IDE to SCSI Adapter - discrete logic - reverse engineering candidate?

[Trash80toHP_Mini]

I've started trying to identify the components on the IDE Zip250 to SCSI adapter board on a very interesting Winstation product: WFS2E1W13. The board appears to be a discrete component IDE to SCSI bridge implementation that could be equivalent to the ASIC on ACARD adapters?

Assumption would be that some of the ICs must be programmable logic, but the board might be a good model for development. Programmable logic IC could be dead ends, but where they fall in the schematic might define the underlying architecture. If anyone recognizes any of the ICs offhand, please post them.

Major ICs:

U2 - What should be a stickered ROM

U25 - qlogic - FAS209 - SCSI 10MB/s SE 8bit

U5 - Intel - S0C31BHI - ????? - appears to be a stock part equivalent to same part number from Mfr. TMS - anybody have info on that company or the chip's function? Searching came up dry so far.

Thinking here is that  this could be a first step in creating a SATA to SCSI board? SATA to PATA chips seem readily available or if not, could be easily harvested from very inexpensive adapters.











View attachment 29977



This is so far out of my league as to be ridiculous, but someone might take an interest?

 

[Cory5412]

Neat, I didn't realize someone built SCSI Internal Zips (Iomega built SCSI Externals) - it looks like this is probably the same configuration used by the Winstation internal LS-120 drive, which was similarly an Imation LS-120 drive, with an IDE-SCSI board tacked on.

Adapters to use SATA disks on SCSI buses already exists, Acard built them. For example: http://www.acard.com/index.files/Page861.htm

EDIT: Here's a more generic listing of Acard's SCSI bridges: http://www.acard.com/index.files/Page385.htm /EDIT

And: https://www.ebay.com/itm/ACARD-ARS-2320-Ultra320-SCSI-to-SATA-2-Enclosure-68-pins-New-Old-Stock-Tested/274033010323 (You could enclose a 2.5" SATA disk in this adapter and put it in a server where you'd use a u160/u320 scsi drive.)

It would be interesting to get a perspective on whether cloning the Classic Mac OS compatible SATA cards would be easier or more difficult than this, because most Macs that really "need" more speed than their onboard SCSI have PCI slots. The rest of them are using IDE, and IDE-SATA/CF/SD adapters do exist.

The other main option here is SCSI2SD v6 and in my experience doing day-to-day stuff on an 8600/300 with OS 9.1 and a scsi2SDv6, that's a very good solution if you did not have a pci slot or SATA card for some reason.

 

[Trash80toHP_Mini]

Yep, but everything ACARD seems very expensive. Adapters have hit the market at very reasonable price points, but ISTR they were determined to be for the likes of opticals, not mass storage?

LVD SCSI to IDE adapters are in the $70 range on Amazon.



I'm wondering if something more compatible with Mac cubic might be designed or even with EMMC storage on a board built in the 2.5" PowerBook form factor? ISTR, you saying SCSI2SD is great for snappiness of response times, but throughput pales as compared to CF or even spinning rust?

Dunno, it's all theoretical discussion for now and probably too much work to do for benefit, but very cool if workable.

 

[Bolle]

The Intel chip could be something like an 80C31BH MCU.

If it is then it has no internal program memory and relies on an external source for its program memory. This would make copying easy.

 

[Trash80toHP_Mini]

Would that be supplied in ROM for the program or in discrete memory for buffer? The board's design use as 250MB Zip interface or LS1io as Cory suggested would be cool. Interfacing faster ATAPI cds and even DVD in SCSI only or Limited function, single drive only IDE Macs would be interesting applications.

Interfacing EMMC in a laptop drive form factor I find interesting. EMMC would be better than SD or CF, if not a bona fide SATA SSD no?

 

[Gorgonops]

The Intel chip could be something like an 80C31BH MCU.

If it is then it has no internal program memory and relies on an external source for its program memory. This would make copying easy.

That chip with the "SLS/7.43" sticker has 28 pins, which does indeed make it a good candidate for being the ROM.

I haven't read the label on every chip but every one I have looked at seems to be a generic 74xx series part, so there may well be no programmable logic on this thing besides the CPU. I'm going to demur on whether this is actually a useful thing to try to clone; it has a pretty high part count and the availability of the SCSI chip is probably a significant issue. (I see one guy selling a box of 11 of them on eBay. One guy. That's not a good sign.)

 

[Unknown_K]

ACARD uses a custom chip to make their adapters simple and cheap to build. The older adapters used more chips and space, plus those chips are probably not availbale these days.

 

[Cory5412]

STR, you saying SCSI2SD is great for snappiness of response times, but throughput pales as compared to CF or even spinning rust?

It depends on the machine and what you're doing.

In my experience, SCSI2SDv6 is just about fast enough to be used for something like video capture on the 8500/8600 (which is reputed to need about 4.5 megabytes/second of transfer.)

This is the post with my MacBench 4 disk benchmark results:



So like, it was good enough for booting and running applications and probably even dev work would be fine on it, but, and I haven't tried this, it might not meet expectation if you started throwing video at it.

At that point, yes, something like a real hard disk would be better, or something on SCSI with a better adapter.

The use cases for that are moderately niche, however, and if you're setting up an 8600 for retro analog standard def desktop video editing, you're probably also hunting for period-appropriate bits like a Micropolis disk or a Seagate Elite 9/23/47 for capture, for example.

but ISTR they were determined to be for the likes of opticals, not mass storage?

There are different adapters with different purposes, though, to be honest, what with the utter glut of cheap-good IDE and SATA CD-ROM drives that exist, someone would do well to pick up a couple of those adapters if they're cheap and integrate a few into, say, ex-syquest external enclosures -- if the inventory of external SCSI CD-ROM drives was at some point drying up.

There's also the CF powermonster, which might be a design that can be put back into production, and would provide a performance boost and modern part, but those were selling for more than SCSI2SDs even years ago when SCSI2SD cost a lot more than it does now and the powermonster (aztecmonster?) was still being built.

EMMC would be better than SD or CF, if not a bona fide SATA SSD no?

eMMC is usually very similar media to SD. I wouldn't say it's any better, but SD is a super stratified market at this point, as I keep mentioning all over the place every single time someone says their SCSI2SD isn't as fast as some people say it is -- that's always because they bought an SD card that's not high end enough.

eMMC itself runs the gamut from "run away as far as possible" to "passable, but you'd absolutely rather have SATA or NVMe".

I'm wondering if something more compatible with Mac cubic might be designed or even with EMMC storage on a board built in the 2.5" PowerBook form factor?

PowerBooks are an entirely different use case. Out-and-out highest end possible raw performance really isn't what anyone needs strictly speaking to use all the functionality of any SCSI-having PowerBook. (Notably here, the PowerPC PowerBooks all use IDE, for which known good and widely available m.2 and CF adapters are available.)

Anyway, for SCSI PowerBooks, there's the SCSI2SD PowerBook Edition: https://store.inertialcomputing.com/product-p/scsi2sd-v5-2.5-inch.htm 

And the regular SCSI2SD devices all fit within the boundaries of a 3.5-inch disk, meaning they fit fine in any 68k Mac.

So, like, I'm not against the idea of another kind of adapter existing, but, I do think we should think about what our needs are and what is/isn't available and how it does/doesn't work.

I'll admit, aside from perhaps "really fast NuBus PowerPC Macs" I struggle to think of a scenario we don't have reasonably covered, although perhaps some of our wiki documentation needs to be updated.

 

[Gorgonops]

Re: the performance discussion, it bears pointing out that the adapter in the OP is almost certainly *far* slower than an Acard-style adapter, and very likely significantly slower than the better variants of the SCSI2SD. It's based on an 8-bit MCU designed in 1980 as a competitor for devices like the Z-8. It's clocked pretty high for an 8 bit CPU (... of the era, not now) but it doesn't look like there's any external RAM, which would almost certainly rule out DMA operation. An 8-bit MCU with a very small amount of RAM task-switching between doing PIO with that SCSI chip and driving the IDE bus is anything *but* a recipe for high-speed operation. Color me surprised if this thing could keep up with even the fairly mediocre performance of the ZIP 250 drive it was connected to.

 

[Trash80toHP_Mini]

Like I said, could be more bother than it's worth, but interesting discussion nonetheless. Is there a soft(?) SCSI controller design available for  integrating into an FPGA? That's seems like part of what's going on in ACARD's adaptation. Fast/Wide UltraSCSI would be optimal and even faster than Fast/Narrow of SCSI2SD v.6, which I can't wait to get my hands on at some point, BTW.  :approve:

ACARD uses a custom chip to make their adapters simple and cheap to build. The older adapters used more chips and space, plus those chips are probably not availbale these days.

Yep, mentioned ACARD's ASIC based implementation in the first sentence of the post. I was looking at this more as a possible block diagram for implementing an updated version in FPGA.

There are different adapters with different purposes, though, to be honest, what with the utter glut of cheap-good IDE and SATA CD-ROM drives that exist, someone would do well to pick up a couple of those adapters if they're cheap and integrate a few into, say, ex-syquest external enclosures -- if the inventory of external SCSI CD-ROM drives was at some point drying up.

I'm interested in adapting modern slot loading opticals from laptops. ATM I've got them set up with IDE adapters, but having one adapted to SCSI is one of my goals.

 

[Gorgonops]

Yep, mentioned ACARD's ASIC based implementation in the first sentence of the post. I was looking at this more as a possible block diagram for implementing an updated version in FPGA.

I'm sure whoever owns ACARD's intellectual property at this point isn't particularly interested in sharing. Their old marketing spiel says their converter ASICs had a built in "High Speed ACARD RISC Microprocessor" on it, what the actual specs of what that is are and whether it used an ISA that has FPGA implementations openly available are utter unknowns to me. (Although perhaps you might be able to figure it out by analyzing a firmware file. For cheap thrills I downloaded one and ran "strings" on it; it looks like it has the SCSI ID text fields in plain text so it might not be encrypted in any way, so if you could make an educated guess what the instruction set is and could run a disassembler program on it in principle you could learn a lot about its internal architecture. Those are really big ifs, though. It's not unusual for ASIC-embedded CPUs to use highly customized ISAs; it might be derived from, say, MIPS, but contains weirdness in it that renders it cryptic without the customized support toolchain.)

 

[Trash80toHP_Mini]

That's waaaay beyond my ken.  :blink: I knew Bolle would find this interesting and I was hoping folks like you, @trag and the FPGA/CPLD (whatever, they're just acronyms to me) gunslingers might be interested, if only on a theoretical level. Never considered ACARD sharing anything, I was wondering if on open source SCSI controller was out there?

 

[Unknown_K]

I do kind of wonder what the difference between the IDE and PATA SCSI adapters are outside of just firmware. If somebody could crack that then the much cheaper PATA adapters could be reflashed and make life easier.

Also the Addonics adapters I have seem to be the same as the Acard variety of the same model name and firmware.

 

[Gorgonops]

I was wondering if on open source SCSI controller was out there?

There's this project from 2008 on OpenCores that was active for a week and a half before they suspended it. That's not too promising.

https://opencores.org/projects/scsi_chip

That said, there are plenty of software bit-banging solutions out there for doing SCSI with just some buffers. (There's SCSI2SD, that project for doing it with the GPIO ports on a Raspberry Pi, etc.) An FPGA god probably wouldn't have a *lot* of trouble converting the code that drives those buffers into hardware state machines. It's just a lot of picky, excruciating work for someone to do out of the goodness of their heart. And the end result would probably have to sell for a *lot* more than a SCSI2SD sells for if you want something *fast* and "production quality". (Based on several projects I'm vaguely interested in having it looks to me like the minimum ballpark price for a complex FPGA "hobby" project is in the $100-$200 range. That's a lot to hook up an old IDE CD-ROM drive.)

 

[Unknown_K]

I think most of the ATAPI models were for old SCSI only systems where people wanted to replace the dead SCSI CDROM drive with something more easily available like an IDE CDROM, CDRW, or DVD drive or burner. 

I stuck a DVD drive into a Q950 for fun and it worked from what I recall (connected to a Jackhammer card).

 

[Trash80toHP_Mini]

I was wondering if on open source SCSI controller was out there?

There's this project from 2008 on OpenCores that was active for a week and a half before they suspended it. That's not too promising.

- Synchronous Data Transfers un to 4MB/sec is even less promising. That would be more on the order of pathetic Macintosh SCSI 1 implementations?

I do kind of wonder what the difference between the IDE and PATA SCSI adapters are outside of just firmware. If somebody could crack that then the much cheaper PATA adapters could be reflashed and make life easier.

Also the Addonics adapters I have seem to be the same as the Acard variety of the same model name and firmware.

View attachment 30001

Edited 15 hours ago by Unknown_K

Addonics appears to have distributed a 96k modem for PC from the OEM(?) that made my PowerBook specific 56k modem. Same very cool case, compatible with the 56k modem's ADB PTO, PowerBook Charcoal and all. I'd be very surprised that you'd find anything but an Acard logo under that Addonics sticker.

Model number "7722" between Addonics and Acard is nearly the same, but the LF suffix combined with LVD spec on the IBM aCard IDE to LVD-SCSi Bridge Adapter AEC-7722 give me pause. Have you got the SCSI Spec on yours? LVD was introduced in Ultra-2 which appears to be compatible with LVD 1.2 and HVD ≥5 voltages while Ultra3 appears to be LVD 1.2 only. What's the date on your adapter? 1994 to 1996 would be Ultra-2. I wonder about the IBM/Acard version dates, if made after 1996 it could very likely be Ultra-3 and low voltage only and so incompatible for my purposes? Wondering here about pre-1996 Mac voltage compatibility and optional G4 series SCSI card?

Wikipedia

Ultra-2
This standard was introduced c. 1997 and featured a low-voltage differential (LVD) bus. For this reason ultra-2 is sometimes referred to as LVD SCSI. LVD's greater resistance to noise allowed a maximum bus cable length of 12 meters. At the same time, the data transfer rate was increased to 80 MB/s. Mixing earlier single-ended devices (SE) and Ultra-2 devices on the same bus is possible but connecting only a single SE device forces the whole bus to single-ended mode with all its limitations, including transfer speed. Ultra-2 SCSI actually had a relatively short lifespan, as it was soon superseded by Ultra-3 (Ultra-160) SCSI.

Ultra-2
Also known as Ultra-160 SCSI and introduced toward the end of 1999, this version was basically an improvement on the ultra-2 standard, in that the transfer rate was doubled once more to 160 MB/s by the use of double transition clocking. Ultra-160 SCSI offered new features like cyclic redundancy check (CRC), an error correcting process, and domain validation, a way to negotiate maximum performance for each device on the chain.

Curiouser and curiouser. Got linkage to the less expensive "PATA" adapters? What was the acronym for IDE with optical drive support? Drawing a blank again.  :mellow:

 

[Unknown_K]

LVD80 standard. If you look at the picture the BIOS has an ACARD label.

 

[Trash80toHP_Mini]

HAH! makes sense. That's too bad, good for single device only "chain" for all practical purposes in a Mac w/CD then? I guess you could use the Optical Adapter (ATAPI) with it. Dunno about flashing the BIOS, but some SMD rework could test your hypothesis. I imagine reading the ROM/BIOS in a bare nekkid state and making copies might be a good way to go. Methinks you're right, different BIOS for low level product differentiation? Sticker peeling and ASIC ID check would have to be the first step. A slower version of the ASIC would be a better way of differentiating product levels. Could also be a fuse burn in the same part as Moto disabled failed test FPU sections to create perfectly good LC040 versions along with intentional disabling for product level differentiation.

just missed edit window: was correcting the second Ultra-2 to Ultra-3 and the dates to 1996 and late 1997 when the hourglass flipped. Also found ATAPI acronym in the description on your box. [:I]

 

[Gorgonops]

- Synchronous Data Transfers un to 4MB/sec is even less promising. That would be more on the order of pathetic Macintosh SCSI 1 implementations? 

Let the record show that 4MB/second is the equivalent of a 26.6666x-speed CD-ROM drive. Why do you need something super-fast if your goal is to drive laptop PATA CD-ROM drives, again?

(Sure, that is only about 3x if you're using DVD as your benchmark but, still, it's fast enough.)