Skip to main content
shopping_basket Basket 0

Restoring a Sun SPARCstation IPX Part 1: PSU and NVRAM

Andrew Back
Open source (hardware and software!) advocate, Treasurer and Director of the Free and Open Source Silicon Foundation, organiser of Wuthering Bytes technology festival and founder of the Open Source Hardware User Group.


November 19, 2020 08:53

Also the KiCAD schematic in case you want to make changes.

0 Votes

November 23, 2020 08:23

@ms70 that's very kind of you, many thanks for sharing these!

November 19, 2020 08:53

I'm sure there are errors, but maybe it will still help someone.

0 Votes

November 19, 2020 09:01

  • Moderated

@ms70 Interesting and thanks for sharing! Sure, would like to see the schematic. I think you should be able to attach it in a comment on this post.

0 Votes

November 12, 2020 08:50

Greetings Andrew,

I found your blog here since I was doing very similar work on an IPC. It has been very helpful. I also have to say that your NVRAM battery hack is much nicer than mine!

The main issue I had to deal with was the power supply. I also replaced all the caps (except the big line cap), and then started experiencing the same problem. It would click, kick the fan slightly, then turn off. But I think I fixed it.

In my case, the system would turn on just fine ~10% of the time. It seemed to be random. I was convinced at first that it was a problem with the funky magnetic thermal switch (an interesting piece of kit, actually, though the design looks a little touchy), but that checked out. Nothing was showing up on the thermal camera, either, and everything I could test in-circuit seemed fine. So I made a rough schematic of the power supply to track down the issue. (It has some voltage readings on it as well. I'm happy to share it.)

I'm pretty sure the fault ended up simply being the 5V adjustment. It was adjusted slightly high (around 5.2 volts when it was working). You might just try turning yours down a bit (CCW). When it's too high, it triggers a fault. Since adjusting that to about 5.02, mine has turned on reliably. You might have similar luck.

0 Votes

August 13, 2020 11:09

Normally a SMPS only has the transformer and the feedback optocouplers between the mains side and the low voltage side. The SparcIPX power supply is very unusual in that it also uses two safety capacitors to bridge the gap. This gives improved noise performance from memory. However if both fail then bad things will happen. The idea is that with two of them it will take two to fail for bad things to happen which is much less likely. However these capacitors are now ~30 years old and much more likely to fail than when they where new and they will fail silently. You won't see this very often if at all on a modern PSU. I would recommend replacing them. They are marked C111 and C112 on the silk screen.

0 Votes

August 17, 2020 07:09

@jabuzzard Many thanks, that's an interesting insight and a really useful tip. I'm not with the machine just now, but think I can see those in one of the above photos. Are they just ceramic caps? Or maybe tantalum? Would seem odd if the former was prone to failing. In any case, will look to replace them and do this as the same time as swapping out the fan, which works OK, but I think a small investment in a fresh new, more modern and less noisy one will be well worth it.

August 25, 2020 08:45

@Andrew Back They are X1/Y1 suppression capacitors. Something like RS Stock No. 133-5623 would be a suitable replacement (if I am correct in thinking they are 2.2nF. I am guessing a bit as my board has been totally depopulated - long story). These sorts of capacitors are supposed to be very reliable, but they *do* fail from time to time. Earlier this year I have a X2/Y2 suppression capacitor (which should be even less prone to failure) fail on my in spectacular magic smoke fashion on an old strimmer. Also a couple of years ago a suppression capacitor failed on my mothers sowing machine. It just silently failed in a shorted condition, which proceeded to blow a fuse but not good. In both cases the equipment was of a similar age to any SPARCstation IPX. Like I said won't see it on any modern design and I am sure my 2nd edition of Horowitz and Hill describes it as a "bad idea".

August 25, 2020 08:41

@jabuzzard Thanks, so they look to be Y2 class. I've now replaced them and you can read about the most recent work on the IPX here: I also have an old Tektronix (late 80s?) scope where an X class went bad. *Lots* of smoke... At the time I found a pretty funny YouTube video, where someone demonstrated a solution for this exact problem and scope model, where they basically just pointed a large noisy fan at the scope to blow the smoke away. Can't seem to find it now and suspect it was taken down for some reason.

August 4, 2020 07:10

You will be interested to know that a large portion of the photomasks, used to make integrated circuits, manufactured today are built with equipment still using Sun Sparc 2s, 20s and Ultrasparcs. It's a testament to the quality of Sun computers. In fact, a lot of the equipment in mask shops and wafer fabs still run DOS or Win NT. The replacements, if there are any, are too expensive for the mature market.

0 Votes

August 4, 2020 16:37

@deaddog Interesting indeed, thanks. I guess this is much the same as how PDP-11 computers often ended up in process control and remained there for many years, quietly and reliably doing their job. I'm also told that Transputers got used a lot in MRI machines and systems based on these were maintained way beyond when people stopped using them in new designs. All these architectures seem to find one or more particular niches and committed advocates who appreciate what they bring to that area.

July 13, 2020 08:01

You can set your sparcstation to boot disk via command at OBP:
setenv boot-device disk

0 Votes

July 14, 2020 07:36

@Topkek456 Yes, indeed. Planning to take a look at OBP a little more in the next post in the series. It's pretty cool and the OpenBoot reference manual is a nice read.

July 2, 2020 14:42

@Boss and Andrew, here is a short pdf that helps predict reasonable life expectations for electrolytic caps:

From that source,
Aluminum electrolytic capacitors slowly degrade over time and once the capacitor has degraded beyond a specified amount, the capacitor is considered to have failed. Most capacitors are considered a failure when the capacitance has changed by 20 to 25% of its initial value. Aluminum electrolytic Capacitors load life’s ratings are generally expressed between 1000 and 10000 hours at their rated voltage, maximum temperature rating and with maximum ripple current applied to the capacitor. This means that the capacitance of the capacitor will not change by more than the amount indicated under the load life rating when the capacitor is operated at the stated conditions. Although the life expectancies appear be a short amount of time the following can increase them. When the capacitor is operated at temperatures other than the maximum rated temperature for the capacitor the expected life of the capacitor will increase. The rate of increase in operating life is for the life to double for every 10°C decrease in temperature (Arrhenius’s law).

The paper points out the life is also affected by applied voltage vs rated voltage, and by temperature rise due to ripple current, and provides formulas.

0 Votes

July 6, 2020 08:35

@BradLevy, thank you, That gives me some confidence and explains why some of my older equipment still functions! Think operating temperature has been in my favour as well as s good safety margin in the voltage rating. I like the point about "Excessive reverse voltage"... a colleague inserted a small electrolytic the wrong way round and the end of the capacitors' life was instant!!! A very loud bang and the can and contents spread over a large area. This was only at 5 or 15 volts, nothing high voltage. I guess if you want long service life and space is not a constraint you need to design out electrolytics. I read an interesting article that PV industry does not use electrolytics and use 'film' types for long life. I wonder what the satellites use or indeed the military by choice? I recall back in the 80's tantalum was the choice, but did not investigate the reasons. Anyway a very good paper Brad and an interesting article Andrew.

July 2, 2020 14:42

@BradLevy Thanks, Brad! I've read bits of advice on e.g. temperature related effects, but that looks like it could be pretty comprehensive and will take a look, appreciated.

July 2, 2020 09:47

Good job!
I always find electrolytic's and 'life' a concern and a mystery! Even one of your replacements states "Panasonic FC Series Type A electrolytic capacitors offer a high endurance, low impedance solution. These aluminium capacitors feature endurance between 1000 and 5000 hours." so although this is at worst case temperature, but an expected life of 40 to 200 days in continuous use?
I could not see a spec that showed endurance against temperature, but hopefully much better!

0 Votes

July 2, 2020 14:43

@Boss I'm relieved to hear that I'm not the only one who finds electrolytic lifetime ratings a little mysterious, but expect it must mean worst combination of operating conditions (temperature + ripple current etc.) and degradation by some degree — hence not "dead" at this point and rather instead just not performing so well (still a way to go until ruptured and with electrolyte leaking, although hopefully this would never happen...)

Related Content

DesignSpark Electrical Logolinkedin