Hello,
I am fairly new to MoHPC, but have been using HP calculators for over 2 decades.
I came across the HP30B/20B repurposing project, and was wondering if it was close to being made available to testers. I see that it now has an entry as of Jan 2011 on sourceforge (http://sourceforge.net/projects/wp34s/).
Thanks!
Hi Svisvanatha,
things are about ready to be ported on the real 30b, as you may have seen in previous threads at the turn of the year. This porting is a task Thomas Chrapkiewicz wanted to attack since Pauli doesn't have access to the necessary equipment anymore. I didn't hear or read anything about Thomas' progress recently, however - intercontinental projects have their intrinsic challenges.
Preliminary documentation is found at the sourgeforge address you mentioned. It is (and will be) changing without notice for the time being until we've got the feedback our modifications are up and running smoothly on the 30b.
In any case, we appreciate whoever wants to join us in our quest to get this job done properly. The site at sourceforge was installed and is maintained for sharing information and data.
Best regards,
Walter
Hi,
is there any code out there. over at SF, i can only find the documentation - which looks pretty good btw.
thanks,
-- hugh.
Thank you for the reply Walter. I am very interested in the progress of this project, and the documentation on SF is very promising as to the potential impact this could have for enthusiasts.
Once a flashable image is available, will that show up on SF as well? Instructions, pre-requisites, and the like?
Thanks!
I set up a source forge repository for the code:
http://sourceforge.net/projects/wp34s/
The only download currently is Walter's very comprehensive documentation. The source code is available via subversion but only works on Linux and Mac via a command line terminal.
- Pauli
Hello Pauli,
Thanks! I downloaded the code via svn. My machine is running Ubuntu, and when I try to 'make', I get the following output. It looks as though the file curses.h is missing from the svn, and then there are errors within keys.c that are really warning, but being treated as errors. Any thought on what I could do to make the project? Many thanks!
------------------------------------------------
user@computer:~/hp/wp34s/trunk$ make
make[1]: Entering directory `/home/user/hp/wp34s/trunk/decNumber'
make[1]: Nothing to be done for `all'.
make[1]: Leaving directory `/home/user/hp/wp34s/trunk/decNumber'
gcc -Wall -Werror -g -fno-common -fno-inline-functions -fno-defer-pop -DINCLUDE_AGM -DINCLUDE_ZETA -DINCLUDE_DIGAMMA -DINCLUDE_ELLIPTIC -DINCLUDE_BESSEL -DINCLUDE_DIGAMMA -DINCLUDE_DBLFACT -DINCLUDE_SUBFACT -O0 -g -DUSECURSES -DDEBUG -c -o keys.o keys.c
In file included from keys.c:20:
lcd.h:57:20: error: curses.h: No such file or directory
cc1: warnings being treated as errors
keys.c: In function ‘main’:
keys.c:1983: error: implicit declaration of function ‘getch’
keys.c:1983: error: ‘ERR’ undeclared (first use in this function)
keys.c:1983: error: (Each undeclared identifier is reported only once
keys.c:1983: error: for each function it appears in.)
keys.c:1992: error: implicit declaration of function ‘clear’
make: *** [keys.o] Error 1
Edited: 28 Jan 2011, 8:09 p.m.
You need to install the ncurses development library.
- Pauli
The keyboard to layout mapping isn't obvious either. It is roughly presented in doc/layout.
- Pauli
OK, I installed the ncurses library, and now face the following:
user@computer:~/hp/wp34s/trunk$ make
make[1]: Entering directory `/home/user/hp/wp34s/trunk/decNumber'
make[1]: Nothing to be done for `all'.
make[1]: Leaving directory `/home/user/hp/wp34s/trunk/decNumber'
gcc -Wall -Werror -g -fno-common -fno-inline-functions -fno-defer-pop -DINCLUDE_AGM -DINCLUDE_ZETA -DINCLUDE_DIGAMMA -DINCLUDE_ELLIPTIC -DINCLUDE_BESSEL -DINCLUDE_DIGAMMA -DINCLUDE_DBLFACT -DINCLUDE_SUBFACT -O0 -g -DUSECURSES -DDEBUG -c -o keys.o keys.c...snip...
gcc -Wall -Werror -g -fno-common -fno-inline-functions -fno-defer-pop -DINCLUDE_AGM -DINCLUDE_ZETA -DINCLUDE_DIGAMMA -DINCLUDE_ELLIPTIC -DINCLUDE_BESSEL -DINCLUDE_DIGAMMA -DINCLUDE_DBLFACT -DINCLUDE_SUBFACT -O0 -g -DUSECURSES -DDEBUG -g -o calc keys.o display.o xeq.o prt.o decn.o complex.o stats.o lcd.o int.o date.o xrom.o consts.o alpha.o charmap.o commands.o string.o -LdecNumber -ldecNumber -L. -lconsts -lcurses
/usr/bin/ld: cannot find -lconsts
collect2: ld returned 1 exit status
make: *** [calc] Error 1
The build process should have created a libconsts.a in the truck directory. That is what isn't being found. Try a "make clean" and then "make" to rebuild everything. If that doesn't work, there is something screwy going on. "make consts.h" should build the library alone. It does require "ar" and "ranlib" but you should have those two commands already but still worth checking that they are there.
- Pauli
Edited: 28 Jan 2011, 8:37 p.m.
OK, thank you! Fantastic work, and very exciting.
./calc now brings up an emulator for the wp34s. I can do some basic functions using the keymap in /doc. Some questions:
1. Is there a more complete keymap, for instance, for shifted keys, etc?
2. What is the best way to help test? Using the sourceforge bug tracker and forum, or posting here?
3. What is the remaining process to get this onto the 30B? (sorry for the naive question, but my forte does not lie in electrical or computer engineering)
Edited: 28 Jan 2011, 8:50 p.m.
ok, i had a go at making a windows build. after a bit of a struggle i've got all the consts, cats and other libs going.
right now, im shooting a lot of gcc specific syntax. if im keen i might finish the job.
are you interested in windows changes?
1/ The complete keyboard map is in the documentation PDF file either in the doc directory or available as a download.
2/ Just send me and/or Walter a message on this forum. We're not really advanced enough to justify proper bug tracking yet. Plus, I'm not checking the source forge site often.
3/ To get this on the hardware required merging HP's development sources with what we've done. The basic methodologies of the two are the same & I did start doing it but decided I was going about it incorrectly and would need to start that bit again. Then life got in the way.
There isn't a huge amount to do to finish the port but it will take some programming ability/knowledge and some hardware.
- Pauli
Edited: 28 Jan 2011, 9:02 p.m.
Most definitely interested.
You've got mail coming soon.
- Pauli
Thanks again. Is the hardware required expensive? From the 20B repurposing project, the list of hardware is detailed here . I assume similar is required for the 30B?
The cable costs an email to me with your shipping address (including phone number). Unless you want hardware debugging, the rest is not used.
TW
svisvanatha wrote:
Quote:Promising? If this calculator comes true it's nothing short of sensational! It has all the nice things and functions I would like to have added to my beloved HP-41. Simply great!
...the documentation on SF is very promising as to the potential impact this could have for enthusiasts.
And, finally, there's a decent function set for statistical distributions and their inverses. I simply don't know why these absolutely basic functions are missing in most calculators that claim to be "scientific" - they are basics in mathematics as well as in most social sciences! Walter, can you say something about the implementation of, say, the inverse Normal and Student distribution? Did you use a set of rational approximations? An iterative approach? Or something else?
There are just two things:
First, it's a shame that such a great calculator (if it eventually comes true) doesn't provide any external storage. No memory card, no PC-transfer, no anything. Or did I miss something?
And finally - what about the keyboard and new key labels? In the HP41-days there was a way - the 82200A. ;-)
Dieter
Done, and thanks!
Quote:I am pretty reserved usually :)
Promising? If this calculator comes true it's nothing short of sensational!
...And finally - what about the keyboard and new key labels? In the HP41-days there was a way - the 82200A. ;-)
Hallo Dieter,
thank you for your kind words. With respect to the two points you raised:
Quote:You only missed we cannot perform magic. We have to take the 30b hardware as is.
... it's a shame that such a great calculator (if it eventually comes true) doesn't provide any external storage. No memory card, no PC-transfer, no anything. Or did I miss something?
Quote:We tried finding a way getting the keys relabeled. Pauli did some experiments, and they looked promising. An adhesive keyboard overlay (it has to be adhesive due to the design of the 30b, no other ways for attaching like e.g. in HP-41) may be an alternative, but so far we didn't investigate - it's way down on our agenda and looks viable so far. I assume the prototypes will get some sticky paper labels.
... what about the keyboard and new key labels? In the HP41-days there was a way - the 82200A. ;-)
Quote:
And, finally, there's a decent function set for statistical distributions and their inverses. I simply don't know why these absolutely basic functions are missing in most calculators that claim to be "scientific" - they are basics in mathematics as well as in most social sciences! Walter, can you say something about the implementation of, say, the inverse Normal and Student distribution? Did you use a set of rational approximations? An iterative approach? Or something else?
The Normal uses the error function which in turn uses gamma. The inverse normal uses the built in solver so it will be slow but should be reasonably accurate. I couldn't find a series approximation I liked & using the solver (which has to be there anyway) is much more space efficient.
Student's T is a special case of the incomplete beta function which comes down to a series approximation. The inverse is the solver again.
Quote:
And finally - what about the keyboard and new key labels?
I experimented with a 20b and clearing the labels off the keys is quite easy with a Dremel and a fine sanding & a polishing disc. Engraving and painting new labels would be straightforward but I never got to that stage since Walter keeps improving the keyboard.
- Pauli
Quote:
can you say something about the implementation of, say, the inverse Normal and Student distribution? Did you use a set of rational approximations? An iterative approach? Or something else?
Now I'm back home I'll give a more detailed response.
For those distributions where there are explicit formulas for the cumulative distribution (CDF) or its inverse (the quantile function or QF), I've used that. This is the case for the Weibull distribution, the exponential distribution and the geometric distribution's CDF. I don't take a possibly huge summation as an explicit formula so the binomial and Poisson distributions aren't calculated that way.
For the others, I've used generalised functions (the incomplete gamma and beta functions) calculated by convergent series. Functions using the incomplete gamma function as their basis are both Normal distributions, the Chi squared and the Poisson distribution. Functions using the incomplete beta are: Student's T, Snedecor's F and the binomial distribution. In all these cases and the geometric's QF, the solver is used to evaluate the QF.
Even though the solver is used to calculate these inverses, these functions can be used inside user code solver routines (although user routines cannot nest SOLVE).
- Pauli
Pauli, thank you very much for this information. Let me add a few more thoughts.
Using the solver for the quantile functions is an option if
(and only if) the whole calculation is done with the internal
15-digit precision. Otherwise this method is not accurate since
the quantile may change by several ULPs and still provide the
same probability returned by the solver.
Example:
Probability p = 0,18
Normal quantile z = 0,9153 6508 7842 8140 4978 ...12-digit result z = 0,9153 6508 7843
Value for z Normal CDF (12 digits)
------------------------------------------
0,9153 6808 7840 0,1800 0000 0001
0,9153 6508 7841 0,1800 0000 0000
0,9153 6508 7842 0,1800 0000 0000
0,9153 6508 7843 0,1800 0000 0000
0,9153 6508 7844 0,1800 0000 0000
0,9153 6508 7845 0,1799 9999 9999
------------------------------------------
But I think you will have access to the full internal 15-digit precision, so the solver should be okay.
Both for the Normal CDF as well as its inverse there are rational approximations, leading to fast execution and very accurate
results (full double precision, i.e. 15-16 sign. digits). For instance, take a look at the well-known AS241 algorithm. It uses three (7;7) rational approximations with 45 numeric constants. As long as these constants are given with 20 significant digits the relative error is less than 10-16 over the whole double-precision range, i.e. down to p = 10-308. At the working limit of our HPs (10-499) the error still is less than 0,2 ULP.
Of course you can also find your own rational approximation optimized for your specific working range and with exactly the precision you need. In many cases usual spreadsheet software like Excel is all you need. ;-)
Another possible way is a much simpler approximation, giving 7 significant digits, followed by one single Newton-Halley iteration, which should provide the desired 12-digit result.
If you're interested in series approximations, in any case take a look at the papers by Shaw and Steinbrecher where several possible solutions can be found. For instance, the Normal Quantile can be evaluated to full 12 significant digits (+/- 1 ULP) for values below approx. 10-90 with a simple series evaluated to just three terms.
Yes, it's an interesting topic, this quantile thing... :-)
Dieter
Quote:
Using the solver for the quantile functions is an option if
(and only if) the whole calculation is done with the internal
15-digit precision. Otherwise this method is not accurate since
the quantile may change by several ULPs and still provide the
same probability returned by the solver.
Actually, we don't have 15 digit precision. Our basic reals are 16 digits and internal computations are carried out with 39 digits. To the best of my knowledge, most of the internal functions are correct to all 16 digits (although achieving this over their entire range & proving such is pretty much impossible). The statistical distributions aren't fully accurate unfortunately, however they are good enough for most applied statistical work.
In your example Q^-1(1 - 0.18) displays 9.15365087843E-1 and the actual value in X is 0.9153650878432420. So reasonably accurate but not perfect.
I'm aware of several rational approximations. My problem here is a trade off of accuracy vs time vs space. I'm not terribly concerned about the time at the moment but accuracy and space are important. Unfortunately, we've pretty much completely filled flash memory so space efficient algorithms are important. Using the solver is very space efficient. Storing 40 constants and having extra code to calculate the approximation isn't :-( I'm getting better value for the space by having the incomplete beta and gamma functions and the solver.
I do agree that this is an interesting topic.
- Pauli
Well, I have a HP 30B on order now, and Tim was kind enough to offer to ship me a flashing cable.... so that is a start. I will try to spend some time to test out the wp34s on my Ubuntu machine and provide feedback.
Thanks in advance - we are looking forward to your findings.
Hi all,
I'm a recent convert to HP RPN (thanks to a good engineer at work).
I received my 30b today. It looks and feels great. I'm excited about the 34s project. Let me know how I can test. (I am checking out the source now.).
Does the original 30b image exist somewhere? So I can go back if needed.
Last, I also have the 12CP with the arm processor. It looks like the serial connector pads are covered with black tape. Can it be removed easily? (I just got it from Fry's so dont' want to mess it up if I wanted to return it.)
Ok, one more thing. I also received my 49g. I like it.
Ok, this is seriously last. my 48gx also works great :)
thanks,
erturk
I'm awaiting a cable for my 20b & 30b, too. It would be interesting to know how to backup the actual flash contents as a way to fall back to the original functionality. I'm planning to use my 20b for the experiments. Can it be "upgraded" to a 30b?
Quote:From the user's point of view, the main advantage of the 30b hardware is a significantly improved keyboard (see "rotate and click"). You won't get this with any 20b so far.
I'm planning to use my 20b for the experiments. Can it be "upgraded" to a 30b?
Nope. We built the 30b software to make sure it won't run on a 20b.
Well, I guess it will run, but before it moves into any 30b functionality it validates you are running on the 30b. I guess you could probably hack it to force it to work. . . but isn't a good keyboard and a better screen worth the bit extra? :-)
TW
Hi Tim,
I don't want to take my 30b out of service by using it as a development platform. Just for software testing, a 20b should suffice. Running the 30b firmware on the 20b hardware would be one of these tests. You convinced me that I can skip this step.
Looking forward to the cable.
I bought my 30b from amazon for $22. 20b was $18. Definitely worth the extra bucks.
On my side of the pond, prices were definitely not in the range mentioned when I got my two machines.
Looks like free trade again. One country gets the bargain, the others pay for it :(
Current prices at dynatech.de:
20b: 43,50 €
30b: 49,95 €
The price difference isn't to high. I wonder who will buy the 20b given he knows about the differences in hardware and functionality.
What's the best way to build the guessing game in the sdk for the hp20b/arm?
I downloaded the free IAR kickstart version. I'm able to open the project file but get linker errors during build.
Are these due to the IAR free version (missing libraries?) or is there something else that needs to be setup?
Thanks,
erturk
Linking
Error[Li005]: no definition for "__aeabi_uidiv" [referenced from c:\Documents and Settings\Administrator\Desktop\hp30b\Developers\firmware\
AT91SAM7L\ewp\HP_20b\Obj\application.o]
Error[Li005]: no definition for "__aeabi_uidivmod" [referenced from c:\Documents and Settings\Administrator\Desktop\hp30b\Developers\
firmware\AT91SAM7L\ewp\HP_20b\Obj\application.o]
Error[Li005]: no definition for "__aeabi_lmul" [referenced from c:\Documents and Settings\Administrator\Desktop\hp30b\Developers\firmware\
AT91SAM7L\ewp\HP_20b\Obj\application.o]
Error[Li005]: no definition for "strlen" [referenced from c:\Documents and Settings\Administrator\Desktop\hp30b\Developers\firmware\
AT91SAM7L\ewp\HP_20b\Obj\application.o]
Error[Li005]: no definition for "__aeabi_memclr4" [referenced from c:\Documents and Settings\Administrator\Desktop\hp30b\Developers\
firmware\AT91SAM7L\ewp\HP_20b\Obj\application.o]
Error[Li005]: no definition for "__aeabi_memcpy" [referenced from c:\Documents and Settings\Administrator\Desktop\hp30b\Developers\
firmware\AT91SAM7L\ewp\HP_20b\Obj\application.o]
Error[Li005]: no definition for "sprintf" [referenced from c:\Documents and Settings\Administrator\Desktop\hp30b\Developers\firmware\
AT91SAM7L\ewp\HP_20b\Obj\application.o]
Error[Li005]: no definition for "__aeabi_llsl" [referenced from c:\Documents and Settings\Administrator\Desktop\hp30b\Developers\firmware\
AT91SAM7L\ewp\HP_20b\Obj\graphics.o]
Error[Li005]: no definition for "__aeabi_idiv" [referenced from c:\Documents and Settings\Administrator\Desktop\hp30b\Developers\firmware\
AT91SAM7L\ewp\HP_20b\Obj\main.o]
Error[Li005]: no definition for "__aeabi_llsr" [referenced from c:\Documents and Settings\Administrator\Desktop\hp30b\Developers\firmware\
AT91SAM7L\ewp\HP_20b\Obj\graphics.o]
Error while running Linker
Tim has sent me the cable, so the first step is done. :)
In fact, I've got a spare cable which I can share with someone here in Germany who wants to join in.
I've asked for a price quote from IAR for the full compiler version but I fear that might end up too high for me. We'll see.
Marcus
I've used the free GNU tools (GCC etc.) for lots of ARM development, though not specifically on the 20b/30b. To get the HP SDK sample code working with GCC would require a little bit of work. Regrettably I do not presently have any spare time to assist in such an effort.
The easiest way to get the GNU tools working, on Windows or Linux, is to use the CodeSourcery G++ Lite for ARM EABI distribution:
http://www.codesourcery.com/sgpp/lite/arm/portal/release1592.
I'm not sure how the code size optimization with GCC compares with IAR, but they don't have the size limit of the IAR Kickstart edition.
Edited: 8 Feb 2011, 3:51 a.m.
Hi Tim,
Quote:Seems I missed something: What do you mean by "better screen"? AFAIK the displays in both 20b and 30b are identical. TIA for enlightenment d:-)
isn't a good keyboard and a better screen worth the bit extra?
IAR has answered. For the price of the full compiler I could by a stack of iPads. :(
I've downloaded the gcc version Eric has proposed. But it will take time to adapt the SDK.
I was using HP-GCC for testing code size & was intending on using that for later development. Still, most any gcc version ought to be okay.
- Pauli
The displays are similar. I think the one in the 30b has a higher resolution. The biggest difference though, is the fact that the 30b's screen is below the surface plastic, which is flush with the body of the calculator. Therefore, there is no gap between the body and the display for dust to get into. On the other hand, the 20b's screen has the plastic set right against it and there is a lip where the body meets it, similar to other calculators.
I think there are other differences that people should acknowledge that aren't physical. For example, the 30b has a many more functions than the 20b - or any other financial calculator, for that matter.
Mark
Quote:?!?
The displays are similar. I think the one in the 30b has a higher resolution.
I have a 30B, but no 20B near to compare. I would say, from available information, that both displays have the same size, number of segments (the restrictions were commented by Cyrille some time ago), and hence resolution. Possibly the 30B has better looks, mounting method or contrast; but all information suggests they are functionally the same.
The ATMEL CPU supports exactly 400 LCD segments. The display is a 100% implementation of this constraint. On both machines.
Okay, I can see how what I said was confusing...
Resolution isn't the best word. The 20b and 30b have visually similar displays. Meaning, that the way their screens are arranged are identical. One line is dot matrix. One line is segmented. The annunciators are in the upper right side of the screen. This is how they are similar. However, I think the plastic above the display affects the way the 30b displays information in a positive way. It has the effect of making the lcd segments appear darker with better contrast than the 20b. Also, it completely protects the display from dust contamination.
Here are some images that you can look at to see the differences (keep in mind that these aren't the best images around) The darkness of the characters on the 30b compared to the 20b doesn't show well in these pictures. However, I have seen both head on and the difference is quite real.
20b from angle:
30b from angle:
Regards,
Mark
Edited: 8 Feb 2011, 5:07 p.m.
Gosh, it would be wonderful to report at HHC2011 this coming September 24-25 (in San Diego, by the way, with HP hosting again) that we have at least a handful of active, exciting user projects ongoing - in the 34S, Monte Dalrymple's NEWT/41CL project and Meindert Kuipers' latest MLDL2000 iteration. Keep saving your pennies....
Jake Schwartz
Richard Ottosen and I will probably be presenting some new stuff as well, but as always, we're secretive about it until the actual conference. We don't want to promise something and then fail to deliver. That happened to us for HHC 2010, so we were glad we hadn't promised anything. Under promise and over deliver!
I have not seen either in real life. But in your pictures, the 20b screen looks more contrasty to me!
I also don't understand your dust comment. With the Pioneers, it's the screen cover that causes the dust problem. No cover, no space between to collect dust.
Similarly, with the Pioneers, the screens w/o the dust cover have higher contrast.
I am not trying to be argumentative, just trying to understand.
Quote:
Under-promise and over-deliver!
This is a simple recipe for cheerfulness all around, because...
'Happiness = Reality/Expectations', which admittedly assumes that the expectations were positive.
Another form is 'Happiness = Reality - Expectations', but this allows for no glee if high expectations are actually achieved.
Suggestions for a more mathematically-robust formula?
-- Karl
Edited: 8 Feb 2011, 11:17 p.m.
These pics are not mine. They were two of the best angled shots I could find from the web. Also the 30b shot was at a different angle than the 20b, so it isn't exactly comparable. I did mention that the shots do not show the difference well. However, I have seen them both head-on and the 30b has a much darker look to the screen. To me, the 30b is more pleasing on my eyes.
Regarding the dust issue, there is a crack in between the upper casing and the screen on the 20b. This is illustrated in the pics where you can see the ledge. On the 30b, the upper casing and the plastic over the screen are one and the same. It is seamless with no crack. Therefore, there is no place for dust to get in.
Speaking of pioneers, I have my 20S in front of me and it also has the upper cover and a recess where the display meets the cover. This has a crack where the screen and the cover meet. The 20b is similar.
Mark
Edited: 9 Feb 2011, 1:31 a.m.
I've always thought Happiness = d(Reality)/dt because we adjust over time. Never thought about an Expectations term. Hmm...
Quote:From another post of yours, it seemed you had saved the original flash contents and restored it after flashing a demo rom. What procedure did you use to save the existing flash contents from the calculator? (I am on a 30B)
I'm awaiting a cable for my 20b & 30b, too. It would be interesting to know how to backup the actual flash contents as a way to fall back to the original functionality. I'm planning to use my 20b for the experiments. Can it be "upgraded" to a 30b?
I'm not sure that there's any way to save the original ROM contents without a JTAG cable (entirely different than the serial cable). Flash programming is done using "SAMBA" in the masked ROM of the AT91SAM7L128, but I don't think there's any way to force it into SAMBA that doesn't erase the Flash memory.
The screen on the 30b is a much higher quality LCD display. It isn't just the fact that it is recessed.
I think what you are trying to convey is that the pixel coloartion is consistent across the entire screen. On the 20b some pixel areas sometimes seem slightly lighter than others. The 30b doesn't have that issue and so is much easier to read.
TW
Edited: 10 Feb 2011, 1:28 a.m.
Or you can just ask and get the zip file. . . ;-)
TW
Tim,
That is exactly what I was trying to convey. It is weird. Sometimes, I have absolutely no problem conveying my thoughts and other times, I have great difficulty. I suppose it doesn't help that I've been sick for the last week. If I responded to this today, I would have probably been able to give a much better response.
Thanks for jumping in,
Mark
The 20b image came with Cyrille's SDK. Since you can only start the SAM-BA serial boot loader by clearing the flash, there is no way to use this tool for a backup of your current setup. I guess it will be possible through the optional JTAG connector.
Thank you for the responses! The 20B documentation clearly indicates that a serial-usb adapter will not work for this purpose, and that a hardware serial port is required. Is this indeed the case? My machine does not appear to have a plain jane serial port.
There are two ways to interface the calculator:
1. Serial pogo connection below the batteries.
2. Optional JTAG connection, which you will have to go through the trouble of installing. Personally I don't like this option as it will result in defacing the device.
If you want to back up the software or want to debug, you will need the JTAG. However, Tim is offering you the backup of the 30b software on a zip file, so it really is unnecessary to install the JTAG (unless you want to do more technical stuff).
The cable that works with the pogo connection will plug into a USB port with a serial adapter. You should then be able to flash the device and push the new software to it or push a backup of the old software. Since you can get a backup from Tim, you won't need to back it up yourself.
Regards,
Mark
Edited: 10 Feb 2011, 3:49 p.m.
Thank you. I will buy a serial-usb adapter and be off. Tim has sent me the zip file. I am sure I will have questions when I start to play around with this more!