UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Specific Vintage Equipment > Vintage Computers

Notices

Vintage Computers Any vintage computer systems, calculators, video games etc., but with an emphasis on 1980s and earlier equipment.

Closed Thread
 
Thread Tools
Old 23rd Jun 2011, 6:44 am   #1
Kat Manton
Retired Dormant Member
 
Kat Manton's Avatar
 
Join Date: Jul 2005
Location: West Yorkshire, UK.
Posts: 1,700
Default Sun SPARCserver 1000E Saga

A bit of history...
In 1992, Sun Microsystems introduced the sun4d architecture. This uses SuperSPARC processors, SBus I/O cards and the XDBus system bus. This system bus was the result of a collaboration between Xerox and Sun; named after the earlier Xerox Dragon multiprocessor workstation project.

The first system to appear was the SPARCcenter 2000 (codenamed Dragon). In 1993 this was joined by the smaller SPARCserver 1000 (codenamed Scorpion/Baby Dragon) and the third sun4d system, the Cray CS6400 (codenamed SuperDragon.)

In October 1994, the SPARCserver 1000E (Scorpion+) and SPARCcenter 2000E (Dragon+) were announced, offering a 50 MHz XDBus; a whopping 25% faster than the original 40 MHz bus. (Or in today's terms, 'a bit less slow'.) It appears that these were marketed alongside the 40 MHz versions rather than being replacements.

On announcement, an entry level SPARCserver 1000E with two 60 MHz CPU, 64 MB of memory, 2 x 1.05 GB disks, CD-ROM, and Solaris was priced at $54,300.
This particular 1000E has two system boards with two 60 MHz CPU on each, 256 MB of memory, 2 x 1.05 GB disks, CD-ROM drive... and I paid £21 for it (including some other Sun stuff) in 2005. Not much else depreciates as rapidly as computer hardware!

Still, it's closely related to a Cray and has developmental links to Xerox PARC; that makes it interesting... I think...

It's a 5U rack-mountable desktop case with a backplane between the front and rear sections. Up to four system boards may be installed in the rear section; the lower half of the front section contains the power supply, above this is the SCSI tray to which the system control board is fitted. It all comes apart fairly easily, handy if it happens to have an elusive intermittent fault...

When I got it, it wasn't happy. It'd refuse to do anything (apart from sounding like a hovercraft.) Eventually, after several minutes repeatedly hitting the reset button, it'd almost start but then lock up again. After a little longer, it could be persuaded to start up and ran long enough for me to install Solaris, but would still randomly lock up.

I shoved it to one side and ignored it until a couple of weeks ago. It hadn't fixed itself in the meantime.

In case anyone's interested, Oracle (who now own Sun) still have manuals available for download here.

I'll end this post with some pictures of the innards. I think I've got to the bottom of the problem; but as I'm waiting for parts I've got time to write up the saga so far. Watch for the next exciting installment, coming soon!
  • Exploded diagram (pinched from the manual.)
  • Power Supply.
  • SCSI tray and control board.
  • First system board (two CPU modules, three SBus modules, 128 MB RAM.)
  • Second system board (two CPU modules, no SBus modules, 128 MB RAM.)
Kat.
Attached Thumbnails
Click image for larger version

Name:	SS1000E_diagram.gif
Views:	203
Size:	4.7 KB
ID:	52847   Click image for larger version

Name:	SS1000E_01.jpg
Views:	222
Size:	41.1 KB
ID:	52848   Click image for larger version

Name:	SS1000E_02.jpg
Views:	357
Size:	78.0 KB
ID:	52849   Click image for larger version

Name:	SS1000E_03.jpg
Views:	260
Size:	121.2 KB
ID:	52850   Click image for larger version

Name:	SS1000E_04.jpg
Views:	239
Size:	118.6 KB
ID:	52851  

Kat Manton is offline  
Old 23rd Jun 2011, 6:00 pm   #2
Zelandeth
Heptode
 
Zelandeth's Avatar
 
Join Date: Jun 2006
Location: Milton Keynes, Buckinghamshire, UK.
Posts: 931
Default Re: Sun SPARCserver 1000E Saga

Interesting, always had a soft spot for computer equipment from that sort of era. Not got anything in terms of actual machines from that level, though I do have a 1.2Gb SCSI hard disk from 1991 if I remember rightly, which was in active use until only a few years ago. What you have there I imagine is the sort of market that was aimed at.
Zelandeth is offline  
Old 23rd Jun 2011, 7:21 pm   #3
Kat Manton
Retired Dormant Member
 
Kat Manton's Avatar
 
Join Date: Jul 2005
Location: West Yorkshire, UK.
Posts: 1,700
Default Re: Sun SPARCserver 1000E Saga

Part II

The first thing I tried was the usual obvious stuff when dealing with old computers. Often, it's just a dodgy connection somewhere; just unplugging and reconnecting everything is enough to get 'em going again. I removed the memory SIMMS, cleaned the edge contacts and refitted them, removed and replaced the processor modules and SBus cards and so on. This made no difference.

As it's comprised of several of the same thing, I also tried swapping the system boards and running the machine with just one of them and then the other. This, too, made no difference whatsoever.

I also tried lifting it a couple of inches and dropping it, swearing at it and threatened it with being hurled off Beachy Head. Still it continued to sulk.

Each system board has a row of LEDs visible through a hole in the back panel. These display a number in binary as POST works through the tests. When it failed to start, the first system board showed the value 0x05.

POST User's Guide (801-2916-10.pdf) - Section 2.1 - Early Post Tests
  • Finally, POST does a basic BootBus SRAM read/write test. POST tests 8 bytes of SRAM at the SRAM base address +8. If a failure is detected, POST attempts to print a message on TTYA, then falls into and remains in a write/read scope loop for as long as the failure persists. If this test fails, you see the value 0x05 in the LEDs.
So the SRAM is failing unless the system has been on for a while? On both system boards whether one, the other or both are installed? That doesn't sound right..!

Up to this point I was running it with a keyboard and monitor connected to a framebuffer SBus card. Unlike a PC, which initialises the display hardware fairly early and can report errors during POST (or emit coded sequences of beeps if that fails), this machine initialises the framebuffer card quite late and most of the time wasn't getting that far. If it did get as far as initialising the framebuffer (about a minute after power-up or hard reset), it would continue to work.

It does, however, initialise the first serial port on the first system board early and will produce diagnostic information via that if the front-panel keyswitch is turned to the third 'diagnostic' position.

I have two Wyse serial terminals but they're buried in the depths of the back room. A few years ago I'd been given a dead DEC VT510 (with no keyboard) and had chucked it in the garage; I half-jokingly speculated that it might be quicker to dig that out and fix it than unearth one of the working terminals. Amazingly, that was the case; the DEC merely had an o/c resistor in its PSU and took minutes to repair (then several hours to clean.) As an added bonus, these things work with a standard PS/2 keyboard, of which I have many.

(See Obsolyte - Serial Console on a Sun or SGI System for cable wiring etc.)

With the terminal connected, I could see what was going on. Or not. Nothing appeared on the terminal. I kept hitting the reset button and eventually, a couple of characters appeared. Then a few more. Then part of the first diagnostic message. Finally, several lines of diagnostic messages appeared then abruptly halted with this on the terminal:

Click image for larger version

Name:	SS1000E_05.jpg
Views:	494
Size:	54.4 KB
ID:	52893

At last! A clue! Exciting, isn't it?

Kat
Kat Manton is offline  
Old 23rd Jun 2011, 8:24 pm   #4
Kat Manton
Retired Dormant Member
 
Kat Manton's Avatar
 
Join Date: Jul 2005
Location: West Yorkshire, UK.
Posts: 1,700
Default Re: Sun SPARCserver 1000E Saga

Hi Zel,
Quote:
Originally Posted by Zelandeth View Post
Interesting, always had a soft spot for computer equipment from that sort of era. Not got anything in terms of actual machines from that level [...]
I might be able to help out with that...

I have 14 Suns, 6 SGIs and a DEC AlphaServer! It's really too many and I need to thin the collection out somewhat. There's something wrong with most of them but, rather than chuck them I figured I'd try to sort them out and see if anyone's interested in working machines, maybe even with an OS installed.

It's partly why I'm going to post about them as I sort them out; maybe I can generate a bit of interest in them and re-home a few? To my mind, they're a lot more interesting than old PCs (which are pretty much like newer PCs only slower); they run one or another flavour of Unix, they're proper multi-user, multi-tasking systems, they're easily networked, some open-source software will run on them and there's plenty of information available if you know where to look.

Ataris and Amigas and other home computers have their following; these old Unix boxen date from around the same time but were, of course, vastly more expensive and out of reach to most. Anyone who did use them at the time may have less than fond memories of slaving away over one at work rather than playing 'Lemmings' until 4am every night...

Okay, time for a coffee then I'll sort out a diagram and some more photos for Part III of this tale of excitement, adventure and... yes, I know, it's just a big old Unix box...

Kat
Kat Manton is offline  
Old 23rd Jun 2011, 8:44 pm   #5
donutty
Pentode
 
Join Date: Aug 2010
Location: Sheffield, South Yorkshire, UK.
Posts: 163
Default Re: Sun SPARCserver 1000E Saga

Kat, you should probably check out the Vintage Computer forum if you haven't already.

http://www.vintage-computer.com/vcforum/forum.php


I've got a MicroVAX 3300 and currently looking for a HDD for it, and a couple of SparcStations and an SGI Indy, oh yeah and a great big hulk PDP 11/70
donutty is offline  
Old 23rd Jun 2011, 9:18 pm   #6
wave solder
Hexode
 
wave solder's Avatar
 
Join Date: Feb 2007
Location: Chesterfield, Derbyshire, UK.
Posts: 481
Default Re: Sun SPARCserver 1000E Saga

Very interesting Kat, I love reding about proper computers such as these and have a selection of my own. The problem is space. On a smaller scale I am currently repairing a Sun Voyager (portable) which is missing it's HDD. this HDD is not a very common one being 2.5" 50 pin SCSI, so if anyone happens to have one.......
wave solder is offline  
Old 23rd Jun 2011, 9:53 pm   #7
julie_m
Dekatron
 
Join Date: May 2008
Location: Derby, UK.
Posts: 7,735
Default Re: Sun SPARCserver 1000E Saga

Mmm, interrupt weirdness! That could be as simple as a bad connection, or a dodgy decoupling cap on a logic IC. Scope the power rails; you never know, you might just see something.

By the way, when you get Solaris up and running, you'll probably find your favourite commands don't support all the options you expect them to
__________________
If I have seen further than others, it is because I was standing on a pile of failed experiments.
julie_m is offline  
Old 24th Jun 2011, 12:56 am   #8
Kat Manton
Retired Dormant Member
 
Kat Manton's Avatar
 
Join Date: Jul 2005
Location: West Yorkshire, UK.
Posts: 1,700
Default Re: Sun SPARCserver 1000E Saga

Part III

'AC Power Failure Detected' seemed to be the clue. The longer it was left on, the longer it'd take before it stopped with the same message every time.

Okay, I have no schematics and probably little chance of getting hold of any, either. But the System Board Manual (801-2900-12.pdf) Section B.5 includes the pinout of the system board backplane connector (four rows of 60 pins) and one pin, C50, is labelled 'AC_Fail'.

It seemed reasonable to assume that 'AC_Fail' came from the PSU but maybe it passed via the control board first? Anyway, I needed to know where it originated and there was only one way to find out, a spot of reverse engineering.

I emptied the cabinet and removed the backplane. All four system board connectors had pin C50 commoned; they connected to pin A56 on the control board but were not connected to the PSU. So I needed to trace the signal back through the control board.

Pin A56 connected to pin 4 on U0204, a 74BCT8244. That's basically a '244 octal buffer with a JTAG port. The input is on pin 21 and this traced back via a bit of filtering and a pullup resistor to pin B58. Back to the backplane...

Pin B58 on the control board traced back to pin D14 on J2, the lower of the two connectors on the PSU. So now I knew where it came from, but what, if anything, was it doing?

Access for probing around with a 'scope is pretty much impossible, the chassis doesn't come apart enough. So I soldered bits of wire onto pin B58 and a convenient ground on the control board and reassembled the system.

Now with the 'scope connected to the 'AC_Fail' output from the PSU, I powered up the system. Bingo! A TTL level 50 Hz square wave!

As I watched, the positive-going pulses got narrower with their width wobbling around. Slowly but surely, they narrowed further and started to appear less often. Eventually there was just an occasional pulse and after a few minutes these were less frequent and almost vanished. At this point, the system would run through the full set of POST diagnostics and get as far as attempting to boot fron the network (which wasn't connected, so I just got error messages.)

50 Hz? That seemed reasonable; I expected the 'AC_Fail' signal to be derived from somewhere fairly early in the PSU, possibly straight from the mains input, so the system has time to do something before the reservoir capacitors discharge. (I've designed systems along these lines, using a power fail signal to trigger a non-maskable interrupt, giving the software just enough time to save information to non-volatile memory then halt.)

That narrowed things down a lot, the fault must be somewhere in the PSU and that 'somewhere' should be a fairly simple bit of circuitry which I could trace out without reverse engineering the entire PSU.

And that brings Part III to a conclusion. I'm pleased to note there's some interest in this; it makes it worth writing up. Plus the techniques used can be applied to other bits of equipment so it's hopefully educational. Lack of schematics doesn't prevent repair; if anything it makes the detective work more interesting

The next exciting instalment will bring us to the point I'm at now, where I just need to fit some new bits then see if I've fixed it...

Kat
Kat Manton is offline  
Old 24th Jun 2011, 3:16 am   #9
Kat Manton
Retired Dormant Member
 
Kat Manton's Avatar
 
Join Date: Jul 2005
Location: West Yorkshire, UK.
Posts: 1,700
Default Re: Sun SPARCserver 1000E Saga

Part IV

This is one serious PSU. It's heavy and that case is full.

Some details:
  • Zytec power supply, 650W, part number 22917100
  • Sun part number 300-1103-03
  • 900W input
Output:
  • +5V/100A
  • +1.2V/40A
  • +12V/2.25A
  • -12V/0.25A
  • +12V/7.5A, 12A pk/25sec
(That last one is for hard drives, I suspect.)

Removing all the visible screws on top as well as all the screws around the top edge of the sides reveals... the back of a PCB.

Click image for larger version

Name:	SS1000E_06.jpg
Views:	354
Size:	59.7 KB
ID:	52910


Next, these two groups of 4 screws need to be removed, these connect the 5V/100A supply and return bus-bars to the vertical PCB with the two backplane connectors on it.

Click image for larger version

Name:	SS1000E_07.jpg
Views:	367
Size:	40.3 KB
ID:	52911


Now the upper PCB can be removed; this appears to contain just the 5V supply.

Click image for larger version

Name:	SS1000E_08.jpg
Views:	347
Size:	45.9 KB
ID:	52912


Finally, after removing another few hundred screws, the lower board can be removed.

Click image for larger version

Name:	SS1000E_09.jpg
Views:	393
Size:	69.9 KB
ID:	52913


And that's one short of the image limit. Continued in Part V...

Kat
Kat Manton is offline  
Old 24th Jun 2011, 6:05 am   #10
Kat Manton
Retired Dormant Member
 
Kat Manton's Avatar
 
Join Date: Jul 2005
Location: West Yorkshire, UK.
Posts: 1,700
Default Re: Sun SPARCserver 1000E Saga

Part V

Tracing from pin D14 on the lower connector led me to the collector of Q1 (2N4401) and these highlighted components:

Click image for larger version

Name:	SS1000E_10.jpg
Views:	218
Size:	91.2 KB
ID:	52914


Tracing from there led me to the optoisolator (4N26) and these highlighted components:

Click image for larger version

Name:	SS1000E_11.jpg
Views:	246
Size:	91.2 KB
ID:	52915


I've condensed the important bits of my awful freehand ballpoint-pen sketches and notes down to a a proper schematic, including the backplane and control board:

Click image for larger version

Name:	SS1000E_AC_Fail.png
Views:	287
Size:	46.5 KB
ID:	52916


The TO92 'transistor-looking' thing near the opto-isolator is a TL431AC Adjustable Precision Shunt Regulator IC (you should be able to find a datasheet by googling for the part number and 'datasheet'.) Here, it's used as a comparator; an application noted in the Philips datasheet. An interesting and potentially useful device; I'll file that one for future reference.

The power fail circuit is fed with full-wave rectified mains. Just rectified, not smoothed; the capacitors are on the 'other side' of some beefy devices on a heatsink which I suspect are SCRs. So the circuit should be getting lumpy DC; that would be 100 Hz pulses. The 'AC_Fail' line had 50 Hz pulses on it though...

Anyway, I checked the components. Resistors were all correct, I hadn't got a replacement for the 1uF capacitor so I removed it and measured the value with the trusty Marconi TF2700 bridge then checked it for leakage with the Megger BM8/2 then put it back. Given the circuit was producing something, it seemed reasonable to assume the TL431AC and 4N26 were okay.

50 Hz, not 100Hz... this was bugging me and I don't like working on the live side of SM power supplies. Maybe one diode in the bridge rectifier had partially failed, recovering as it warmed up? That'd put 50 Hz pulses of half-wave rectified DC on the power fail circuit...

For less than 2 quid for a 1000V direct replacement for the 600V/25A original (GBPC2506W), it seemed worth a try. I ordered one then tackled something simpler, building one good Sun Type 5c keyboard from two knackered ones. When it arrived, I stuck the new bridge rectifier in, reassembled the thing and tried it again. Guess what? It made no difference. Oh well, the new one can stay put and the old one can go in my diodes bin.

I gave in. It was time to drag the isolating transformer downstairs and attempt to hang a 'scope on bits of the live side of this thing with it running, without either killing myself or blowing up the PSU.

I figured out how to detach the back half of the keyswitch and unscrewed the LED PCB so I could leave the front panel off. Then I soldered some bits of wire onto strategic points of the circuit and fitted the upper PCB but left the top cover off.

With the 'scope ground connected to the 0V return on the live side, one probe connected to the +ve side, the terminal still connected, keyswitch set to 'diag' and rear panel mains switch set to 'on', I stood back and switched the mains on at the socket on the isolating transformer. (Yes, I'm a coward.) Full-wave rectified mains on the 'scope and... masses of POST messages scrolling up the terminal. So it's decided to work now?

I turned it off, reconnected the 'scope to the 'AC_Fail' line with ground to the 0V DC return and fired it up again. 100 Hz pulses and nothing on the terminal. 100 Hz..? I turned it off again then plugged it straight into the mains. 50 Hz pulses...

Now this was getting strange. Next I plugged it back into the isolating transformer and connected the 0V return of the live side of the bridge to chassis with a croc-clip lead. 0V on the 'AC_Fail' line and POST messages. Remove clip from chassis, 100 Hz pulses and 'AC Power Failure Detected' messages scrolling up the terminal...

That had to be several clues. It exhibited different behaviour depending on whether it was powered from the mains (neutral near ground) or the isolating transformer (live/neutral both floating w.r.t. ground) and behaved itself with the live side 0V return grounded. This odd behaviour meant that something which should be isolated, wasn't..?

I disconnected everything then connected my Megger between the 0V DC return and the 0V live-side return and measured 5 megohm on the 50V test range... Aha..!

I pulled the PSU out of the chassis and found the leakage was still present. It was still there with the upper PCB removed. I removed the main board again and took it back to the bench. All the capacitors to ground in the mains filter checked out okay. The two vertical boards in there have opto-isolators on them; I wondered if one of them was breaking down and removed both boards.

Click image for larger version

Name:	SS1000E_13.jpg
Views:	240
Size:	143.9 KB
ID:	52917


The leakage, now down to 4 megohm, was still present; meaning the fault was somewhere on the main board.

Finally I spotted it. With the two control boards removed, I could see traces of something spreading out from around an electrolytic capacitor. Traces of electrolyte; conductive goo spreading across the board over the isolation gap and all around the power failure detection circuitry...

I wiped a cotton bud over a bit of the board; now the leakage resistance was over 50 megohm. The stuff was all over the place so I removed a few more components and cleaned the film of goo off thoroughly with IPA. Now I had no reading (infinite resistance) on the 1000V range. Better...

Click image for larger version

Name:	SS1000E_12.jpg
Views:	239
Size:	102.2 KB
ID:	52918


C121 was the culprit; 2200uF, 25V, 105 degC; no sign of bulging or other distress but notably located down-wind of a heatsink. It was also one of those tall thin ones of a type I haven't got; so I started drafting this lot after ordering some.

While it was in bits I noticed several small 85 degC electrolytics dotted around; though these seemed healthy according to my ESR meter they're old and live somewhere warm. My experience is that it's almost always the physically small low-value ones which fail so I'll replace these with 105 degC ones while I'm at it.

"Has she fixed it?", I hear no-one shout...

I have no idea. This massive missive brings the SPARCserver saga right up to date. The capacitors turned up yesterday afternoon but I resisted temptation in order to finish writing this. I'll install yet more vital live-giving weapons-grade coffee, fire up the Weller TCP and glue some capacitors in and let you know...

Kat
Kat Manton is offline  
Old 24th Jun 2011, 9:24 am   #11
Paul Stenning
Administrator
 
Paul Stenning's Avatar
 
Join Date: Dec 2002
Location: Cardiff
Posts: 9,060
Default Re: Sun SPARCserver 1000E Saga

It is certainly a substantial and well constructed piece of equipment. I can see why it was so expensive new.

Once we got to the PSU I was expecting it to be electrolytics. Though not quite in this way.
__________________

Paul Stenning
Forum Admin/Owner and BVWS Webmaster
Paul Stenning is offline  
Old 24th Jun 2011, 12:36 pm   #12
Herald1360
Dekatron
 
Herald1360's Avatar
 
Join Date: Feb 2007
Location: Leominster, Herefordshire, UK.
Posts: 16,528
Default Re: Sun SPARCserver 1000E Saga

Whilst I've little direct interest in the workings of old computers, I do enjoy these writeups and the problem solving that goes with them.

Is our Kat normally a nocturnal creature, or is it just a matter of getting carried away with the fascination of the particular job? Perhaps the lack of "Has she fixed it?" shouts is just that most of us were fast asleep at the time

Looking forward to the next instalment be it about this or some other recalcitrant bit of obscure old technology! (Here or elsewhere- how's the Viva/Shuvit saga going?)
__________________
....__________
....|____||__|__\_____
.=.| _---\__|__|_---_|.
.........O..Chris....O
Herald1360 is offline  
Old 24th Jun 2011, 1:27 pm   #13
julie_m
Dekatron
 
Join Date: May 2008
Location: Derby, UK.
Posts: 7,735
Default Re: Sun SPARCserver 1000E Saga

Got to admit, that's not quite what I was expecting to be the cause!

The bogus interrupt message had me thinking some (external to the CPU) interrupt-control logic was glitching, and causing a legitimate interrupt to be mistaken for something else (AC power failure sounded to me like the sort of thing that should generate an interrupt).

Still, well done Kat for persevering ..... I love a good detective story! (Where's the smiley with a deerstalker when you need one?!)
__________________
If I have seen further than others, it is because I was standing on a pile of failed experiments.
julie_m is offline  
Old 24th Jun 2011, 2:41 pm   #14
Kat Manton
Retired Dormant Member
 
Kat Manton's Avatar
 
Join Date: Jul 2005
Location: West Yorkshire, UK.
Posts: 1,700
Default Re: Sun SPARCserver 1000E Saga

Part VI

At long last, the moment of... something. I've replaced all the small electrolytics along with the one which caused all this, re-soldered any even slightly suspicious joints, cleaned the flux off then reassembled it (with no screws left over!)

I hadn't removed the the wires attached to the control board so I hooked the 'scope up to the 'AC_Fail' line, crossed everything and powered up. The 'AC_Fail' line sat at 0V and stayed there, POST messages scrolled up the terminal and everything behaved as it should.

Well, almost. It has no idea what time it is and won't remember what it's supposed to boot from, among other things. This is a common problem with old Sun hardware; the lithium cell in the clock/NVRAM IC has died. I've fixed a couple of these; this time I'll take photos.

It now gets this far with a keyboard/monitor attached:

Click image for larger version

Name:	SS1000E_14.jpg
Views:	316
Size:	57.0 KB
ID:	52924

Solaris (2.6) starts up if I boot it manually. I must've only got as far as installing it several years ago; it seems like a fairly minimal installation and has barely been configured.

It works!

(And I've been awake for well over 24 hours now. I get carried away...)

Kat
Kat Manton is offline  
Old 24th Jun 2011, 7:12 pm   #15
Zelandeth
Heptode
 
Zelandeth's Avatar
 
Join Date: Jun 2006
Location: Milton Keynes, Buckinghamshire, UK.
Posts: 931
Default Re: Sun SPARCserver 1000E Saga

Nicely done! Good fault that one which surely must have produced a very good "...?" moment when suddenly it started working on the isolating transformer!

I wonder how often it is that major problems with bits of equipment as substantially put together as this turn out to be faults with simple, cheap components like that...

Now let's see what it can do!
Zelandeth is offline  
Old 24th Jun 2011, 7:19 pm   #16
wave solder
Hexode
 
wave solder's Avatar
 
Join Date: Feb 2007
Location: Chesterfield, Derbyshire, UK.
Posts: 481
Default Re: Sun SPARCserver 1000E Saga

Over the years I have had IT equipment that has seemed to be past the point of no return with the most obscure faults which after much tracing and head scratching has come down to a very simple fault such as this. Well done Kat, time to get the Solaris installation completed.
wave solder is offline  
Old 26th Jun 2011, 6:05 am   #17
Kat Manton
Retired Dormant Member
 
Kat Manton's Avatar
 
Join Date: Jul 2005
Location: West Yorkshire, UK.
Posts: 1,700
Default Re: Sun SPARCserver 1000E Saga

Hi,

Now I've caught up on a few things (like those "sleeping" and "eating" things which get in the way of resurrecting dead technology), it's time I replied to a few posts...

Quote:
Originally Posted by donutty View Post
Kat, you should probably check out the Vintage Computer forum if you haven't already.
I probably should; I'm aware of them via The Retr0bright Project (which I'll be trying and writing about soon, some of my Sun bits are a little off-colour.)

Quote:
Originally Posted by donutty View Post
I've got a MicroVAX 3300 and currently looking for a HDD for it, and a couple of SparcStations and an SGI Indy, oh yeah and a great big hulk PDP 11/70
I'll be digging through hard drives soon; I'm not sure if I have something suitable but they'll appear in the 'Offered' section once I've made a list. I wish I had more space; I've often fancied a MicroVAX and/or a PDP 11. Maybe once I've thinned the collection out a bit I'll find room for something small that'll run VMS...

Quote:
Originally Posted by wave solder View Post
On a smaller scale I am currently repairing a Sun Voyager (portable) which is missing it's HDD. this HDD is not a very common one being 2.5" 50 pin SCSI, so if anyone happens to have one.......
I'm afraid I haven't got one of those. But, will it net boot? Most/all of my Sun kit will. It renders it rather less portable but at least you could use it as long as it's attached to your LAN and can boot and mount filesystems from a server.

Quote:
Originally Posted by ajs_derby View Post
By the way, when you get Solaris up and running, you'll probably find your favourite commands don't support all the options you expect them to
I'm familiar with this source of confusion; it doesn't help that I could be logged on to Solaris/IRIX/etc. machines remotely from my main Linux system and have an array of X apps, X sessions and xterms on different machines in front of me. One thing which helps is installing a few things from Sunfreeware (bash and vim are going on this thing asap, for a start!)

Quote:
Originally Posted by Paul Stenning View Post
It is certainly a substantial and well constructed piece of equipment. I can see why it was so expensive new.

Once we got to the PSU I was expecting it to be electrolytics. Though not quite in this way.
The build-quality of hardware like this is amazing; obviously built up to a quality rather than down to a price. I think it's one reason why I like this kind of thing; it's a pleasant change from consumer hardware. I find it hard to get excited about PCs after playing around with Sun, SGI and DEC hardware.

As a comparison with something a few people might be a bit more familiar with; the SPARCserver 1000E is about the same size and weight as an RCA AR88 communications receiver. (I used to have an AR88LF and couldn't lift that either.)

I left out a few bits in the interests of brevity (really!). Electrolytics were naturally suspects; when I first pulled the PSU to bits I waved my ESR meter over all of them and they all checked out okay.

Quote:
Originally Posted by Herald1360 View Post
Whilst I've little direct interest in the workings of old computers, I do enjoy these writeups and the problem solving that goes with them.
Thanks very much; I enjoy writing them (especially if I appear to have an audience.) I have to admit I've probably amassed quite a bit of experience; writing fairly detailed blow-by-blow accounts of repairs/restorations seems like a good way of passing some of it on. I still get the odd real puzzler - like this one!

Quote:
Originally Posted by ajs_derby View Post
Got to admit, that's not quite what I was expecting to be the cause!
Quote:
Originally Posted by Zelandeth View Post
Nicely done! Good fault that one which surely must have produced a very good "...?" moment when suddenly it started working on the isolating transformer!
I trimmed down my investigations rather a lot; it was a real head-scratcher!

Now the cause is known, what was happening makes a lot of sense. The poor old thing was getting 'AC_Fail' interrupts every 20 ms, something I suspect the firmware programmers might not have allowed for. The line's supposed to remain low normally then go high once shortly before all supply rails disappear.

As to why; I think the input impedance of that TL431AC used as a comparator is pretty high and it's fed via 150k. The input should sit at 6.2V, the threshold is 2.0V, so it just needs a negative-going swing of 4V imposed on it and it'll trigger. With AC mains and rectified mains nearby, there's a lot of volts around; then throw in a 'distributed resistor' spread over the top of the PCB...

Powering from the isolation transformer would result in different voltages w.r.t. ground floating around compared to powering it straight off the mains; I suspect it would've also done something different if I swapped live and neutral round; which I was tempted to try. Grounding the return side of the rectified mains would change things again... and none of this should've made the slightest bit of difference. I've had some weird faults; this is a corker!

It doesn't even look that bad; look at the second photo in post #10. There's a splodge below and to the right of the TL431AC and below the lower of the two control boards but it's not screaming out for attention.

That's now filed for future reference; never underestimate the conductive properties of capacitor electrolyte and if in doubt, clean the board!

Quote:
Originally Posted by Zelandeth View Post
I wonder how often it is that major problems with bits of equipment as substantially put together as this turn out to be faults with simple, cheap components like that...
Quote:
Originally Posted by wave solder View Post
Over the years I have had IT equipment that has seemed to be past the point of no return with the most obscure faults which after much tracing and head scratching has come down to a very simple fault such as this.
It's fairly common IME (though I do tend to end up with dead stuff either bought cheaply because it's dead or just given to me, so my sample is biased.) But indeed, one cheap component can render hugely expensive (or formerly hugely expensive) very complicated hardware completely useless. It's why I'll have a go at the really complex stuff; it almost always comes down to something simple in the end. It's fun and very satisfying when it all starts to work again

Kat
Kat Manton is offline  
Old 26th Jun 2011, 7:03 am   #18
Kat Manton
Retired Dormant Member
 
Kat Manton's Avatar
 
Join Date: Jul 2005
Location: West Yorkshire, UK.
Posts: 1,700
Default Re: Sun SPARCserver 1000E Saga

Quote:
Originally Posted by Zelandeth View Post
Now let's see what it can do!
Quote:
Originally Posted by wave solder View Post
Well done Kat, time to get the Solaris installation completed.
I gave up on the Solaris 2.6 installation which was already on the two internal 1 GB drives. It was about as minimal as it can be, probably as the thing wouldn't keep running for long enough for a full installation (or installing a later version.)

I have a Sun 711 StorEdge MultiPack enclosure (fast wide SCSI) with 3 x 4 GB drives in it and there's a QLogic fast wide SCSI SBus card in the machine, so it seemed a good idea to use that. I've installed Solaris 8 now, the last version which supports sun4d hardware (and it took ages using the internal CD drive. I think it's a 4x one...)

Once I'd configured networking and it was on my LAN and could access the Internet, I gave Netscape 4.74 a try (it's included with Solaris 8.) Useless! But at least I could just about access the Opera website, download an old version and install it. Some of the Internet now works a bit. I'll see if any later versions work; I just took a guess that Opera 6 might be okay on Solaris 8.

So, what will it do?

Well, without doing any further configuration I tried remotely logging in from X running on my PowerBook.

Click image for larger version

Name:	admiral_login.jpg
Views:	306
Size:	18.3 KB
ID:	53006

I'd already created an account for myself so I just needed to log in.

Click image for larger version

Name:	admiral_DT1.jpg
Views:	351
Size:	91.1 KB
ID:	53007

That's CDE (The Common Desktop Environment.) Now consider it was contemporary with Windows 3.1...

For a start, you get four desktops (selected with the 'One' through 'Four' buttons.) I had a console (su'ed to root for installing stuff) and file manager on 'One', Netscape on 'Two' and eventually, Opera on 'Three'.

Click image for larger version

Name:	admiral_DT3.jpg
Views:	354
Size:	97.3 KB
ID:	53008

It's quite usable; this is a system dating from 1995 with four 60 MHz processors, 256 MB RAM and it's communicating with the X server on my Mac via a 10Base-T network connection. This being a 'proper computer' and using the X Window System, several users could log in and use the system simultaneously, maybe from SPARCstations or X terminals (I've got some Tektronix X terminals; I might set it up for those.)

(I had to reduce the size of the window grabs and save them as .jpg for the forum; it's much sharper at the original resolution on the PowerBook's LCD.)

Kat
Kat Manton is offline  
Old 26th Jun 2011, 8:02 pm   #19
neon indicator
Retired Dormant Member
 
Join Date: Jun 2010
Location: Co. Limerick, Ireland.
Posts: 1,183
Default Re: Sun SPARCserver 1000E Saga

I've a load of 9Gbyte fast & wide SCSI for price of postage if any use. Retired due to upgrade.

Not sure how many. Could be as much as 5
neon indicator is offline  
Old 27th Jun 2011, 8:56 am   #20
richrussell
Heptode
 
richrussell's Avatar
 
Join Date: Jul 2008
Location: Selby, North Yorkshire, UK.
Posts: 979
Default Re: Sun SPARCserver 1000E Saga

It's nice to see CDE again - at university (York, 1992-1995) I spent most of my first year on VT220s connected to the VAX cluster, but after a while migrated to the electronics department's Sun 3/50 workstations. And then onto SGI Indigos. Jumping between VMS, Solaris, IRIX and Slackware Linux was certainly a rapid way to learn things. Certainly the opinion of Windows 3.11 was that 'well, it's prettier than a text console, but not as useful as VMS or Unix - adequatefor the students doing Arts degrees I guess'
richrussell is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 1:37 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.