple of glitches, one gotcha, and you'll need to load the communities R<>P and trig utilities, but hey, it works, its fun, its functional, and its allowed on the NCEES FE PE PS exams! ... which means, you need to understand it, but hey, you can trust it and rely on it.

PS Having said all of that!--- based on my work with the 34s emulator, the 35s doesn't hold a candle to it functionally (especially the programming paradigms) but again... my son isn't going to be able to use the WP34s I'm repurposing for him on the NCEES exam. Its sad, but true. so... what's an engineering father to do...?

Cheers