Thread: hp-41cx / NoV-64d / Hepax XF multi-function

[kj.vandelooij]

Hello everyone,

I have a question about Hepax multifunctions on a HP-41cx with a NoV-64d module. When I key in a little program like

LBL TST
XF
0

and exit the program mode to run it, the result is 'nonextistent'. After this, entering program mode again shows

LBL TST
XROM 07,53
0

Only after turning the calculator OFF and ON again, with

H:FAT FULL

appearing in the display, the program is changed back into

LBL TST
XF
0

and executes fine, returning the length of the string in the alpha-register in the x-register

Note: there are no additional modules or hardware plugged in. Neither have I added any modules to the NoV-64d. I cleared all four RAM chips manually as per the procedure described in the NoV-64d manual.

Questions: 'Is this behavior normal? Is there a way around it? And if there is: How can I change things back to normal?'

Thanks for reading,

Kees Jan

[rprosperi]

Using these HEPAX functions to see what the module thinks is in HEPRAM:

[XEQ] "HEPDIR" List of files in HEPRAM
[XEQ] "BCAT" List ROM Pages (3-F) to see how HEPAX is seeing what is in each page.

It sounds like it is not actually cleared, but these results may help you to understand what's going on.

[kj.vandelooij]

(09-04-2023 03:02 PM)rprosperi Wrote:  Using these HEPAX functions to see what the module thinks is in HEPRAM:

[XEQ] "HEPDIR" List of files in HEPRAM
[XEQ] "BCAT" List ROM Pages (3-F) to see how HEPAX is seeing what is in each page.

It sounds like it is not actually cleared, but these results may help you to understand what's going on.

Thanks Bob,

I got the following results:

HEPDIR
H:DIR EMPTY

HEPAX 002
3: -ext fnc 2d
4: no rom
5: -time 2c
6: -hepax 1d
7: no rom
8: hepax ram
9: @@@@@@@
A: hepax ram
B: hepax ram
C: no rom
D: no rom
E: no rom
F: no rom

Page 9 is not what it should be and indeed suggests that the memory has not been cleared. After this block dir I applied the CRF method:
1. enter + on: does not work.
2. <- + on: does work.

HEPAX 002
3: -ext fnc 2d
4: no rom
5: -time 2c
6: no rom
7: no rom
8: no rom
9: no rom
A: no rom
B: no rom
C: -hepax 1d
D: no rom
E: no rom
F: no rom

I would expect 'hepax ram' in 8 - B, but clearing hasn't happened yet, so this might be normal.

Next, still following the procedure, for each chip, I cleared the pages:
000 -> 4100
OK -> alpha
8 -> X
CLRAM
9 -> X
CLRAM
10 -> X
CLRAM
11 ->
CLRAM
and the same for 001 -> 4100, 002 -> 4100, 003 -> 4100

After a power cycle, this resulted in:
3: -ext fnc 2d
4: no rom
5: -time 2c
6: -hepax 1d
7: no rom
8: hepax ram
9: hepax ram
A: hepax ram
B: hepax ram
C: no rom
D: no rom
E: no rom
F: no rom
for each of the four chips. While this result looks promising, keying in the test program and running it, like before, still results in:
1. nonexistent
2. 9:@@@@@ in the block dir

In my mind there are three possibilities:
1. I interpreted the procedure for clearing incorrectly.
2. Something is broken.
3. The XF functionality does not work in a HP-41 CX.

I hope somebody can confirm one of these possibilities, or can come up with a completely new one.

Thanks again,

Kees Jan

[Ángel Martin]

I can confirm that the XF and HEPAX multi-functions are indeed compatible with 41CX models, so this problem you're experiencing looks like an issue in the initialization of the HEPAX file system, or maybe an issue with the NoV64 configuration itself.

Also, HEPAX 000 is the header for that group, so strictly speaking is a "NONEXISTING" situation.
XF 000 on the other hand would be ALENG, so it should work just fine (it does on the HEPAX original, and the CL, DM-41X)

[kj.vandelooij]

Dear Ángel,

many thanks! I can forget about option 3 then; the XF function should work on a 41CX with NoV-64d. In a way it does work in my case, that is, apart from the fact that I need a '<- on' to return to normal behavior.

While superficially the file system seems to work (I can create files and perform operations on them, HEPDIR does work and list files), I would still like to check the initialization / integrity of it. Is there a way to do this?

Secondly: apart from the control word at H4100, are there other options for configuration of the NoV-64d module? I mean, options different than those offered by the Clonix Config program?

Greetings, and thanks again

Kees Jan

[Diego Diaz]

Hi Kees Jan, all,

As confirmed previously, the XF function is fully operative on NoV-64d.

Also, your HEPRAM erasing procedure works as expected, since you get HEPAX RAM on pages 8-B.

It will be good to RUN HEPROOM (should return 2,610) after erasing and power cycle. And remember you need to power cycle after erasing every chip for the HEPAX to build the FILE SYS.

At this point, it may help to check if you got the same problem when running this program with a different HEPAX RAM chip.

Please, place 001, 002 or 003 into H'4100 and test your program again.

It would be the first case but it is indeed possible that chip 000 is faulty.

Look forward to reading from you.

Best wishes.

Diego.

[kj.vandelooij]

Hi Diego,

Good to hear from you! I hope you are doing well.

I Switched to chip 1. An existing program

01 LBL TST
02 XF
03 0

with some string in alpha, does return the length of the string. Next, I erased to program and keyed it in again.

01 LBL TST
02 XF
03 0

Back stepping shows

01 LBL TST
02 XROM 07,53
03 0

Running the program yields 'nonexistent'. After a couple of times <- on, normal behavior is restored, and the program can be viewed as

01 LBL TST
02 XF
03 0

again, and it runs fine.

I got the same result for chip 2 and 3. So, yes the same problem occurs in all four cases, that is, with each of the four RAM chips.

Next, I cleared all four chips by
1) setting the control word
2) power cycle
3) "OK" 8 CLRAM
4) "OK" 9 CLRAM
5) "OK" 10 CLRAM
6) "OK" 11 CLRAM
7) powercycle
8) HEPROOM
9) HEPCHN?

In my case at least, the second power cycle was needed because without it, HEPROOM resulted in H:NO FILESYS. After the power cycle HEPROOM returned 2610 and HEPCHN? in 089AB0.

I am curious to the reason of why you suspect chip 0 might perhaps be faulty. Is there a more extensive way of testing it?

Greetings, and best wishes

Kees Jan

P.S. Congratulations on your latest product. If I need to continue running CLRAM as often as I did the last couple of days, I might need to consider a purchase :-)

[MeindertKuipers]

Hi all,

I have seen identical behavior on my NoVRAM. I must add that I had done a quick reconfiguration earlier and created a ROM conflict in a development setup. I wil try to do a full reset and clear of my NoVRAM unit and then try again.

[kj.vandelooij]

(09-06-2023 05:49 PM)MeindertKuipers Wrote:  Hi all,

I have seen identical behavior on my NoVRAM. I must add that I had done a quick reconfiguration earlier and created a ROM conflict in a development setup. I wil try to do a full reset and clear of my NoVRAM unit and then try again.

Hello Meindert,

That would be great.

Groeten,

Kees Jan

[Sylvain Cote]

I always use HEPAX 1D with my NoV-64d modules and I have no issue.

Testing HEPAX 4H on one of my NoV-64d modules with a 41CX Fullnut (ROM: 0:NFL 1:DCC)

Module programming:

1. Download the latest ROM files from Monte website (rom_files_230725.zip)
2. Unzip rom_files_230725.zip
3. Copy all the ROM files into the "Clonix & NoV Configuration" folder
4. Start ClonixConfig61.exe
5. Select NoV-64(d)
6. Select HEPAX 4H (4 pages for HEPAX-4H and 1 page for LIBRARY4 is selected) [warning: to my knowledge, LIBRARY4 only works with 41CX OS]
7. Assign HPX_PLUS.ROM to page E bank 1
8. Program the module
   

Module testing:

• OFF ; insert module ; ON
• CAT 2
  → -TIME 2C
  → -CX TIME
  → -HEPAX'4H
  → -HEPX_PLUS
  → -HEPX_APPS
  → -EXT FCN 2D
  → -CX EXT FNC
• Create Erase HEPAX RAM program:
  
  Code:
  

  LBL "EHR"
"OK"
8.011
LBL 00
VIEW X
CLRAM
ISG X
GTO 00
  Code:
  

  SF 11
OFF
HEPROOM
VIEW X
PSE
HEPCHN?
AVIEW
END
• Init all Nov-64d RAM
  • HEXEDIT ; ADR 4100 enter 000 ; exit HEXEDIT ; XEQ "EHR" ; wait until off ; ON // display 2610 then 089AB0
  • HEXEDIT ; ADR 4100 enter 001 ; exit HEXEDIT ; XEQ "EHR" ; wait until off ; ON // display 2610 then 089AB0
  • HEXEDIT ; ADR 4100 enter 002 ; exit HEXEDIT ; XEQ "EHR" ; wait until off ; ON // display 2610 then 089AB0
  • HEXEDIT ; ADR 4100 enter 003 ; exit HEXEDIT ; XEQ "EHR" ; wait until off ; ON // display 2610 then 089AB0
    
• Testing
  
  Code:
  

  input         display           comment
[PRGM]        00 REG XXX        enter program mode
LBL "TST"     01 LBL "TST"      program label
ENTER         02 ENTER          byte filler, just in case
ENTER         03 ENTER          byte filler, just in case
XF 000        05 0              display ALENG then 0 , we are 1 line too far, go back 1 line
BST           04 XROM 07,53     should display XF, something strange here
  Code:
  

  [PRGM]        0.0000            exit program mode
[ON]                            off
[ON]                            on but does no power on even after several times.
                                remove module
[ON]                            now it power on
[PRGM]        04 XROM 07,53     go back to program mode, normal the module is not there
[ON]                            off then insert NoV module
  Code:
  

  [ON]                            it power on
[PRGM]        04 XF             as expected but there is something going on here, this setup is not stable
[PRGM]        0.0000            exit program mode
[ALPHA]
ABC
[ALPHA]
XEQ "TST"     3.0000            ALENG of "ABC" is 3, it works, but still

I have to go, I will do more testing later.

Sylvain

[kj.vandelooij]

Hello Sylvain,

Thanks for trying.

Up till now I have not experienced any issues either.

While carried out manually instead of the clearing program you present, my procedure for wiping the memory in effect is identical.

I can also confirm that what you describe in 'testing' is very, very similar to what I experience:
1) The back step results in XROM 07,53
2) The 41 not powering on after a single 'ON'. It only does power on after several tries.
3) In the end, the program working like expected.

Please let me know if I can help doing some more testing.

Good luck, greetings,

Kees Jan

[Sylvain Cote]

Testing with HEPAX 1D on a second NoV-64d module, same 41CX Fullnut (ROM: 0:NFL 1:DCC).

Module testing:

• OFF ; insert module ; ON
• CAT 2
  → -TIME 2C
  → -CX TIME
  → -HEPAX 1D
  → -EXT FCN 2D
  → -CX EXT FNC
• Init all Nov-64d RAM (HEPCHN? has been removed from the test program)
  • HEXEDIT ; ADR 4100 enter 000 ; exit HEXEDIT ; XEQ "EHR" ; wait until off ; ON // display 2610
  • HEXEDIT ; ADR 4100 enter 001 ; exit HEXEDIT ; XEQ "EHR" ; wait until off ; ON // display 2610
  • HEXEDIT ; ADR 4100 enter 002 ; exit HEXEDIT ; XEQ "EHR" ; wait until off ; ON // display 2610
  • HEXEDIT ; ADR 4100 enter 003 ; exit HEXEDIT ; XEQ "EHR" ; wait until off ; ON // display 2610
    
• Testing
  
  Code:
  

  input         display           comment
[PRGM]        00 REG XXX        enter program mode
LBL "TST2"    01 LBL "TST2"     program label
"ABC"         02 "ABC"          test string
CLST          03 CLST           cleanup stack
  Code:
  

  XF 000        05 0              we are 1 line too far, go back 1 line
BST           04 XF             as expected
[PRGM]        0.0000            exit program mode
XEQ "TST2"    3.0000            ALENG of "ABC" is 3, it works
[OFF]+[ON]                      several power cycle, no issue            
XEQ "TST2"    3.0000            still working

With this limited testing, it seems that there is an issue with -HEPAX'4H and/or HEPX_PLUS , more testing is needed.

Sylvain

[MeindertKuipers]

I see nearly the same behavior after clearing the HEPAX memory.

However, my HP41CX does power off correctly, and when powering on shows FAT FULL. A subsequent power cycle turns everything back to normal. I had only HEPAX-1D loaded, not the other HEPAX support modules.

I have now hooked up my newly builtHP41 bus analyzer and will share the results.

In addition I see that the HEPRAM page in page B is messed up, the BCAT of Page 9 shows "@@@@@@@@@@@". Details of this page:
9000 - 00C
9001 - 02F
9002 - 000
9003 - 000

The HEPAX chain looks quite OK, with all pages initialized (200 in xFF3). A bit funny is that Page 9 is the first page in the HEPRAM chain, then A, B and finally 8. HEPROOM gives the correct result.

I have put the traces (hexcodes only, no disassembly) on my GitHub page here: https://github.com/mjakuipers/NoVRAM-HEPAX-traces

I did a quick scan, and in the trace where XF 000 is entered in the program, I see a bankswitch instruction to Bank 3 (traceline 7020 in the file '3-enter XF in PRGM mode.txt'), but never a bankswitch back to bank 0. This would perfectly explain the behavior that a BST cannot find the function again, as the main HEPAX bank with the FAT is gone. Power cycle will more or less fix this but with the wrong bank active strange things may happen.

Why is this happening? May be due to something in PRGM mode since the HEPAX XF function does ask for a 3-digit prompt like it is executing, but it never executes, the XF code is entered in program memory. But a quick test on my HP41CL works as expected (and it would be difficult to make a trace as this will probably not show on the HP41 bus).
Alternatively the NoVRAM firmware may have the enabled bank a bit more sticky than intended and does not automatically switch back to bank 0 in this situation.

My tests are done with my NoVRAM64d, loaded with HEPAX-1D and 4 HEPRAM pages in an HP41CX, nothing else.

[Sylvain Cote]

(09-07-2023 10:38 AM)MeindertKuipers Wrote:  My tests are done with my NoVRAM64d, loaded with HEPAX-1D and 4 HEPRAM pages in an HP41CX, nothing else.

Hello Meindert,
What ClonixConfig version did you used ? (v4.2 2014 or v6.1 2020)
What firmware version did you used ? (pre 2020 bank switching strategy or the new one)
Sylvain

[MeindertKuipers]

(09-07-2023 11:09 AM)Sylvain Cote Wrote:  

(09-07-2023 10:38 AM)MeindertKuipers Wrote:  My tests are done with my NoVRAM64d, loaded with HEPAX-1D and 4 HEPRAM pages in an HP41CX, nothing else.

Hello Meindert,
What ClonixConfig version did you used ? (v4.2 2014 or v6.1 2020)
What firmware version did you used ? (pre 2020 bank switching strategy or the new one)
Sylvain

I am using ClonixConfig 6.1, and the firmware that more or less automatically loads, this is NoVRAM-H. The source seems to have a fix to a bankswitching issue.

Actually just tried XFA and HEPAX prompting functions in PRGM mode, and these show the same erratic behavior.

[Sylvain Cote]

(09-07-2023 11:31 AM)MeindertKuipers Wrote:  

(09-07-2023 11:09 AM)Sylvain Cote Wrote:  What ClonixConfig version did you used ? (v4.2 2014 or v6.1 2020)
What firmware version did you used ? (pre 2020 bank switching strategy or the new one)

I am using ClonixConfig 6.1, and the firmware that more or less automatically loads, this is NoVRAM-H. The source seems to have a fix to a bankswitching issue.
Actually just tried XFA and HEPAX prompting functions in PRGM mode, and these show the same erratic behavior.

Hello Meindert,

Thanks for the reply and for the in-depth analysis that you did.

After my article on HP-41C Bank Switching Schemes in 2020,
Häkan started an email discussion with Diego, Monte, Ángel, Robert and I, regarding NoV/Clonix page 4 bank switching scheme issues with new ROM images.
Unfortunately I do not think that you were included in the exchange, but I remember having validated with you the MLDL2000 bank switching scheme.

Historically page 4 was only used for take-over ROM and for the IL printer disabled mode,
then Ángel started to use it for his Library4,
then Monte followed with his own page 4 library
and of course, Ángel came back with a bank switched page 4 library.
This last step broke the Clonix/NoV bank switching scheme.

In the original PIC code, a bank switch request were switching all the pages at the same time, the ones that was under the control of the Clonix/NoV module of course.
The new PIC code (2020+) follow the more common way, system pages (0...7) are individually switched and port pages (8...15) are port switched.

HEPAX 1D was made 1987/88, do not use page 4 bank switching, use bank switching (4 banks) and page remapping. [works with all 41 OS]
Ángel's HEPAX 4H is fairly new, use page 4 (LIB4) and use bank switching (4 banks), but not sure if t has page remapping. [only works with a CX OS based calc, I think]

I cannot do it right now, but I will load and test HEPAX 1D with the old and new page switching firmware tonight (UTC-4h) and will come back with the results.

Sylvain

Edit: I did a fast test,
"HEPAX 1D" with pre-2020 NoV firmware works, but
"HEPAX 1D" with 2020+ NoV firmware has the issue.

[kj.vandelooij]

Hello all,

I can confirm Meindert's findings of:

1) the @@@... BCAT item for block 9. That is: this string appears sometimes. In my case @@@... strings also show up in CAT 2 in 'place of' each module function.

2) Similar behavior in the case of XFA and HEPAX - when programmed.

My versions: ClonixConfig 6.1, HEPAX-1D

Cheers,

Kees Jan

[Ángel Martin]

(09-07-2023 12:41 PM)Sylvain Cote Wrote:  

(09-07-2023 11:31 AM)MeindertKuipers Wrote:  I am using ClonixConfig 6.1, and the firmware that more or less automatically loads, this is NoVRAM-H. The source seems to have a fix to a bankswitching issue.
Actually just tried XFA and HEPAX prompting functions in PRGM mode, and these show the same erratic behavior.

Hello Meindert,

.......

Historically page 4 was only used for take-over ROM and for the IL printer disabled mode,
then Ángel started to use it for his Library4,
then Monte followed with his own page 4 library
and of course, Ángel came back with a bank switched page 4 library.
This last step broke the Clonix/NoV bank switching scheme.
..........

Ángel's HEPAX 4H is fairly new, use page 4 (LIB4) and use bank switching (4 banks), but not sure if t has page remapping. [only works with a CX OS based calc, I think]

I cannot do it right now, but I will load and test HEPAX 1D with the old and new page switching firmware tonight (UTC-4h) and will come back with the results.

The picture gets denser as more and more advanced solutions are released, let me try to shed some light on this.-

First, it is correct that the HEPAX 4H needs the Library#4 for *some* cases, so it is recommended to always have the Library#4 configured when the HEPAX 4H is used. And as a consequence, the CX OS would eventually be also needed because the Library#4 requires it (more abut this below).

But I believe Kees is having these issues with the HEPAX-1D, which needless to say has nothing to do with the Lib#4 - at all!. It should work on any 41C model, no Lib#4 required.

As to the Bank-switched version of the Library#4 - this is a recent release that was put together initially to be used with the MAXX modules. It removes the dependency from the CX OS (which is a good thing), but the price paid is that some CX routines were added to the Library, in a secondary bank (the main page was already full). Incidentally, I believe the current BS scheme on NoVo's does support page#4 BS as well, but I might be wrong?

Hope this clears the picture a bit...

ÁM

[Sylvain Cote]

Hello Ángel,

Regarding HEPAX 4H ...

My comment about only running on 41CX was made in the context of ClonixConfig software.

My understanding is that to run HEPAX 4H on a HP-41C/CV you need the following ROMs:

Code:

XLIB4_B1.ROM ; XLIB4_B2.ROM ; HEPX4H_1.ROM ; XHEPX_B2.ROM ; HEPX4H_3.ROM ; HEPX4H_4.ROM

My understanding is that to run HEPAX 4H on a 41CX you need the following ROMs:

Code:

LIBRARY4.ROM ; HEPX4H_1.ROM ; HEPX4H_2.ROM ; HEPX4H_3.ROM ; HEPX4H_4.ROM

Unfortunately, right now ClonixConfig current HEPAX 4H config only support the 41CX setup not the 41C/CV setup.
I have not yet check if it is possible to create a custom config that would run on a 41C/CV.

ClonixConfig (v4.2 & 6.1 2020) current HEPAX 4H option only support these ROMs:

Code:

LIBRARY4.ROM ; HEPX4H_1.ROM ; HEPX4H_2.ROM ; HEPX4H_3.ROM ; HEPX4H_4.ROM

As I wrote before, the new Clonix/NoV bank switching scheme (2020+) should support independent page 4 bank switching.

Regarding HEPAX 1D ...

ClonixConfig v4.2 2015 HEPAX 1D uses the following ROMS and the old bank switching scheme (pre-2020)

Code:

HEPX1D_1.ROM ; HEPX1D_2.ROM ; HEPX1D_3.ROM ; HEPX1D_4.ROM

ClonixConfig v4.2 2020 HEPAX 1D uses the following ROMS and the new bank ClonixConfig v6.1 2020 HEPAX 1D uses the following ROMS and the new bank switching scheme (2020+)

Code:

HEPX1-1E.ROM ; HEPX2-1D.ROM ; HEPX3-1D.ROM ; HEPX4-1D.ROM

More testing is needed, but my basic tests hint that a module build with CC2015 works as expected and the one build with CC2020 have issues.
I still need to compare the ROM's and the ASM files between these versions, but my free time is sparse and I have to focus on my HHC presentation and on MAXX testing.

[J-F Garnier]

(09-07-2023 12:41 PM)Sylvain Cote Wrote:  I cannot do it right now, but I will load and test HEPAX 1D with the old and new page switching firmware tonight (UTC-4h) and will come back with the results.

Edit: I did a fast test,
"HEPAX 1D" with pre-2020 NoV firmware works, but
"HEPAX 1D" with 2020+ NoV firmware has the issue.

This is a resurgence of an old HEPAX emulation quirk, found in many emulators in the past.

As pointed out long ago, the "problem" can also be reproduced on a real 41C/HEPAX system with the help of the HP-IL devlpt ROM.
(or on latest V41 if you don't have the physical modules - V41 is now very accurate).

With the HP-IL dev ROM and HEPAX installed, do XEQ "ON",
- then enter a XF or alike command in a program, and check the entered line,
- or in run mode, execute HEPAX 000, then try to CATALOG the HEPAX functions (particularly funny on a halfnut 41CX)

J-F