XT-IDE on PCjr

[jmetal88]

I did get a valid signed quote back from State Electronics for AMP 532773-2 (60 pin stacked box connector) $8.90 and AMP 532771-2 (60 pin shroud) $0.75. I think I might order 25 of each to have on-hand. I have another request for quote out. Going to wait for it first. Would save tearing up side-cars.

That's awesome, dude! It would make them a lot easier for other people to put together. I couldn't find both connectors anywhere!

[alanh]

The other quote from Aesco came back as $11.65 & $2.70 respectively. So I ordered 25 from State Electronics. Not sure how long it will take to get in or if it will actually be the real deal since Amp hasn't made them in well over a decade. I'm going to cross my fingers anyway. At least this will make it easier not just for this build but if we wanted to do another side-car design.

Now if I could just find a place to do color matched plastic we'd have it all! The back plates are an easy get from frontpanelexpress.com.

[alanh]

Got the connectors! And they are indeed the real deal. Brand new in a Tyco box even. I have 25 now. Of course being 2 separate part numbers, there are no screws. But that's pretty minor.

[alanh]

Few items for general discussion:

A) It appears as if the sidecar bus buffers are floated during accesses to D0000->EFFFF. Nothing I seem to do will return valid data in those ranges. I don't suspect anything wrong with the board at this point. Everything else seems to work fine including memory fill up to 736K and within the C0000->CFFFF range. So this may put an end to the idea cartridge images could be selected from the boot menu and shadowed from paged flash into a RAM fill in the cartridge regions. It also ends the idea of depopulating the ROM chips and serving up a modded system bios from the side-car. And it also diminishes the general usefulness of a full 512KB extra RAM chip. Alliance makes a 128KB. And a few bucks could be saved using that one. But this may be the end of UMBs on the PCjr too. Very disappointing

B) I used the new super cart to replace the system BIOS with one I modded. I changed the main memory check loop to count up to 736KB and adjusted the lower checksum to match. When it gets to 736K, the screen goes magenta and the system halts. If I back the check test down to 704KB, everything works. It counts up to 704 and DOS reports 704. I can peek/poke around in debug and everything looks fine all the way up to b7FFF. Any idea what could be going wrong?

-Alan

[southernbob]

It has been awhile since I played with this PCJR stuff so I may be wrong. I think the video buffer
begins at B8000. Since 736 x 1024 = 753664, converted to hex = B8000, so I think you are running
into the video buffer.

Also, not being able to access the cartridge space, this may not help you but maybe it will give
you some ideas. I wanted to be able to write/read cartridge space D0000 - EFFFF so looking
at hardware drawings in the Tech Ref I saw that the logic only selected this space on a read
so I put a "patch" on the motherboard that got the -MEMW from another module and thru
some logic OR'ed with -MEMR signal and and put that output into -MEMR pin (after cutting
the pin connection to the board) at module ZM47 (System board logic page 2 of 17- lower
left hand- on the page.) This allowed me to make up a card with 128K static RAM that
plugged into a cartridge slot and I was able to do read/writes to this space.
What I modified allowed the CS's to the cartridge slots for both reads & writes so that
probably won't help you with your problem but there may be somthing similiar for
the Bus Address.

Bob

[alanh]

As far as I can tell, both memory strobes going to the side car connector off the main board either never go active on memory reads or writes to D0000->EFFFF or the data bus transceiver never gets enabled. There is a ROM chip on the board that sources bus transceiver enables based on the upper address lines. I still need to confirm via the schematic, but it doesn't look like it ever gets gated.

I might improve upon the super cartridge design to allow it to work in conjunction with the JR-IDE side car and it's BIOS. Instead of dip switches selecting the cartridge image, a specific read pattern could similar to the way the RTC cartridge options work. So you plug it in, it presents a boot menu of everything on the cartridge, then you can select which image you want to activate and jump.

[alanh]

As far as I can tell, both memory strobes going to the side car connector off the main board either never go active on memory reads or writes to D0000->EFFFF or the data bus transceiver never gets enabled. There is a ROM chip on the board that sources bus transceiver enables based on the upper address lines. I still need to confirm via the schematic, but it doesn't look like it ever gets gated.

I might improve upon the super cartridge design to allow it to work in conjunction with the JR-IDE side car and it's BIOS. Instead of dip switches selecting the cartridge image, a specific read pattern could similar to the way the RTC cartridge options work. So you plug it in, it presents a boot menu of everything on the cartridge, then you can select which image you want to activate and jump; without hardware mods. It would require an on-board MCU though. I don't think a PLD would cut it. But the upside with an MCU is you could have an SD card slot on the user facing side - or USB port.

[jmetal88]

It has been awhile since I played with this PCJR stuff so I may be wrong. I think the video buffer
begins at B8000. Since 736 x 1024 = 753664, converted to hex = B8000, so I think you are running
into the video buffer.

Ah, yeah, that could be.

Alan, what do you think would happen if you stopped the detection just one byte short of 736k?

[alanh]

That's not the reason. I can read/write all the way up to B7FFF in debug where the post memory check should stop. Even if it over ran by a byte, the worst that would happen is a corrupted top left character. I'm not sure why changing the stop check value from A0 to B8 fails but B0 works.

[jmetal88]

Hmm. Yeah, that's perplexing. B0 is the highest you can set the stop point? Well, I guess that makes sense, considering the PCjr BIOS checks the memory in 32k chunks.

I wish I knew more about BIOS programming so I could come up with some ideas on this.