[jan-jaap]

Picked up a new addition to my collection last weekend: a Professional IRIS 4D/50

I met the original user, who used it between 1998 and 1991. Then it went home with him, but not before the entire machine was put through a degausser.

He stored it in his basement, where it sat untouched since 1991. Last week he fired it up, and apparently it came to life. Of course the hard disk is (still) completely erased. He put it on eBay and now it's mine



The system is basically immaculate. I just need to remove 30+ years of dust and grime. Of course I will not just start using and pray nothing blows up. I'm going to have to do a deep inspection and testing, starting from the power supply

---

Took off the skin panels to give them a gentle wash. The power supply is in the bottom of the drive bay stack. It's a 1000W MightyMite unit made by L&H Research.

The system disk is a 145MB Toshiba. I hope it can be brought back to life after being degaussed? I may need to low level format it to rebuild the bad sector map

---

There's only four boards inside the 12-slot card cage:

• An IP4 @ 8MHz. It looks like I've got 8MB RAM. The IP4 was SGI's successor to the MIPS R2300 MBC (multi board computer) which occupied up to four slots
  
• Clover1 graphics: GF3, DE3 and TB2. The (optional) ZB2 Z-buffer board is not present
  

The DE3 is only ~ 3/4 populated, unlike what Gerhard Lenerz's site shows. I may be missing alpha planes. Perhaps this is 'B' graphics -- 'G' without Z and alpha? The memory may look like 30-pin SIMMs, but the RAM chips are dual ported, Toshiba MB81461-12

The PROM is an older version than what's in my 4D/70 GT. It's

Code:

Version 4D1-2.1 PROM IP4 OPT Tue Nov 10 04:00:37 PST 1987 SGI

It put a copy of the PROM here: https://vdheijden-messerli.net/sgistuff/...onal_IRIS/

[robespierre]

The system disk is a 145MB Toshiba. I hope it can be brought back to life after being degaussed? I may need to low level format it to rebuild the bad sector map

Degaussing the drive (if the magnetic field really penetrated) would scramble the servo information. Most drives can't write their own servo information because it is by nature an analog signal. Typically the servo is written slightly off-track to both sides, so that when the head stack is centered on a track, the head reading the servo surface receives two overlaid waveforms with equal amplitude. That head may not have any write amplifier connected to it.
Most drives used a dedicated surface for servo, but a few used embedded servo information recorded inside the inter-record gap of actual data tracks. The latter prevents low-level formatting in most cases.

[jpstewart]

What a spectacular machine to find! It's also nice to know it'll be in good hand with you, jan-japp. Congrats.

I'm sure you'll have tons of fun with it. Be sure to keep posting updates and photos for those of us who will never get to see one in person.

[weblacky]

It's great to still see some "rare find" stories coming up! We all keep seeing longer and longer stretches between scores, we all love it when someone shares a sweet score like this one. Especially such a rare model and such a mint cosmetic condition!

Definitely keep us up-to-date with each phase of your refurbishment.

[robespierre]

The geometry data for the drive is

Code:

MK156FB     SCSI          10 heads        830 cylinders        35 sectors         0 Write-Precomp    830 Landing-Zone     145MB Capacity

The operation manual for the related MK156FA (an ESDI drive) is here. Of note:

Use of a center stack servo system improves head positioning accuracy across the full environmental operating range and allows servo writing to be performed with the disk stack mounted in the drive.

So it has a dedicated servo surface on an inner platter.

My 4D/70G was slightly different: the power inlet was a Hubbell twist-lock, not an IEC connector, the base compartment of the side tower had space for a tape drive, and there was a DC-37 port on the back for some unknown purpose (ESDI? Parallel?)

[jan-jaap]

So it has a dedicated servo surface on an inner platter.

Thanks, we'll see what comes of it in due time.

My 4D/70G was slightly different: the power inlet was a Hubbell twist-lock, not an IEC connector, the base compartment of the side tower had space for a tape drive, and there was a DC-37 port on the back for some unknown purpose (ESDI? Parallel?)

Initially the Professional IRIS shipped as the 4D/60 with the 8MHz R2300 MBC from MIPS. The R2300 relied on a VME ESDI controller and the CMC ENP-10 VME ethernet card. All Professional IRISes come with bulkhead wiring for ESDI and ethernet on the main tower. It is my understanding that ESDI wiring to the drive stack was external. I don't know if the tape drive of the 4D/60 was ESDI or something else (QIC-02?)

Then only 4 months later in July 1987, SGI announced the 4D/60T, the "turbo" upgrade for existing 4D/60 owners. This replaced the R2300 MBC with the 12.5MHz IP4 SBC. Complete systems were sold as the 4D/70, and a little later the 4D/50 with 8MHz CPU. Unlike the R2300 MBC, the IP4 board has SCSI. It is connected via the user-defined pins of the VME P3 connector and internal wiring to the drive tower. Additionally, there's a sticker on my chassis that says it's an "SBC CHASSIS".

Now, I imagine that if you replace a CPU board, you leave the rest of the system untouched as much as possible. So a 4D/60 upgraded with an IP4 to a 4D/60T would probably still use it's ESDI disk and doesn't strictly require SCSI. So one can speculate about the chassis of the original 4D/60. Did the "MBC CHASSIS" already have SCSI wiring? And what about the interconnect between the modules in the drive tower? With so little time between the R2300 and the IP4, the IP4 must have been under development before the 4D/60 was even released. So what makes an "MBC CHASSIS" -- ESDI wiring?

The twist-lock was for 110VAC countries I believe. The metal box with the circuit breaker and power inlet comes out easily and plugs into the wiring harness of the PSU. The harness itself is much more elaborate than most other systems. The PSU requires an external fuse, so there are three main power cables going to it: the top one feeds it from the power inlet, the bottom two go back to the circuit breaker. There's got to be a different circuit breaker in the 110VAC version as well (30 vs 15A). My more immediate problem is that the PSU doesn't switch on if I supply power directly and bridge the fuse. I would at least expect it to create 300+ VDC internally after the primary rectifier. I must be missing something.

I had it open to inspect e.g. electrolytics. There's a RIFA cap on the input filter that's cracked and I suspect a bit of leakage on the 12V output filter caps so I think it's prudent to replace them now that I'm in here anyway. But everything looks spotless and untouched -- it's like I unearthed a NOS Professional IRIS.



You can also see the 'intrusion detection' switch mentioned on Gerhard Lernerz's site and also in "This Old SGI" if I remember correctly. There's another on the main tower. If you open it the circuit breaker will trigger and kill power to the system

[jan-jaap]

Update: the PSU starts.  What it needed was some load on the main 5V rail. The builtin power resistors across the output capacitors are not enough. An electronic load set to 10A was enough, it happily supplies 40A as well

[arhiman57]

This is insane, I love it.

I've seen the listing on ebay.de, it was not too far, but It's far more responsible to let this machine to an expert Can't wait the next step

Maybe a good addition (but I think it's from Power series era)
https://www.kleinanzeigen.de/s-anzeige/s...1-228-7862

Do you have any background on this particular machine, original usage etc ?

[jan-jaap]

The system disk is a 145MB Toshiba. I hope it can be brought back to life after being degaussed? I may need to low level format it to rebuild the bad sector map

Degaussing the drive (if the magnetic field really penetrated) would scramble the servo information. Most drives can't write their own servo information because it is by nature an analog signal. Typically the servo is written slightly off-track to both sides, so that when the head stack is centered on a track, the head reading the servo surface receives two overlaid waveforms with equal amplitude. That head may not have any write amplifier connected to it.
Most drives used a dedicated surface for servo, but a few used embedded servo information recorded inside the inter-record gap of actual data tracks. The latter prevents low-level formatting in most cases.

I took the disk out of the drive tower module. Initially it didn't even spin up, but I managed to unstick it by jolting it in the rotational direction of the platters and letting inertia do it's thing. it spins up now, but then it makes a "dakdakdakdakdakdakdakdakdakdakdakdak" noise that it doesn't seem to get out of.

I guess the degaussing has successfully destroyed this hard disk beyond recovery?

[jan-jaap]

Maybe a good addition (but I think it's from Power series era)
https://www.kleinanzeigen.de/s-anzeige/s...1-228-7862

I have two meters of those brown binders already

Do you have any background on this particular machine,  original usage etc ?

Not much. The person I got it from, who was the original user, was a software development consultant. I know he also discovered the flight sim (dog). My guess would be something related to the military. A university or business would not write off the machine in 3 years and (in 1990!) care enough about security that they would degauss the whole thing.