NEW: HP 15C Collectors Edition

[cdmackay]

My first one had a mark on the bezel, from contact with the box, and the logo was applied crooked. That went back, and the replacement was fine, as was my second which I bought elsewhere, from José at the Calculator Store (who I strongly recommend, as a long-time happy customer).

[adamj]

Hi all

I haven't posted on here before, but my searches haven't turned up anything so I wondered if forum members could help check this out for me.

I've got a new HP15C CE (Christmas gift!) and it's looking very nice. Nicely produced/presented manual too. A little bit of key bounce sometimes as some people have mentioned, but I do like the feel and my technique I think is improving.

So, my question relates to matrix processing. I ran through the example at the top of Section 12 (pp 138-139) to solve simultaneous equations. This generated the correct result into matrix C, but I found that the last 2 elements of my input matrix A were clobbered.

The input matrices are set up as follows:

Code:


A
  3.8      7.2
  1.3     -0.9

B
 16.5
-22.1

At the end of the example, the state of the three matrices was as follows:

Code:


A
  3.8      7.2
  0.3421  -3.3632  <---- these elements changed

B
 16.5
-22.1

C
-11.2887
  8.2496

For people who don't have the manual to hand, the calculation is C = inverse A x B (or B/A).
Obviously, I've run through this a number of times and it's repeatable on my unit. On an emulator on my phone, the input matrix A is not changed by the calculation. So while it is certainly possible that this is user error on my part (I need to reset or something), it doesn't feel like correct behaviour.

For reference, running the self test checksum test (2), returns:
ChE--9090h
2023-01-05
C 2008 hP

Serial number: 02870.
Could someone else check this on their calculator?

[adamj]

From studying a bit further, it turns out that this is designed behaviour, which the manual explains on p148. The matrix A is replaced by an 'LU decomposition' of itself, which is part of the solving implementation. So in reading up about that I learnt some new maths! :-)

But since the emulators I tried don't generate the LU decomposition (Retro 15C on iPhone, and https://hp15c.com) it's still worth the note in case anyone else is confused by this.

[Eddie W. Shore]

(11-02-2023 06:05 AM)Eric Rechlin Wrote:  

(11-02-2023 05:00 AM)dmh Wrote:  Thanks for the detailed information and photos Eric !

A previous post mentioned a button was still used, which line has the button connected?

It appears the wiring is direct and not going through the chip at all now, is that correct?

So if you can roll your own pogo type connector could you could make a new cable the same way?

So, where do I get the firmware (original and patched) and what software is now required?

Thanks again.

The Reset button and Erase button are both still connected. You don't need Reset because you can just use the reset hole on the calculator instead, but the button is more convenient. It's connected in the same place as the old cable, which I believe is pin 2, which would be to the left of the black (ground) wire, lower left corner, in the last image above. I believe Erase is pin 6, which would be to the right of the black ground wire, lower right corner. It's also unchanged from the old cable.

Wiring is indeed direct, bypassing the chip. The chip was only used for the TTL to RS-232ish voltage level conversion. Since the new calculators speak USB directly there is no need for any level shifting.

As long as you have something to work for the pogo connector, which is the toughest part, you can make your own cable.

Software is just a newer version of the SAM-BA software used for programming the older calculators. I have been using version 2.14; I don't see that for download anywhere, but I do see version 2.16 is still available, so maybe you could use that. It still claims to support the sam4l chips so I assume it will work.

To flash the calculator, here are the directions I found worked best:

1. Install SAM-BA.

2. Remove the battery door from the calculator and plug in the 6-pin connector from the cable.

3. On the cable, hold down Erase, then press and release Reset, then release Erase. This just puts the calculator in USB linking mode. Unlike on the 20b/30b, this won't actually erase your firmware, so you can safely do this even if you fail to connect in the later steps.

4. Plug the USB connector from the cable to your PC.

5. Wait for it to detect as a USB Serial Device. Then check Device Manager for the COM number; usually I saw COM4 or COM5.


6. Start SAM-BA, make sure the right COM port is selected, and select at91sam4l-ek as the board type:


7. Click Connect. A new window will appear, where you must enter 0x04000 as the address, pick the ROM file to upload, and then click Send File:


8. After the file is done transferring (less than 10 seconds), press the Reset button on the cable (or use the reset hole on the calculator) and then turn the calculator on.

If you have a problem anywhere prior to the point where you click "Send File", pressing the Reset button will let you get back to the "Pr Error" display without losing the firmware. If a failure occurs after clicking "Send File", then your calculator may not operate until you can successfully load a new firmware.

I just received the cable today and installed SAM-BA 2.16 (can't find 2.14 anywhere). I have Windows 10. The software installed but when trying to open the software I'm not getting a single screen at all.

Update: I had to be in an Administrator account. 2.16 didn't have the at91sma4l... board. I'll have to try the sam4lek[not factory program] later.

2nd Update: The sam4lek[not factory program] works for only one time. Where does the HP_IAR.bin file come from? Do we have to create a that file? I thought there were three patches to fix, not just the 0x04000 address.

Apologies for all the questions.

[Divasson]

Hi, you can find the way to perform all changes here (there are 4 pages in total describing the downloading of current firmware, the modification and the upload of the modified firmware)

Using the programming cable for HP15c CE

[rkf]

In between of all discussion I just want to express my gratitude to José for providing the community not only with the HP15C CE, but also the cable needed to flash the units having the HEX display bug when working in the unsupported HP16C mode.

I ordered the cable presumably as one of the firsts 12th Nov, as it was priced yet a bit above todays 20€ for existing HP15C CE customers. Finally, I received the cable yesterday, and thanks to the wonderful and elaborate instructions given on José's website I had today the chance to patch and flash the firmware of my unit - bug removed!

I know many of you are disappointed for waiting that long for the cable, but I can say of firsthand experience that it's worthwhile the wait. And I'm sure the reasons of the delay are out of José's reach. And - to be honest - it's much better to have a fully functional unit at the end of the day than to be left alone with a buggy one!

[Eddie W. Shore]

Thank you Jose for posting the instructions and the cable. I got my HP 15C CE updated last week. Yay!

Yes, the cable is worth the wait.

[Harald]

(02-06-2024 09:55 AM)rkf Wrote:  I ordered the cable presumably as one of the firsts 12th Nov

I beat you to it. Ordered on November 8th.
I didn't want to hack up my WP34s cable and while I couldn't wait for the cable to arrive, the delays didn't really matter. This is a hobby after all

[Divasson]

We have received a new lot of calculators, and these come again with a plastic applied to the bezel, as the initial prototype. That will ensure less damage during transport. (we'll still test the keyboard in the Enhanced version)

       

(yellowish color is solely due to low color temperature lamp on my desk). These are already part of the second production batch - see serial number. Same original firmware though - the cable is still needed!

[Michael Lopez]

Hi all,

Received my cable a few weeks ago but only just had the opportunity to download the firmware to fix the HP16c mode DEC bug. Have been following the excellent instructions on the Calculator Store & everything worked as described by Jose.

However, I am having trouble locating the memory addresses that need to be modified. Downloaded & installed free Hex Editor Neo for Windows but can't seem to find the following addresses:

> At 0x0EF5E we find 46 03 - and this is not what we expected (346)! Then we remember that the coding here is "Little Endian" so all byte nibbles are reversed - so it is correct! We need to replace the 46 03 by 00 00.
> At 0x0EF60 we find CB 01 - which is the Little Endian 2-bytes form of the 10-bit number 1CB. We need to replace the CB 01 by 00 00
> At 0x0EFB0 we find 7A 01 - same as above, this is the Little Endian form of 17A. We need to replace it by AC 03.

I've only dabbled a little in the past with Hex Editors so probably just my lack of experience :-(.

Would any of you please suggest how I might find the addresses within the editor?

Thanks,

Michael

[BruceH]

(03-31-2024 06:39 AM)Michael Lopez Wrote:  Would any of you please suggest how I might find the addresses within the editor?

I don't have it installed, but going from the screenshots on Neo's home page the value on the left is the start address of each group of 16 octets and each column shows each successive octet in turn.

So to find address 0x0ef5e you scroll down until you are on row 0000EF50 and then move across the columns until you are in column 0e. The value in the top left corner should then show 0000EF5E as a confirmation.

The display is slightly confusing in that the column labels are shown as two hex digits 00 - 0F when they should be 0 - F as they represent just the last digit of the address.

[Michael Lopez]

Thanks very much BruceH. Your explanation of how NEO displays the addresses is very helpful. Will give it another go shortly.

Cheers,

Michael

[Michael Lopez]

Hi all,

Firmware update went well & 16c DEC bug now addressed :-).

Have seen reference in the Calculator Store to increasing the key bounce time tolerance from 80 to 100 on the units they are now selling. I presume this also requires the firmware to be modified? If so, are there any references to the addresses to be altered? Finally, is it worth doing?

Thanks,

Michael

[Johnh]

I'm also interested in such a key-bounce fix. It's helpful even just understanding that such a time tolerance addresses the issue. But to increase the tolerance time a bit, even to somewhat longer than 100ms , maybe 150ms? would hopefully knock the issue on the head.

Personally, I bought my Hp15C CE , to be just that, as designed. It's interesting to read about switching to Hp16 mode etc, but that's not something I'd want to do.

[tommi60]

Just received my HP-15C CE from Amazon today. Seems it's from the first run, serial number is 8795.

No plastic film were applied to the bezel and when I turned on the calculator I noticed FIX 6 was selected. Strange I thought, shouldn't it be FIX 4? Checking 'g MEM' I noticed there was a program also inserted!

001 LBL A
002 STO 1
003 LBL 1
004 1
005 0
006 X>Y
007 GTO 2
008 1
009 ENTER
010 INT
011 STO 1
012 FRAC
013 R/S
014 1
015 0
016 *
017 RCL 2
018 +
019 STO 2
020 GTO 1
021 LBL 2
022 Σ+

Not even an interesting program, just someone playing around with my supposedly brand-new calculator.

Apart from bouncing keys I don't see any marks or blemishes, so I might keep it. Feeling lazy to send it back. And maybe the bouncing key problem can be solved in firmware? Still waiting for the POGO pin programming cable.

P.S.
My guess this is a returned item to Amazon. Should have gotten a discount at least.

Edit:
I really tried to like the calculator, but it's going back. With keys that intermittently don't register and with bouncing keys at random intervals, it's totally hopeless. I'm scared now of getting a replacement, lost the confidence somehow. Maybe get a Swissmicros DM15L instead?

[Voldemar]

(04-02-2024 11:17 AM)tommi60 Wrote:  when I turned on the calculator I noticed FIX 6 was selected. Strange I thought, shouldn't it be FIX 4?

I ordered the first one in May, it arrived at the beginning of August. Calculator in an unopened package, it didn't look like the package had been opened. When I turned it on for the first time, the display showed 6.000. I didn't pay attention to it. Maybe during a factory check?

[Divasson]

I have reprogrammed and tested more than 200 units, so I am familiar with unopened units. All were somehow tested at the factory and had a number on screen - most single digits but some had e or pi. So, be confident that it is an unopened machine. I never checked for a program, though!

150ms debouncing was tested but then we missed strokes when we wanted to repeat numbers. Far too slow.

[Michael Lopez]

Thanks Divasson.

I was interested in trying to increase key bounce tolerance from 80 to 100 mS. Is there a post we can refer to on how to make this change? I already have your POGO pin cable.

Cheers,

Michael

[chromos]

(04-01-2024 06:37 AM)Michael Lopez Wrote:  ... Have seen reference in the Calculator Store to increasing the key bounce time tolerance from 80 to 100 ... are there any references to the addresses to be altered?

(04-02-2024 06:19 AM)Johnh Wrote:  I'm also interested in such a key-bounce fix.

(04-03-2024 08:54 AM)Michael Lopez Wrote:  ... I was interested in trying to increase key bounce tolerance from 80 to 100 mS. Is there a post we can refer to on how to make this change? I already have your POGO pin cable.

I also bought a cable from thecalculatorstore.com and would also like to increase the debouncing time from 80 ms to 100 ms. On the calculatorstore.com website, I only found information about the firmware modification to fix the dec bug, but nothing about debouncing keys. Please, I beg you Mr. Divasson to publish this information. Thank you.

[ctrclckws]

If you purchased a second B grade 15C ce, it should come with the debounce change as well.

Then use the cable to extract the firmware from that one and load to the other one.

I leave other possibilities to the user.