HEPAX - from the treasure chest

[SteenPetersen]

All,

KimH stopped by today and I borrowed a working HEPAX module (1B) - thank you Kim! We can read the content of my disks, so he has taken them with him to make a copy and so we can get the contents online, which I am not able to. I will try to find out/remember what is on each image and hopefully we can find this out together.

I managed to read the BOOTROM image from my tape - it has the following content:
- BOOT ROM A
BCAT
CLFCN
CLROM
CODE
COPYROM
DECODE
DECOD1
DECOD2
DECOM3
DECOM4
DELETE
FIND
HEXKEY
HEXLIST
HEXLOAD
INSERT
MCP
MCSCAN
P>ROM
PP>ROM
RAMTEST
READROM
REG>ROM
ROM>REG
ROM>X
ROMTEST
WRTROM
X>ROM
AND
OR
RXL
RXR
SXL
SXR
X+Y
XOR
Y-X
1CMP
2CMP

I am not 100% sure if this is something I got from somewhere else or if it is the first set of tools in my tool box - outside the ASM3 which I used as assembler. Anyone else who has seen this before?

Thanks,
Steen

PS: I will use this thread to talk about what I dig out of my treasure chest...

[Diego Diaz]

Hi Steen.

Good news indeed that you and Kim managed to recover these files from your disks and tapes.

Sure every M-coder here (I'm not in this selected group) will be very happy to have a look at such content when available.

At first glance the catalog listing you've shown looks like a customized developer's tools box.

More authoritaive voices will sure offer a deeper analysis and insight about this.

Best wishes and congratulations for your latest findings. :-)

Diego.

[J-F Garnier]

Hi Steen,

I'm particularly interested in the hardware aspects, if you can remember the details. How did you move from the breadboard to an IC? What kind of IC is it? Did you program it by yourself or was it done by the IC manufacturer?

Some years ago (well 10 years already!), Diego and I found a nice little secret of the Hepax, that is the Standard and Advanced modules are only differentiated by a very tiny detail... That was very clever, did you customize the boards depending on demand (if you can remember)?

J-F

[SteenPetersen]

Hi J-F,

it looks like the epoxy was not strong enough ;-)

We did not have a lot of money, so we only created one model of the ceramic substrate. We needed to be able to produce four - actually five - different versions of the module with only one substrate. So we got the small company who assembled the modules to "program" the modules before assembly. This was done by cutting at exactly the right spot before applying something which was very hard to remove - I think it was epoxy, but I am not sure. I think it was a laser doing the cutting, but I do not remember - I was busy coding and did not see where this was actually done.

I did leave you a small clue together with the picture I posted of the breadboard - the instruction for the DIP switch settings - switch #8 controls the 8K/16K setting. As you can see there are no DIP switches on the real module, so here we have to do a permanent cut instead. So basically the hardware was identical for all modules, except no EPROM on the MEM modules. By the way - the fifth version was a ROMONLY device for custom usage (I think without the bank switching - there is a control point for that as well) - unfortunately we never got to sell that.

As far as I remember then the chip was a custom gate array made by Fujitsu (we worked with a local danish company - Nordisk Elektronik - who had the relationship with Fujitsu). The chips were soldered at Bruel & Kjaer, while the connector (which we bought from HP) was soldered by the small company who also made the plastic moulding (as far as I recall, but I may be wrong). Here the final assembly also took place - with all the glue ;-)

Regards,
Steen

[Ángel Martin]

Thanks for sharing those details, made me appreciate even more the process you followed during the production of the modules. All those decisions and resulting implementation reaching the appropriate compromises and solving the inherent limitations & financial constraints can only be defined as engineering at its best!

I've always maintained the HEPAX was the "Holy Grail" of the HP-41 world ;-) If somebody thinks that's an exaggeration just consider a few starting points: The first and only MLDL in a plug-in ROM footprint, unparalleled 4-banks implementation never else seen, fully capable to self-relocate to the most optimal configuration, compatible with all HP-41 variants, with a duplicate set of X-Functions arranged in an auxiliary FAT (the original sub-functions!), with the most capable Disassembler and ROM HEX Editors... do I need to go on?

Cheers,
ÁM

[Sylvain Cote]

I agree with everything you said Ángel and more, just a small precision: "The first and only MLDL in a plug-in ROM footprint until NoVRAM, ..."
Sylvain

[Diego Diaz]

Hi all,

First of all, sorry for tampering the "blue spot secret"...

Now seriously, it's great to have all these details and insight on the manufacturing process.

The ceramic PCBs were (still are) really expensive, but certainly a good decision.

I'm surprised about the connectors though, since they are not similar to those in the standard HP modules; I thought you (VM) got them made from a third party manufacturer... much like the plastic housings.

Look forward to reading on the next chapter of the "HEPAX Trasure Chest Adventure" :-D

Best wishes.

Diego.

[Ángel Martin]

(09-22-2020 12:06 PM)Sylvain Cote Wrote:  I agree with everything you said Ángel and more, just a small precision: "The first and only MLDL in a plug-in ROM footprint until NoVRAM, ..."
Sylvain

Sure, almost 30 years later ;-)
Technology has changed a bit since - and I don't mean to minimize the NovRAM merit of course.

[Diego Diaz]

Thanks both for the credit. :-)

First, it's obviuosly certain that NoVs modules couldn't have been possible without the HEPAX.

And though my designs may have provided a second (and prosperous) life to the concept and even expanded some of their features; these achievements, made with 21st century resources are by no means comparable to creating such module from scratch with late eighties available technology.

Cheers!

[KimH]

I could read the floppies - no errors!

My dayjob is getting in the way of moving the files form Floppy to something usable on this site.

I have a plan, but need a couple of hours to make sure I don't ruin the disks while making copies and move files through ILPer to actual postings. This will be over the weekend.

/KimH

(09-21-2020 07:38 PM)SteenPetersen Wrote:  Hi J-F,

it looks like the epoxy was not strong enough ;-)

We did not have a lot of money, so we only created one model of the ceramic substrate. We needed to be able to produce four - actually five - different versions of the module with only one substrate. So we got the small company who assembled the modules to "program" the modules before assembly. This was done by cutting at exactly the right spot before applying something which was very hard to remove - I think it was epoxy, but I am not sure. I think it was a laser doing the cutting, but I do not remember - I was busy coding and did not see where this was actually done.

I did leave you a small clue together with the picture I posted of the breadboard - the instruction for the DIP switch settings - switch #8 controls the 8K/16K setting. As you can see there are no DIP switches on the real module, so here we have to do a permanent cut instead. So basically the hardware was identical for all modules, except no EPROM on the MEM modules. By the way - the fifth version was a ROMONLY device for custom usage (I think without the bank switching - there is a control point for that as well) - unfortunately we never got to sell that.

As far as I remember then the chip was a custom gate array made by Fujitsu (we worked with a local danish company - Nordisk Elektronik - who had the relationship with Fujitsu). The chips were soldered at Bruel & Kjaer, while the connector (which we bought from HP) was soldered by the small company who also made the plastic moulding (as far as I recall, but I may be wrong). Here the final assembly also took place - with all the glue ;-)

Regards,
Steen

[SteenPetersen]

Thank you all for the kind words! It is nice to see 30 years later that we did do a good job back then.

Regarding the connectors, then I am quite sure we bought the actual raw goldplated connectors from HP, but without any plastic - so they were placed inside our plastic mold. We quickly determined that we would not be able to get someone to manufacture the connectors with the correct elasticity to get the right electrical connectivity. This would have been trial and error without assurance it would work in the long run.

Regards,
Steen

[Diego Diaz]

Ok... I misunderstood that point (about the connectors). Now it's clear and have much more sense this way.

Thanks! ;-)

[KimH]

As a teaser here is the CAT :MASSMEM I did on the 9114B with my 71B.

I will do one BOLD for readability ASAP

CAT - HEPAX Floppies

(09-22-2020 05:41 PM)KimH Wrote:  I could read the floppies - no errors!

My dayjob is getting in the way of moving the files form Floppy to something usable on this site.

I have a plan, but need a couple of hours to make sure I don't ruin the disks while making copies and move files through ILPer to actual postings. This will be over the weekend.

/KimH

(09-21-2020 07:38 PM)SteenPetersen Wrote:  Hi J-F,

it looks like the epoxy was not strong enough ;-)

We did not have a lot of money, so we only created one model of the ceramic substrate. We needed to be able to produce four - actually five - different versions of the module with only one substrate. So we got the small company who assembled the modules to "program" the modules before assembly. This was done by cutting at exactly the right spot before applying something which was very hard to remove - I think it was epoxy, but I am not sure. I think it was a laser doing the cutting, but I do not remember - I was busy coding and did not see where this was actually done.

I did leave you a small clue together with the picture I posted of the breadboard - the instruction for the DIP switch settings - switch #8 controls the 8K/16K setting. As you can see there are no DIP switches on the real module, so here we have to do a permanent cut instead. So basically the hardware was identical for all modules, except no EPROM on the MEM modules. By the way - the fifth version was a ROMONLY device for custom usage (I think without the bank switching - there is a control point for that as well) - unfortunately we never got to sell that.

As far as I remember then the chip was a custom gate array made by Fujitsu (we worked with a local danish company - Nordisk Elektronik - who had the relationship with Fujitsu). The chips were soldered at Bruel & Kjaer, while the connector (which we bought from HP) was soldered by the small company who also made the plastic moulding (as far as I recall, but I may be wrong). Here the final assembly also took place - with all the glue ;-)

Regards,
Steen

[rprosperi]

Thanks for sharing these, it is a nice tease!

Some familiar ROMs in there, but also some unfamiliar but quite interesting looking names.

Maybe the ASCII text file includes notes on the files?

I assume the -8064 files are 41 Program files, from the names and small file sizes.

[Dave Frederickson]

(09-29-2020 02:47 PM)rprosperi Wrote:  I assume the -8064 files are 41 Program files, from the names and small file sizes.

Loading the LEX file FILESEXT will display friendly file types for non-71 files.

The program MCOPY will copy multiple files from a floppy to a LIF disc image, where it can easily be shared.

[Diego Diaz]

Hi all,

That catalog, and its files' content (when available), will sure fill a number of the upcoming autum -and winter- evenings for several users (me included) here... ;-)

Thanks a lot for sharing!

Diego.

[SteenPetersen]

All,

this morning I received a tape from KimH with many of my old development images. I got power on my tape drive battery and managed to load the two first images - see below. These appear to be some of my first tools. More to follow.

Regards,
Steen


SP22JUN
-SP RAM A
RRMTEST: ROMTEST program – tests a ROM Image. For a RAM image the Checksum is calculated and stored.
RMMTEST: RAMTEST – tests a RAM page
RMMTOG: RAMTOG – toggles Write Protection of a RAM page
HEXEDIT
INSERT
DELETE
COPYROM
CLRAM

SP25JUN
-SP RAM 0A
ROMTEST
RAMTEST
RAMTOG
HEXEDIT
DELETE
INSERT
COPYB
CLRAM
HPROMPT
CTRAST
ROTYX

[rprosperi]

Thanks for sharing these teasing little peeks Steen!

My first reaction is that if you could add 3 new functions (HPROMPT, CTRAST, ROTYX) in only 3 days, you must have been pretty busy and certainly productive.

[Massimo Gnerucci]

(10-15-2020 01:49 AM)rprosperi Wrote:  Thanks for sharing these teasing little peeks Steen!

My first reaction is that if you could add 3 new functions (HPROMPT, CTRAST, ROTYX) in only 3 days, you must have been pretty busy and certainly productive. :)

Don't forget COPYB. ;)
And poor COPYROM.

[rprosperi]

(10-15-2020 06:42 AM)Massimo Gnerucci Wrote:  

(10-15-2020 01:49 AM)rprosperi Wrote:  Thanks for sharing these teasing little peeks Steen!

My first reaction is that if you could add 3 new functions (HPROMPT, CTRAST, ROTYX) in only 3 days, you must have been pretty busy and certainly productive.

Don't forget COPYB.
And poor COPYROM.

Actually I assumed COPYROM was renamed to COPYB (perhaps for CopyBlock), but later eventually renamed back to COPYROM, as released.