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

[kj.vandelooij]

Dear Ángel

Thanks for clarifying and re-focussing the thread. I am indeed experiencing problems with HEPAX-1D.

Greetings, Kees Jan

[kj.vandelooij]

Dear Sylvain

In the meantime I have tried the pre-2020 version of the firmware and I can confirm that the problem described does not seem to arise in that version. This is consistent with your findings.

What would your strategy for testing be? Please let me know if I can help.

Best whishes, Kees Jan

[kj.vandelooij]

Hi J-F,

Thanks! I read the reference you provided. Would this mean that a fix could be applied to the Nov-64d firmware?

Greetings,

Kees Jan

[Diego Diaz]

Hi all,

And sorry for not being more active in this thread. (*Really* busy these days.)

I'll take a deeper look at the comments and test results posted above when I get back to Dom. Rep. next week.

It seems to me that the latest changes in the BS scheme fixes some unwanted behaviour but causes that particular error.

Please feel free to contact me by mail (or post here) with any info/suggestion about how to deal with this bug(?)

The NoV-64d firmware can of course be modified, however I'm now wondering if there may be some incompatibility issues between the requirements for the current BS scheme and the previous ones required for the XF to work as expected.

As J-F has pointed out, this has been around for a long while, hope we can find (build) a consistent solution.

Will keep track of any news and will run my own tests inside the NoV-64d code, but regrettably, it may take a while.

Best wishes.

Diego.

[brouhaha]

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

That's not quite how system page bank switching works in actual HP-made hardware. Third party ROMs not manufactured by HP, should behave similarly, but there's no guarantee.

The Nut CPU is not involved in bank switching. It treats all bank switch instructions as no-ops. Only the ROM chips interpret bank switch instructions.

For actual HP ROMs, including the 41CX mainframe ROMs, the method is that all the ROMs in one physical integrated circuit, the three-page (120 kbit) ROM, honor a bank switch fetched as an instruction from that ROM only, and no other ROM chips pay any attention to it.

41CX pages 0, 1, and 2 are in one 120 kbit chip, and are not bank switched.

41CX page 3 (single bank) and page 5 (bank switched) are in one 120kbit chip. A bank switch fetched as an instruction from either page 3 or page 5 will switch the page 5 bank, but have no effect on any other pages, including pages 4, 6, 7, or ports. In particular, a bank-switched printer ROM such as 82242A infrared printer module in port 6 will not affect the 41CX page 5 or vice versa.

HP 12K ROM modules in ports normally work as you describe, with bank switching affecting the entire port, but this is only because the 120K chip was deliberately configured that way. In principle, an HP 120kbit chip could have 4K port addressed and 8K outside its page (perhaps page 4), or vice versa, and the bank switching would be triggered from any page in that chip.

[Ángel Martin]

(09-10-2023 02:14 PM)brouhaha Wrote:  ......
41CX page 3 (single bank) and page 5 (bank switched) are in one 120kbit chip. A bank switch fetched as an instruction from either page 3 or page 5 will switch the page 5 bank, but have no effect on any other pages, including pages 4, 6, 7, or ports. In particular, a bank-switched printer ROM such as 82242A infrared printer module in port 6 will not affect the 41CX page 5 or vice versa.
.......

This is a clarifying view, however I don't fully understand how to apply it to the 82242A infrared printer module and page#6 case: where's the 120kbit chip in that instance? We know only page#6 bank-switches, leaving page#7 alone of course. I think that's where the "independent page" bank-switching model comes from.

Best,
ÁM