kiki@uhunix1.uhcc.Hawaii.Edu (Jack W. Wine) (05/03/91)
A lot of current discussion concerns the SCSI and a floppy chip replacement, so this probably unoriginal thought occurred to me: replace the WD 1772 chip with an NCR 5380 SCSI chip. I would agree with an attempt at a floppy chip upgrade if SCSI floppy drives were unavailable, but it seems that the logical and physical interface space should be used for an ANSI standard SCSI port. I managed to get an ST Internals book and it appears that the WD 1772 chip can be replaced with an NCR 5380 SCSI chip. The 5380 has 40 pins so an adaptor board is necessary, but it requires almost no support chips. I think it needs a buffer on the data lines from the DMA chip, because TOS will bootup and think it's configuring a WD 1772. Also, it has three address lines to select among its eight registers, so a free sound chip port bit needs to be connected, but replacing the floppy chip also frees up a couple of port bits that are used for drive and head selection. DMA control lines have to be connected, but I don't have enough info on either the 5380 or the ST DMA chips. If anyone has suggestions about this possible project, please let loose your ideas. Also, I read that Sybex has published a new 1300 page book about the STE/TT computers, but their US sales office claims that it doesn't exist. It may have been written by European authors, so if anyone there can furnish me with its ISBN, I would be very happy. Thanks, Jack
kiki@uhunix1.uhcc.Hawaii.Edu (Jack W. Wine) (05/04/91)
In article <5578@wucc.waseda.ac.jp> ytsuji@wucc.waseda.ac.jp (Y.Tsuji) writes: >I just cannot understand why we need to worry which chip to install on >ATARI ST, Wd1772 or NCR??? Let ATARI manage to procure a floppy disk >controller or SCSI controller. Do you think ATARI will listen to us? >I am not so sure that USENET reflects the opinions of the vast majority of >users (games or music people). This project would be primarily for people who don't have the Mega STe/TT, because Atari supposedly will provide some type of floppy upgrade kit for the Mega STe and TT owners. That leaves several million ST owners who pro- bably want a HD or VHD drive, with the alternative of the Dreampark floppy kit you cited in another post, or an embedded-SCSI floppy with a SCSI adaptor. I can think of these advantages to my proposal: 1) Following ANSI specs will allow connection to peripherals 18 ft. (6 m) away. 2) Enhanced reliability because of ANSI conformance 3) Low chip count, compared to commercial SCSI adaptors. 4) Internal design 5) Cheap (~$10 for NCR 5380 and ~$20 for other parts) Some disadvantages: 1) Like Dreampark kit, requires desoldering floppy chip 2) Requires wire wrap equipment (~$10) 3) Interfacing to TOS may be difficult. 4) Case must be modified to install a connector The hardware part appears to be straightforward, but TOS will be a little confused! If anyone has any ideas about adapting it to TOS, please let me know. For most people, the ICD SCSI adaptor would be a good alternative, but for people using their ST's with numerous peripherals which are constrained by the electrical limitations of the DMA port, this project may be a practical solution. The combination of the ICD SCSI and the NCR chip SCSI would make the ST a pretty formidable machine. Jack
kiki@uhunix1.uhcc.Hawaii.Edu (Jack W. Wine) (05/05/91)
>The trouble is that WD1772 is discontinued now. And ATARI cannot possibly >sell anything if they don't get this chip or its equivalent. And there >is no additional cost to put something like FDC92C36, which consumes less >power and accepts high density. Therefore people will be having 1.44 MB >drives when they buy the cheapest ATARI ST very soon. No worry, I hope. This sounds reasonable. I wonder how much of a supply of WD 1772 chips does Atari have left? Also, how much did it cost to reverse-engineer the 1772 chip? Was doing that cheaper than a board redesign with a chip like the FDC 92C36 you recommend? Who is responsible for these decisions? >I have experience of using NCR53C80, but it is not particularly remarkable. >It is just one of the SCSI chips. It will be a good idea to abandon the >traditional DMA interface and accommodate only a MAC style SCSI port, >urging people to buy MAC peripherals. But I would think a floppy disk >controller will be still useful especially because OS needs no fix. I don't have any SCSI experience, but one thing I noticed in reviews of peripherals using the ST ASCI (and Mac SCSI) port is that they are severely limited by the short cable length brought on by not following ANSI SCSI specs. I think ST owners want to be able to connect hard drives, PCs with SCSI, A/D D/A converters, etc. at a physical distance that is convenient to _them_, not their ST. I am beginning to become more convinced that the OS should not be in rom, but on a JEIDA standard ramcard. They are available in 4 MB capacity and a 12 MB ramcard will be out soon. Since JEIDA specs allow ramcards up to 64 MB, Atari Unix may eventually be distributed on these things. Performing OS updates would just mean downloading some code either from an Atari or local dealer's BBS. Since the new ST Notebook is using them, I am hopeful Atari will put JEIDA sockets on future lines of desktop STs. One idea that I have for older STs is a JEIDA ramcard drive connected to the ASCI-SCSI bus. TOS would be patched to load in an OS stored on a ram- card. The drive would have extra sockets so that you can have hundreds of megabytes online. >I have one internal 3.5" and one external 5" drive, both of which cope >with normal density and high density. What has been most impressive is >that I have never required any software to make use of this. Even Minix >has been very happy with them. If you are satisfied with your present arrangement, then that is great, but I just want to investigate another option. Floppy controllers can only con- trol floppy (or tape) drives; the SCSI controller provides hardware general- ization that blends nicely with the software paradigm of data abstraction. But the SCSI flexibility is crippled when manufacturers take shortcuts in implementing the standard and I believe most people would like to utilize their systems to its full potential. Jack
kiki@uhunix1.uhcc.Hawaii.Edu (Jack W. Wine) (05/06/91)
>Reverse engineering of WD1772 cannot be difficult. My explanation was too >terse: WD1772 has a data separator inside and I said replacing a better >data separtator would be cheaper and more up-to-date. Programming SCSI >interface is not terribly easy. At least for me. If future TOS supports >arbitrary SCSI bus read/write commands, the burden for us will be greatly >reduced. I don't like that NCR thing very much because it's a bit CPU >dependent. Thanks for relating this experience, because it points me toward a possibly better solution to directly replacing the floppy controller chip with the NCR 5380. If a microcontroller (mcu) was connected to the ST DMA chip instead, then compatibility with TOS could still be retained! It would receive and trans- late all the WD 1772 commands to the appropriate SCSI commands for the NCR 5380. A SCSI floppy drive would bootup and execute a program to flip a spare sound port bit which would signal the mcu that all subsequent commands are for the NCR 5380. Maybe a good mcu with huge ram buffer could be connected to multi- ple 5380s? Then something like OS/9 would be required??? The only problems with this approach are that I don't know WD 1772 commands and I don't have any schematics, timing data, and signal definitions for the ST. Does the Atari Developer Docs give complete info concerning the entire line of ST/TTs? If someone could provide a short review of it, it would be appreciated. >I personally think ICD's ATARI-SCSI board is very good for using hard drives >(including st506 things) but when it comes to using it for other purposes like >connecting a streamer tape, image scanner or a printer, the burden for me is >too great to bear. Even minix-st source code doesn't help too much. My experi- >ence is that there are so many vendor specific commands and non-standard op- >tions for SCSI. But I suppose I must learn those things one of these days. Since the TT has a SCSI, am I wrong to assume that Atari does have some sort of SCSI software toolkit? If not, then that should be a top priority project for them, because connecting SCSI peripherals to the Atari should be made simpler and standardized for developers. Easing the porting of SCSI peripherals would increase the value of the ST/TT line immensely. >Sigh. I remember a chap once built a SCSI monitor on his own that displays all >the happenings along the SCSI bus, e.g. such and such command with such and >such arguments was issued by unit what and unit what responded etc . etc. I >just don't feel like building one myself. I am really envious of your attach- >ment to SCSI. A SCSI monitor is a good idea! Maybe Atari or someone will develop one for the TT; it would be great if it was fashioned like a software debugger. Instead of forcing you to remember specific SCSI commands, it should be smart enough to let you point to a device and execute a battery of appropriate commands. Since my STe went to Atari heaven, all I can do is think about how to make it better! Jack
jclark@sdcc6.ucsd.edu (John Clark) (05/07/91)
In article <5599@wucc.waseda.ac.jp> ytsuji@wucc.waseda.ac.jp (Y.Tsuji) writes:
+Reverse engineering of WD1772 cannot be difficult. My explanation was too
+terse: WD1772 has a data separator inside and I said replacing a better
+data separtator would be cheaper and more up-to-date. Programming SCSI
If your going to do go to this then why not go with the NEC 765
which was based on an Intel Floppy disk Controler, and National has
a new and improved model. The chip, in all its varients, has odd-ball
timing delay constrains but it is used in the PC's and so would more
likely be available for some time to come.
--
John Clark
jclark@ucsd.eduytsuji@wucc.waseda.ac.jp (Y.Tsuji) (05/07/91)
In article <19062@sdcc6.ucsd.edu>, jclark@sdcc6.ucsd.edu (John Clark) writes: > If your going to do go to this then why not go with the NEC 765 > which was based on an Intel Floppy disk Controler, and National has > a new and improved model. The chip, in all its varients, has odd-ball > timing delay constrains but it is used in the PC's and so would more > likely be available for some time to come. > -- > > John Clark > jclark@ucsd.edu What is said here is to abandon ATARI ST and goto PC. I have been with Motorola 680x[x] and Western Digital 1791 series for too long to go to the other side of the trench. I admit there is nothing rational in this. NEC's 765 (it's now called muPD72067) cannot read a sector when its size is not known beforehand. I also admit 1791 series users tend to set 'protection' on their disks too easily. I heard that WD has gone over to NEC's side. Bad news. But Fujitsu, the NEC's arch-enemy and the largest computer company in Japan is still on 1791's side. I wonder why ATARI simply don't ask FUJITSU to manufacture a WD1772 clone? ATARI won't sell enough to pay the cost of building another chip plant. I am sure FUJITSU can sell the WD1772 clone with a high speed mode cheaper than any other company. ( I have nothing to do with FUJITSU.) Dr Y Tsuji ytsuji@jpnwas00.BITNET
jclark@sdcc6.ucsd.edu (John Clark) (05/18/91)
In article <5608@wucc.waseda.ac.jp> ytsuji@wucc.waseda.ac.jp (Y.Tsuji) writes:
+ Stuff about NEC 765 and a replacement for the 17xx floppy
+ controler.
+
+What is said here is to abandon ATARI ST and goto PC. I have been with
+Motorola 680x[x] and Western Digital 1791 series for too long to go to
+the other side of the trench. I admit there is nothing rational in this.
I don't think using the NEC chip implies going with the x86
processor as well. I did an OS9/68k driver for the National
incarnation of the 765 most recently. The first driver I wrote for
the chip was for a Z80(ca. '82/83) application so the processor is not
particularly important.
--
John Clark
jclark@ucsd.edu