DMA ISSUE ON H4

[Icky]

Similar with the MFP HDINT signal..

Here is the HDINT - YELLOW line is Pin 8&9 and BLUE is Pin 10

DS1Z_QuickPrint14.png (40.78 KiB) Viewed 4730 times

DS1Z_QuickPrint15.png (40.64 KiB) Viewed 4730 times

[exxos]

BTW: I know that the 74LS245 buffer has been ruled out for this particular problem, but I still wonder how it is ever supposed to work. It uses /HDCS as its enable signal. But /HDCS is not asserted (= constantly high) during the actual data transfer between device and host. /HDCS is only used during the command and status phases of an ACSI transfer, not during the data phase. Therefore, no data can be transferred from or to the device the way the buffer is wired. What am I missing?

I kinda assumed hard drive chip select was always active when the hard drive was. So may need to look into that more if your saying otherwise ?

[exxos]

@Icky Looks like it was the 1772 buffer we tried ? I thought it was the DMA buffer

https://www.exxosforum.co.uk/forum/viewt ... 260#p11631

On the STE nothing actually drives the DMA port buffers, its hard-wired enabled. I was thinking at the time to invert FDCS to drive the DMA buffers, but then saw HDCS and assumed it was just a inversion of FDCS anyway

[czietz]

I kinda assumed hard drive chip select was always active when the hard drive was. So may need to look into that more if your saying otherwise ?

Please see this scope screenshot, /HDCS while reading data from the hard drive. You can see it pulsing low six times in the beginning (command phase) and once at the end (status phase). EDIT: It's barely visible in the screenshot, unless you view it at full resolution.
Inbetween is the data phase where the actual bytes are read from the hard drive and where /HDCS stays high. (The noise on /HDCS is caused by crosstalk from other lines on the ACSI cable I used.)

DS1Z_QuickPrint1.png (28.6 KiB) Viewed 4718 times

[Icky]

@exxos - I'll hard wire enable the buffer, although I did jumper over and it did not work. BUT lets hard enable it and go from there.

[czietz]

Careful: Hard-wire enabling the ACSI buffer might cause problem when reading from the floppy drive, because in that case two buffers drive the data bits towards the DMA. In the STE (at least according to the schematic I have) the ACSI buffer is enabled by an inverted /FDCS plus a flip-flop for synchronization.

[exxos]

I think its probably best to remove the DMA buffer for now and jumper link it over until I have more time to look into it all.

[exxos]

Careful: Hard-wire enabling the ACSI buffer might cause problem when reading from the floppy drive, because in that case two buffers drive the data bits towards the DMA. In the STE (at least according to the schematic I have) the ACSI buffer is enabled by an inverted /FDCS plus a flip-flop for synchronization.

I guess HDCS should be called "Hard Drive Command Select" really then. and FDCS is "floppy Drive Chip Select" .

I was trying to avoid using a inverter on FDCS, which is why I just went with HDCS thinking it would be a inversion of FDCS and also it would actually disable both FDCS & HDCS when there is no access at all. Though I never had chance to look into that anyway as wasn't exactly important if both lines were disabled at the same time or not.

[PaulJ]

I found a excellent description of the ASCI transfer.

[url]http://dev-docs.atariforge.org/files/AC ... 91.pdf/url]

Going to continue the investigation today, going to remove IC6 and jumper the data, maybe give the HP LA a try. Will be reading thread with interest.

[czietz]

Hm, the link only leads to a page titled "Atari History and Evolution of Atari Computers- Atari Fans Club" written by whoever snatched atariforge.org. (EDIT: ... seemingly to promote some online casino. ) Better try the current page: https://docs.dev-docs.org/htm/search.ph ... A+Integrat