HP Portable Plus ... until 2039?

[J-F Garnier]

There have been several discussions on the HP Portable Plus recently ( here , here , here and there ) and I'm very happy of this because I like this machine that I'm using mainly as a super battery-powered HP41CX-display-RAMdisc-HPIL-RS232 integrated system.

A few weeks ago, I rebuilt the battery pac and I had to restart the Portable, and set again the date.
And I noticed that the date can't be set after 2039. Arg! Only 25 years left! What will I do then?

BTW when setting the date from the PAM, it's now faster to roll down from 2039 than from the 1980 starting date...

J-F

[Michael Fehlhammer]

Hello J-F,
you are right, that's a pity. But you can see it that way: 25 years left to rewrite the operating system!

While doing this there would be a good opportunity to upgrade from DOS 2.11 to 3.3 or, even better, to 5.0, thus opening a whole new world of software for the PP.
Another drawback of the PP is its lack of IBM compatibility due to its proprietary BIOS. I couldn't make any of the the programming environments like Turbo BASIC/C or Quick C run. Would a simulation of the PC-BIOS be imaginable? Or does typical PC software ( like the ones mentioned above ) circumvent not only DOS but also defined BIOS calls, addressing compatible PC - components directly?
In that case we would need an add-on board ...

[Jim Horn]

The 110 Portable and Portable Plus still have a warm spot in my heart - I used my Portable for years as a preferred note taker and word processor. While limited compared to later machines, its ~20 hour battery run time and superb keyboard made it really handy for that.

When Sam Chau first engineered how to modify the Portable's motherboard to give it 512k of RAM instead of the standard 272k, we figured we should be able to modify PAM's routines that set the partition between user RAM and Drive A:. But we were surprised to learn that HP had lost the source code, making doing so difficult at best. Sam ended up writing his own disk driver routine to use the extra 240k as drive B:.

We joked that the Portable was "PC Compatible" in that it was theoretically possible to write a program that ran on it and on an IBM PC but that it had never been done. That was a sarcastic stretch as many DOS utilities ran fine. The trick was to use the proper BIOS calls rather than any attempt to access hardware directly.

The macro assembler HP sold as an option worked well too, judging from Sam's work with it.

Do enjoy your Portable family machines! Mine has since been given to one of the very active Forum members along with the many binders of documentation. May they all continue to serve you well!

[Jim Horn]

Oh - if you're in need of higher speed, we found that replacing the Harris 80C86 in the Portable with an NEC V30 series CPU combined with a minor cut and jumper change to run it at 8MHz instead of 5.333MHz roughly doubled the speed of the Portable without noticeable decrease in battery life. It's a far simpler mod than the memory expansion I mentioned (which involves removing all 34 8kx8 SRAMs and replacing them with 16 32kx8 SRAMs, among other things).

[Jim Horn]

Sorry for the rambling earlier. Why do they only allow years to 2039? What we did find in perusing the PAM binaries was that instead of computing and displaying values when varying the partition between user RAM and drive A:, it had a set of all possible values and associated strings in a table. So I strongly suspect the limit of 60 years for the calendar setting is the size of the table they used, not the hardware or other software that actually uses it.

If that's the case, then the PAM code could be reverse engineered and modified to cover a longer range (hard) or a utility written to set the year directly (less hard - still requires figuring out where the year data is). Since the time is kept on the Portable by a dedicated separate microprocessor when it's powered off, the protocol used to do that may be tricky to figure out. Maybe someone has already done that in the 30 years since it came out. But if not, at least the schematics and more are available to help.

[Manolo Sobrino]

Alternatively, on the 1st of January of 2040 you can move the date back to 01/01/1984, that would do until 2096 (two Julian 28y solar cycles) because 2000 was a leap year. 2100 won't be... but I wouldn't worry about it.

[Dave Frederickson]

(05-09-2014 06:58 AM)J-F Garnier Wrote:  And I noticed that the date can't be set after 2039.

I read that somewhere, too. However I have no problems setting my clock well past that date.



Am I missing something?

Dave

[Dave Frederickson]

Okay, I see the bug with PAM. It's documented in this file on Jean-Francois' site.

Dave

[J-F Garnier]

(05-09-2014 08:19 PM)Jim Horn Wrote:  ... So I strongly suspect the limit of 60 years for the calendar setting is the size of the table they used ....

I thought it could be related to the year 2038 bug mainly found in Unix systems.
Anyway, as Dave noted, the workaround is not to use PAM but the DOS date command.

J-F

[Dave Frederickson]

(05-09-2014 07:56 PM)Jim Horn Wrote:  Mine has since been given to one of the very active Forum members along with the many binders of documentation.

Perhaps that Forum member could be persuaded to share the contents of the binders. I know of a museum that would be happy to host the documents.

BTW, Jim, was the code name for the Portable PLUS 'Buzzard', by any chance?

Dave

[Dave Frederickson]

(05-09-2014 08:02 PM)Jim Horn Wrote:  Oh - if you're in need of higher speed, we found that replacing the Harris 80C86 in the Portable with an NEC V30 series CPU combined with a minor cut and jumper change to run it at 8MHz instead of 5.333MHz roughly doubled the speed of the Portable without noticeable decrease in battery life. It's a far simpler mod than the memory expansion I mentioned (which involves removing all 34 8kx8 SRAMs and replacing them with 16 32kx8 SRAMs, among other things).

Sam Chau's description of the mod can be found in this issue of The Portable Paper. I'm considering modding one of my PLUSes.

Dave

[Etienne Victoria]

Well... I've found a D70116D-10 NEC in DIP-40 on TAS but it is 10 Mhz and not 8 Mhz as advised in the article.

I'll keep investigating.

E.

[Etienne Victoria]

Ok, there are also NEC D70116D-8 at 8Mhz, still in DIP 40 at around 6 GBP for 2 shipping included from Greece.

I think these will work and bought 2.

I will report on the mod when I have time to perform it on one of my PPs.

E.

[Michael Fehlhammer]

Yes Etienne, Dave, please let us know if you were successful with that modification!
Obviously this NEC V30 CPU comes in different ... hmm ... flavours:
What's the difference between the 70116D and uPD70116C ( both having DIP40 form factor and 8 MHz clock speed )?
Different power consumption, perhaps? What would be best choice for the PP - speed upgrade?

Another question for our hardware experts:
Would it be possible - at reasonable effort - to equip the PP with an 8087 math coprocessor?
I love fast math floating point routines! :-)

[Dave Frederickson]

Okay, I'm going with the uPD70116HCZ-16 which I can find on TAS for $6 ea. This is the high-speed (16 MHz!), low-power version of the chip. I don't expect any issues with using the -16 part below it's rated maximum clock speed.

Dave

[Jim Horn]

Could an 8087 be added to a Portable or Portable Plus? Certainly not easily - there is no provision in its circuitry for the addition. That said, I know that circuit boards that plugged into a CPU socket and had sockets for the CPU and a coprocessor existed for various CPUs through the years. Whether that is possible with the 80C86/V30 and an 8087 and whether such a board would even fit in the given case, I don't know. The power draw of an 8087 (non CMOS) would be a concern as well. Any feedback out there?

[Michael Fehlhammer]

Power consumption really seems to be a major problem.
If I understood data sheets correctly, the 8087 5 Mhz takes 2.4 W, whereas the PP CPU ( Harris 80C86 ) needs only about 50 mW. A CMOS version of the 8087 is not available.
( There are rumors about an 80C87 from Cyrix, but nobody has ever seen such thing ).
Certainly a reason for HP not to provide an FPU socket.

[Dave Frederickson]

(06-08-2014 09:32 PM)Michael Fehlhammer Wrote:  Yes Etienne, Dave, please let us know if you were successful with that modification!

Hi Michael,

I opted for changing just the CPU, leaving the clock circuit unmodified. Supposedly this should yield a 20-40% increase in performance. The Norton SI benchmark increased from 1.3 to 2.8. Some self tests report failures as reported in the Portable Paper, but this is due to the more efficient CPU and all the hardware works fine including a 1MB Memory Drawer.

Dave