Floodgap ANSwers: The Shiner ESB Prototype

If you'd like to use these pictures somewhere else, please ask me first. Thanks!

The codename for the Apple Network Server during its development was "Shiner," named after a local brand of beer popular with the Austin, Texas-based design team. (The company is still in business.) Other than the prototype-only "Deep Dish" 4U ANS 300, the other prototype system was the Shiner HE ("High End"), which became the ANS 700; the ANS 500 ("Shiner LE") was derived from it. This is one of those prototype ANS 700 systems. Notations on its logic board and components date it to 1995.

Most people are familiar with Apple's engineering validation test (EVT) prototypes, and some of these have famously escaped to the outside world. EVT units are typically feature complete pre-production units, though some aspects may be unfinished and debugging facilities are sometimes still present for test purposes, and some features on EVT units may not make it to the mass production phase. Despite being badged as "ESB" (possibly standing for Extra Special Bitter, in keeping with the beer motif) on the rear, this system is actually an EVT-2, phase 1 being design, phase 2 being prototype and phase 3 being validation prior to stabilization and mass production tests (DVT and PVT).

I picked up this unit from a scrapper in San Rafael, California. Even though it was an EVT unit, it was fully functional at one time, and strings from the hard disk it came with (unfortunately only one of a volume group of at least two) indicate it was a test development server at Netscape/Mosaic Communications Corporation up through Netscape's acquisition of America Online in 1999. Sadly, it was stored improperly afterwards with the Maxell PRAM battery left in place; as these batteries have a tendency to do (which is why us Mac collectors call them "Maxell bombs"), it exploded and leaked acid all over the logic board, ruining it. The acid damage and other oxidation during storage is beyond my ability to repair, but it's still a fascinating unit worth looking at in detail. Compare these pictures with our production ANS 500.

Even from a little distance away you can tell this unit isn't a regular ANS. For one thing, there's no ANS badge, just a blank Apple one where it would go.
Entire original image (79.3KB)

Closer up, you can also see that the keylock is missing all the icons. (The keylock on this unit had to be forced; the front door was locked shut and no spare keys were available.) Also notice the America Online asset tags and the hostname/IP address on the front, which doesn't seem to answer to anything anymore. The FQDN on the hard disk was shiner.mcom.com.
Entire original image (57.9KB)

The blank badge and Apple logo. Likewise, notice from the sheen that the plastic was not textured as it would have been for production beige Macs or ANSes, although it does appear to be injection-moulded.
Entire original image (56.5KB)

However, the power switch and reset/programmer buttons are pretty much the same as the released units otherwise.
Entire original image (80.6KB)

The power supply is also broadly similar to the production 700's power supply, and has a handle like a 700 PSU would, but is in fact bolted in place (see the metal restraining tabs) and can't actually be ejected.
Entire original image (141.9KB)

The rear, however, is almost identical to the production unit, including the same port markings and warning near the rear key lock.
Entire original image (106.2KB)

On the other hand, the backplate where the model and copyrights would go is replaced by a single large legend "SHINER ESB H.E." and a yellow sticker on the other panel indicates there is no FCC approval and that "this unit [is] for evaluation purposes only."
Entire original image (95.8KB)

The SCSI backplane is also pretty much "done" and it looks the same as the production backplane in my 500. The only easily noticeable difference is that the power connector has several additional inputs.
Entire original image (98.3KB)

However, the part number differs: the prototype is 820-0714-03, but the production SCSI backplane is 820-0714-A. This suggests the third development revision of this board, consistent with its early copyright date of 1995.
Entire original image (133.5KB)

Pulling out the logic board on its drawer. The slide mechanism and card guides are pretty much final. Unfortunately you can see a certain amount of oxidation from the poor storage already.
Entire original image (126.9KB)

Inside the case is the unpleasant foam which is always degenerating and the connector to the prototype mezzanine.
Entire original image (106.8KB)

The mezz is codenamed "HENDY." It is labeled as "EVT-2" and also dated 1995. The production mezz is 820-0713-A; this one is 820-0713-07, suggesting quite a number of changes took place here as well.
Entire original image (90.4KB)

The logic board/motherboard is also marked EVT-2, part number 820-0744-02. The colour is also a little more blue than the usual olive green of Apple's contemporary boards, though how much of this is manufacture versus the irregularities of its storage is not clear.
Entire original image (161.6KB)

The damage the Maxell bomb caused unfortunately is obvious; you can see a stream of wreckage where the battery paste came down the board. Separately, several of the connectors were rusted over. Codenames also appear here instead of production markings: GRAND CENTRAL is visible near the Cirrus Logic video chip, which is a PCI device handling serial ports, external SCSI, Ethernet and the floppy controller. The PCI controller is marked with BANDIT, its codename. Both Grand Central and two Bandits are in the Power Mac 9500, but differently connected and at differing addresses. On the production board, Grand Central is unmarked, and the Bandits are labeled PCI BRIDGE 1 and 2.
Entire original image (171.7KB)

Other prototype specific markings include the Cache SIM [sic] and ROM SIM [sic] slots. Both these slots appear on the production board but not with the appelation "SIM." The memory controller is marked with Hammerhead, its codename, and the data path IC is marked Zax. These markings do not appear on the production board either, which simply calls them MEMORY CNTRL and DATA PATH respectively.
Entire original image (166.3KB)

More board damage from the battery. The Fast-and-Wide SCSI controller, marked F/W SCSI on the production board, is given its part number 53C825 on this board.
Entire original image (141.2KB)

The CPU card, a 150MHz 604, has many non-production markings and bears the codename "FIGMENT." In particular, notice the bus speed and 604 power table "LEDGEND" [sic]. The same jumpers in the same positions are also present on my production 200MHz card but the production card doesn't seem to actually have any contacts. This card is also dated 1995 with part number 820-0740-01; my production cards are 820-0740-A, differing only in their resistor blocks.
Entire original image (142.9KB)

generated by Floodgap Super Image Browser 1.788.0

	Even from a little distance away you can tell this unit isn't a regular ANS. For one thing, there's no ANS badge, just a blank Apple one where it would go. Entire original image (79.3KB)
	Closer up, you can also see that the keylock is missing all the icons. (The keylock on this unit had to be forced; the front door was locked shut and no spare keys were available.) Also notice the America Online asset tags and the hostname/IP address on the front, which doesn't seem to answer to anything anymore. The FQDN on the hard disk was `shiner.mcom.com`. Entire original image (57.9KB)
	The blank badge and Apple logo. Likewise, notice from the sheen that the plastic was not textured as it would have been for production beige Macs or ANSes, although it does appear to be injection-moulded. Entire original image (56.5KB)
	However, the power switch and reset/programmer buttons are pretty much the same as the released units otherwise. Entire original image (80.6KB)
	The power supply is also broadly similar to the production 700's power supply, and has a handle like a 700 PSU would, but is in fact bolted in place (see the metal restraining tabs) and can't actually be ejected. Entire original image (141.9KB)
	The rear, however, is almost identical to the production unit, including the same port markings and warning near the rear key lock. Entire original image (106.2KB)
	On the other hand, the backplate where the model and copyrights would go is replaced by a single large legend "SHINER ESB H.E." and a yellow sticker on the other panel indicates there is no FCC approval and that "this unit [is] for evaluation purposes only." Entire original image (95.8KB)
	The SCSI backplane is also pretty much "done" and it looks the same as the production backplane in my 500. The only easily noticeable difference is that the power connector has several additional inputs. Entire original image (98.3KB)
	However, the part number differs: the prototype is 820-0714-03, but the production SCSI backplane is 820-0714-A. This suggests the third development revision of this board, consistent with its early copyright date of 1995. Entire original image (133.5KB)
	Pulling out the logic board on its drawer. The slide mechanism and card guides are pretty much final. Unfortunately you can see a certain amount of oxidation from the poor storage already. Entire original image (126.9KB)
	Inside the case is the unpleasant foam which is always degenerating and the connector to the prototype mezzanine. Entire original image (106.8KB)
	The mezz is codenamed "HENDY." It is labeled as "EVT-2" and also dated 1995. The production mezz is 820-0713-A; this one is 820-0713-07, suggesting quite a number of changes took place here as well. Entire original image (90.4KB)
	The logic board/motherboard is also marked EVT-2, part number 820-0744-02. The colour is also a little more blue than the usual olive green of Apple's contemporary boards, though how much of this is manufacture versus the irregularities of its storage is not clear. Entire original image (161.6KB)
	The damage the Maxell bomb caused unfortunately is obvious; you can see a stream of wreckage where the battery paste came down the board. Separately, several of the connectors were rusted over. Codenames also appear here instead of production markings: GRAND CENTRAL is visible near the Cirrus Logic video chip, which is a PCI device handling serial ports, external SCSI, Ethernet and the floppy controller. The PCI controller is marked with BANDIT, its codename. Both Grand Central and two Bandits are in the Power Mac 9500, but differently connected and at differing addresses. On the production board, Grand Central is unmarked, and the Bandits are labeled PCI BRIDGE 1 and 2. Entire original image (171.7KB)
	Other prototype specific markings include the Cache SIM [sic] and ROM SIM [sic] slots. Both these slots appear on the production board but not with the appelation "SIM." The memory controller is marked with Hammerhead, its codename, and the data path IC is marked Zax. These markings do not appear on the production board either, which simply calls them MEMORY CNTRL and DATA PATH respectively. Entire original image (166.3KB)
	More board damage from the battery. The Fast-and-Wide SCSI controller, marked F/W SCSI on the production board, is given its part number 53C825 on this board. Entire original image (141.2KB)
	The CPU card, a 150MHz 604, has many non-production markings and bears the codename "FIGMENT." In particular, notice the bus speed and 604 power table "LEDGEND" [sic]. The same jumpers in the same positions are also present on my production 200MHz card but the production card doesn't seem to actually have any contacts. This card is also dated 1995 with part number 820-0740-01; my production cards are 820-0740-A, differing only in their resistor blocks. Entire original image (142.9KB)