IDE Interface for TT 030

[pakman]

Schematics for an IDE interface for Atari TT 030.
Data lines are "twisted" according ppera.

Plugs onto the Fast-RAM connector (Fast-RAM can still be used) plus 2 external wires (/DSACK1 and /INT).

[Atari030]

I did wonder if it was possible to make a VME IDE controller for the MEGASTE/TT030. That would be pretty cool. IIRC some of the lines required are not available so it would have to have some lines run from the mainboard. So many possibilities.

[AtariZoll]

I did wonder if it was possible to make a VME IDE controller for the MEGASTE/TT030. That would be pretty cool. IIRC some of the lines required are not available so it would have to have some lines run from the mainboard. So many possibilities.

Popsel did it already years ago. He needed to "patch" some lines, because IDE address space is not accessible on VME connector.

[AtariZoll]

Schematics for an IDE interface for Atari TT 030.
Data lines are "twisted" according ppera.
Plugs onto the Fast-RAM connector (Fast-RAM can still be used) plus 2 external wires (/DSACK1 and /INT).

Cool. And since there is IDE autoboot support in TOS 3.06, it needs to be patched for twist, I guess .

[qq1975b]

A TT IDE interface is very interesting! Is it fast? Is someone going to build this?

[pakman]

A TT IDE interface is very interesting! Is it fast? Is someone going to build this?

There are only 2 prototypes: One is handwired on a veroboard, the other one with a professional made PCB.

Speed is not so bad:

XFERRATE.jpg

[qq1975b]

thanks. Not as fast as SCSI but I think it is fast enough
It will solve one of the TT's most important lacks

[pakman]

thanks. Not as fast as SCSI but I think it is fast enough

How fast can you go with SCSI at the TT?

[AtariZoll]

As I measured, SCSI on TT is max some 2.7 MB/sec . No further comment

[frank.lukas]

XFERRATE measured 1,9MB/s max with a fast IBM or Seagate SCSI-3 big Harddrives or SCSI CF Adapter ...

[qq1975b]

my mistake then, sorry. I thought it was around 7MB/s. Your IDE driver is really fast. In any case I find the SCSI on my TT fast enough and it should be around 2MB/s. I will look for this Xferrate prg.

[frank.lukas]

pakman are using XferRate -> http://www.anodynesoftware.com/freeware/main.htm

[alexh]

I presume the normal SCSI->IDE bridges work with the TT? The prices of them have gone through the roof just recently. You used to be able to get an Acard AEC-7720U for under $80 and they are now more like $250!

There is a SCSI version of CosmosEx being released for the TT which while expensive has lots of additional features (if you need them) such as the USB keyboard, mouse and Joystick adapter.

http://www.atari-forum.com/viewtopic.php?f=103&t=28005

[AtariZoll]

Alexh - your post really not helping here.
This IDE IF for TT is brilliant, and costs only few bucks. Ideal for people who want to build it self and install in TT.
We have longer thread here in forum about IDE in TT, and expectable speed. Now it is reality. No other IF capable of such, or even close speed.
Case of cheapest and fastest

[arf]

Schematics for an IDE interface for Atari TT 030.
Data lines are "twisted" according ppera.

Plugs onto the Fast-RAM connector (Fast-RAM can still be used) plus 2 external wires (/DSACK1 and /INT).

Woohoo! Thanks a lot for sharing this! I’d love to see one in my TT.

[pakman]

This is how the prototype looks like:

PROTOTYP.jpg

[qq1975b]

Schematics for an IDE interface for Atari TT 030.
Data lines are "twisted" according ppera.

Plugs onto the Fast-RAM connector (Fast-RAM can still be used) plus 2 external wires (/DSACK1 and /INT).

It is really cool you made this affordable IDE card. Thanks.

On the other side there is the male connector to plug the fast ram card, isn't it?

[frank.lukas]

Perhaps someone builds a 512MB or 1GB Fastram with IDE Port option card somedays ?

[arf]

This is how the prototype looks like:

PROTOTYP.jpg

That’s very cool, indeed. Is it compatible to existing harddisk drivers like HDDRIVER or CBHD?

Regards
Götz

[pakman]

Is it compatible to existing harddisk drivers like HDDRIVER or CBHD?

Basically yes, but:

The data lines are twisted (D0..D7 swapped with D8..D15) as proposed by ppera.
This way you can get maximum speed when using a DOS compatible format since you avoid SW byte swapping.
Anyhow, it requires a suitable driver, such as ppera's driver.
I don't know whether CBHD or HDDRIVER can handle this.

To make it fully compatible, you can use a twisted cable, which will in fact "untwist" the data lines again.
Then both CBHD and HDDRIVER will work for sure.

[AtariZoll]

Usual drivers can not handle twisted (swapped Lo/Hi) attached IDE device, because must give commands and parameters on high byte instead low.
So no, Hddriver, CBHD, AHDI, Cecile can not for sure.
Just to add that hardware byte swap is used normally by Amiga IDE adapters.

Perhaps someone builds a 512MB or 1GB Fastram with IDE Port option card somedays ?

With note that minimal allowed hard disk capacity is 1000 GB Seriously, what man can do with so much RAM on TT ? I mean something wise.
Idea would be not bad considering that there is lot of diverse DDR modules around. But logic to make it work well with prehistoric 32 MHz computer will be not simple, I guess.

[Arne]

@Holger:

got a question or two regarding the schematics.
I've seen that you connected the CPLD to D16 and D17. Is the purpose of this to configure the CPLD with self-written software?
Why did you use a combination of '541/'573 and not '245? Does the 68030 end the bus-cycle before the IDE device has a chance to take over the data-lines or what is the reason for the latches?

[pakman]

I've seen that you connected the CPLD to D16 and D17. Is the purpose of this to configure the CPLD with self-written software?

Experimental features..
With move.b $F0003F,D0 you can read back the status of the INTRQ (bit 0) and the IRESET# (bit 1) line.
If the driver SW knows about this, then there is no need to connect the XHDINT line from this board to the DMA port pin 10, as usually.
Another feature is that you can force slow IDE access using the address range $F00040..$F0007F.

Why did you use a combination of '541/'573 and not '245?
Does the 68030 end the bus-cycle before the IDE device has a chance to take over the data-lines?

Exactly. The 573 latches make sure that the hold time for the data on write cycles is always ok.
Without latches I would need to insert one wait cycle on the 68030 16MHz bus.
With latches it runs as fast as possible on the 68030 16MHz bus using DSACK.

[Arne]

Experimental features..

I thought so
Interesting Feature.
Last year I talked to Uwe Seimet and he stated that he does not need the INT line on a ST.
Did you already talk to Uwe about this feature?
If it works smootly it should become a de-facto standard IMHO.

Exactly. The 573 latches make sure that the hold time for the data on write cycles is always ok.

So you can finalise the running bus-cycle and start a new bus-cycle while the latches still hold the data-lines to the IDE port?!?!
You might want to take a look at the 74HCT646. Its a octal bus driver with latches (for both directions). A bit more expensive than a combi of '541/'573 but would make layout a bit easier.
Reichelt sells them.

[AtariZoll]

Experimental features..

I thought so
Interesting Feature.
Last year I talked to Uwe Seimet and he stated that he does not need the INT line on a ST.
Did you already talk to Uwe about this feature?
If it works smootly it should become a de-facto standard IMHO. ....

I was thinking about same years ago. But IDE autoboot in TOS uses mode with IRQ active.
IRQ less mode works well, of course. But I don't see how we can make it standard now, so many years later. And byte-swapped mode should be standard right from start - from simple reason: TOS supports FAT16 with Intel byte order. Unfortunately, it resulted in speed loss because need for SW byte swap when attaching DOS partitioned disks (while there is another byte swap in GEMDOS ) .