SCSI for the 512K (or 128?)

[Charlieman]

JDW raised an interesting point in the HD20 thread.

I have found this thread to be quite interesting, especially the technical points hinting at the HD20's interface and protocol. But for myself, I am even more interested in figuring out how to put a bootable flash drive inside my old non-SCSI Mac 512k!

The important thing to remember is that there were SCSI interfaces for compact Macs with 64KB ROMs. Do a search in Google groups for "John Bass", MacSCSI, Fastimes. Or just browse through net.micros.mac. MacSCSI was a free SCSI interface with design information published in Dr Dobb's Journal. On the commercial side, a few SCSI card manufacturers made designs that worked with 64KB as well as 128KB ROMs. So homebrew cards inside a compact Mac are not a ridiculous idea.

Using the John Bass design, which presumably uses obsolete technology, a new MacSCSI card might evolve. Using a commercial SCSI<-->CF interface in conjunction with a homebrew SCSI board, a low power solution would be possible. The MacSCSI design has known limitations, however, so I incline towards cloning one of the third party boards which are more reliable/compatible.

A MacIDE card makes sense too; IDE <-->CF interfacing is trivial, and the John Bass code should give a few clues about connecting to a Mac LB. Hats off to the Apple II IDE/CF adapter crew.

 

[Mac128]

This is amazing. So the SCSI adapters for the 64K ROMs would have had to have the SCSI code encoded on an external chip. Or, at least a software driver to take advantage of any internal circuitry added (in which case the 128K probably couldn't load it), like the HD-20 so it couldn't be a boot disk? The Dove MacSnap doesn't need any software, so it relies on the 128K ROMs to drive the new circuity and theoretically could boot from a SCSI device?

I'd love to get my hands on one of those 64K ROM SCSI cards.

 

[Mac128]

Here's a thought. Would it be possible to use the HD20 model for SCSI implementation on 64K ROMs?

The HD20 cannot be used as a startup disk, but can take over once started up with a floppy containing the 128K ROM HFS & disk drivers code loaded into RAM.

Since the MacPlus 128K ROM uses a similar sequence during startup, i.e. looking for drives during startup, couldn't an HD20-like INIT be written that patched the 128K ROM code into the 64K ROM as loaded into RAM, thus making the 64K ROM aware of the SCSI bus? Then one could theoretically use something like the MacSnap SCSI adapter with the original 64K ROMs. After loading from a floppy disk, the Mac would continue scanning for a SCSI bus and mount any drive it finds, the SCSI drive system taking control and ejecting the startup disk like the HD20.

I would be perfectly happy with such an arrangement. Then I could use a Zip drive with my stock 512K.

I actually wonder now how much of the 128K code is loaded into RAM using the HD20 INIT. Is there any possibility that 100% of the additional 64K code is loaded into RAM so that a 512K could actually recognize a SCSI bus if physically attached? Could Apple's engineers have actually been contemplating a SCSI upgrade for the 512K before deciding to go with the whole rear-bucket/logicboard replacement option they ultimately offered and included the remainder of the 128K ROM code in the HD20 INIT? I mean MacSnap's solution was indeed elegant (without changing any of the functionality of the existing case) and with Jobs gone, it would have been an easy and profitable task for dealers and Apple to implement a similar solution.

Though never intended, the HD20 INIT allowed a 512K to use an 800K drive months before they were released for the Mac. Again, something Apple included in the code, but didn't advertise, preferring instead to sell the 800K drive upgrade kits with the new ROMs.

 

[napabar]

It's hard to say how much of the 128k ROM was loaded into RAM with the HD20 Int. As crazy as it sounds, I think some of the 128K ROM loaded into RAM on any machine the had it. Many years ago, I had 3 Mac 128k's. One of them was a Mac 128ke! Some previous owner had installed the 128K ROM and 800k Floppy drive upgrade from Apple. This baby still only had 128k of RAM, but could boot the HD20. However....it was very flaky with some apps that ran perfectly fine on my other stock 128k's but crashed on the 128ke. I read somewhere that Mac 128ke's had problems, because the larger ROM loaded some info into RAM, and cut down on available memory to applications. I know that doesn't make sense on the surface, but from what I've personally witnessed, it fits.

 

[JDW]

Do a search in Google groups for "John Bass", MacSCSI, Fastimes. Or just browse through net.micros.mac.

When I search for '"John Bass", MacSCSI, Fastimes' in Google, I get one result -- a link to your post in this thread!

When I type "net.micros.mac" in my Safari address bar, I get a server not found error. Please explain.

MacSCSI was a free SCSI interface with design information published in Dr Dobb's Journal.

Don't I know it. And as I posted here, I would certainly appreciate a kind soul in the US to take a stroll through the periodicals section of their local library for us. If I was back in California, I would do it in a heartbeat. But sadly, I just don't have access to old magazines like that here in Japan.

I would be perfectly happy with such an arrangement. Then I could use a Zip drive with my stock 512K.

As would I. I don't mind at all using the HD20 INIT with my 64k ROM 512. And I would absolutely love the options SCSI would open up.

So again, I beg you people, please, someone visit your local library for the good of the classic Mac community. I can't imagine that no one on this site has a library card!

This baby still only had 128k of RAM, but could boot the HD20. However....it was very flaky with some apps that ran perfectly fine on my other stock 128k's but crashed on the 128ke. I read somewhere that Mac 128ke's had problems, because the larger ROM loaded some info into RAM, and cut down on available memory to applications.

The crashing could simply be a result of your running apps that are inherently incompatible with the 128k ROMs. Indeed, that is one reason I still wish to retain the 64k ROMs in my 512.

I would love to read the article that you claim to have read about the ROM being loaded into RAM. Would you mind posting a link to that, or could you PM me so you can send me a scan of the paper article?

 

[Charlieman]

Use this search:

John Bass etc

 

[Mac128]

Do a search in Google groups for "John Bass", MacSCSI, Fastimes. Or just browse through net.micros.mac.

When I search for '"John Bass", MacSCSI, Fastimes' in Google, I get one result -- a link to your post in this thread!

When I type "net.micros.mac" in my Safari address bar, I get a server not found error. Please explain.

I easily searched for every permutation surrounding Charlieman's keywords (as well as some of my own) and think I found all of the information about John Bass & his miraculous SCSI device there is on the net. Nevertheless there was not that much. The net.micros.net forum is probably the best source I found. Odd that Google, one of the biggest search engines in the world, would not discover its own archived forum with the main part of the title as a keyword. http://groups.google.com/group/net.micro.mac/ – actually it looks like a typo on Charlieman's fault. ;-) Once there, you should take the extra effort to search for specific keywords which will produce more intriguing nuggets.

MacSCSI was a free SCSI interface with design information published in Dr Dobb's Journal.

Don't I know it. ... I would certainly appreciate a kind soul in the US to take a stroll through the periodicals section of their local library for us. If I was back in California, I would do it in a heartbeat. But sadly, I just don't have access to old magazines like that here in Japan.

You are aware Dr. Dobbs Journal is still being published and online? It looks like you will need as wellSeptember, October and December issues. Aoresteen said he would be returning home in July, so in the event you want to wait to see what he turns up in his private collection, it won't be long now.

I must admit, it is a fascinating prospect, however, as Charlieman has pointed out, even if the schematics for constructing such an interface exist and are within my abilities to construct, it will need the SCSI driver code necessary for use with it and may only be compatible with one version of Apple's system software since Apple was making broad changes to the code between each release as some of the other articles indicate. His publication is definitely odd timing. It coincides with the release of System 2.1/Finder 5.0 but he most certainly developed it around 2.0/4.1. It only preceded the Mac implementation by 4 months and the old MFS code made obsolete by the simultaneous introduction of HFS with the HD20.

In particular this is fascinating update as it highlights some very real programming problems with respect to individual drives as well as potential System/driver conflicts.

I would love to read the article that you claim to have read about the ROM being loaded into RAM. Would you mind posting a link to that, or could you PM me so you can send me a scan of the paper article?

I am sure he actually did read something to this effect as I have seen references to it myself. However, Apple seems to directly contradict this. http://docs.info.apple.com/article.html?artnum=1990

As I mentioned elsewhere, the 128K aware System files do include 128K ROM redundant code to compensate for 64K ROM System file code which had been added to the 128K ROM code, so theoretically as far as 64K ROMs are concerned, nothing has changed except the size of the System file. So unless Apple is wrong (as they so often are) and I think the specificity of this answer makes that unlikely, this is an urban-legend variety misconception.

 

[Charlieman]

The library where I work has the Dr Dobbs journal that JDW seeks, but when JDW and I discussed getting hold of a copy, the library looked something like this. Dr Dobbs was (and probably still is) in storage so it wasn't a good time to get hold of a copy. Possible but not convenient.

BTW the empty tables in the photo above are now populated with some nifty PCs. They run XP, alas, and are pretty ugly (the library development team initially requested an even uglier design), but they will be easy to maintain for several years.

The library now looks less like this and opened for service at the beginning of this month. Thus I am more than happy to get hold of the Dr Dobbs copies.

 

[JDW]

Charlieman, whenever you do get a chance, I'd love to review a copy of that Dr. Dobbs magazine article!

Mac128, thank you for the links. The posting from 1986 indicates that the original MacSCSI driver code from John Bass didn't support HFS. If it did, I don't see why Lance K. would have needed to hack/modify the driver as he did.

The 1985 post by John Bass reveals very clearly that it is a true SCSI implementation for the original 64k ROMs (since it mentions that it won't work with the newer 128k ROMs). Bass also mentions that the software driver available at the time of that posting only supported one particular SCSI drive, which he mentions in his post. But Bass then talks about a newer version of the software that would soon become available, supporting other SCSI drives (that were widely sold in that day). So even if one of us was able to acquire that newer driver software, it is highly unlikely that it would work with more modern SCSI drives that came out in subsequent years -- perhaps the only SCSI drives we will be able to find now days.

Since Bass also mentions a schedule to build 50 Fastime MacSCSI units in October 1985, that means some of those units may be out there somewhere. I've never seen one pop up on EBAY though. However, many sellers on EBAY don't often know what they're selling either.

 

[Mac128]

Ran across this again and thought I would bring it to the discussions. As much as I would love a SCSI solution for use with 64K ROMs, a solution like this would be both welcome and authentic:

http://www.dcresource.com/FlashPath/index.html

Imagine if this device initially loads 400K MFS disk header and boot info, then loads its own driver from the 400K partition to access the rest of the disk. All using the original 400K drive head keeping the machine 100% authentic while using the latest media. I can dream can't I?

 

[JDW]

I too think FlashPath would be a nifty idea, although not as fast as the "HyperDrive-Flash" that I've long dreamed of (the HyperDrive concept, but based on a flash drive). Since HyperDrive is wired to the CPU, that would theoretically be the fastest implementation. SCSI would be next, followed by FlashPath.

But I also dream of someone other than myself on this site switch on his or her Mac512 (with 64k ROMs) and testing along side me. But alas, it appears that I too must continue dreaming!

 

[napabar]

The crashing could simply be a result of your running apps that are inherently incompatible with the 128k ROMs. Indeed, that is one reason I still wish to retain the 64k ROMs in my 512.

I would love to read the article that you claim to have read about the ROM being loaded into RAM. Would you mind posting a link to that, or could you PM me so you can send me a scan of the paper article?

I don't think that was the case. Those apps ran fine in a Mac Plus with the 128k ROM.

I read that years ago. Sorry, I don't remember where I found it, but it rings true based on my experience!

 

[Charlieman]

I collected my photocopy of the John Bass SCSI article from Dr Dobbs this evening. I only had a quick look at it to confirm that they copied the right thing.

The copyright form that I signed said "for personal research" but John Bass deliberately published it as public domain, so I'll make sure that the usual suspects get copies. I would strongly advise against getting too worked up about the design. Later issues of Dr Dobbs included corrections to the ROM code for the SCSI card (which I can't yet obtain for practical reasons), so I recommend that we see the final design before trying to build one.

 

[Bunsen]

I've got a Sony version of that FlashPath adapter. As I remember, it needed drivers to be used in a 6100. Could be wrong. When it re-emerges from the pile, I'll give it another go. It requires two watch cells to work.

Also:

The Macintosh version of the driver software requires a PowerPC / running at least System 7.5.3 or later. / 68040 (or earlier) aren't supported. / the Mac version is read-only

 

[Bunsen]

How much documentation is available about the HD20 interface itself? Would it be possible to construct a device that plugs directly into the HD20 port, mounts a flash card, and contains a microcontroller for protocol translation? Such that the original HD20 inits are useable.

 

[Charlieman]

How much documentation is available about the HD20 interface itself?.

From Apple, zilch. From other threads over here, we have had some excellent info on how the Rodime mechanism in an HD20 works.

 

[Bunsen]

Hmm, interesting, but a different tack to what I'm talking about ...

... which is throwing aside the whole HD20 drive, and plugging a microcontroller into the HD20 port on the back of the Mac. The micro then pretending to be an HD20 drive, when really it's a flash card reader.

Although I can see that the Hyperdrive processor-direct approach would be the optimal one, that looks like a fair chunk of development.

Serial port?

 

[Mac128]

I've starteda new thread specifically posting the public domain John Bass article's from Dr. Dobb's Journal which detail both the hardware to build a SCSI adapter for the 64K ROMs as well as the software drivers:

viewtopic.php?f=7&t=9800

I had intended to go back and edit my title and entry to better reflect this thread as well as add a link to this thread as well. Unfortunately new forum rules no longer allow edits. Therefore, I've updated this thread to better help connect them should anyone perform a search or come across this thread first.

 

[napabar]

I would love to read the article that you claim to have read about the ROM being loaded into RAM. Would you mind posting a link to that, or could you PM me so you can send me a scan of the paper article?

I am sure he actually did read something to this effect as I have seen references to it myself. However, Apple seems to directly contradict this. http://docs.info.apple.com/article.html?artnum=1990


As I mentioned elsewhere, the 128K aware System files do include 128K ROM redundant code to compensate for 64K ROM System file code which had been added to the 128K ROM code, so theoretically as far as 64K ROMs are concerned, nothing has changed except the size of the System file. So unless Apple is wrong (as they so often are) and I think the specificity of this answer makes that unlikely, this is an urban-legend variety misconception.

I wouldn't say that article contridicts my statement. I would say it confirms it. 1.5K lost may not seem like much on a Mac 512K, but on a RAM strapped 128K, it could make the difference between an application working or not. :beige: