Hacker News new | comments | show | ask | jobs | submit login
Restoring YC's Xerox Alto, day 1: Power supplies and disk interface (righto.com)
117 points by kens on June 22, 2016 | hide | past | web | favorite | 23 comments

Not sure it was mentioned in the last thread on this computer but if anyone wants to actually use a real Alto and many other vintage computers check out the Living Computer Museum in Seattle: http://www.livingcomputermuseum.org/ They have an Alto you can walk right up to and start hacking on--when I last saw it it was running a pretty fun billiards game (worked well on its vertical monitor). There are tons of other vintage computers you can use, including even giant mainframes in a cold room like an IBM System 360, PDP-10, VAX systems, and a lot more.

Keith at the Living Computer Museum has been very helpful with our Alto restoration, providing information from their restoration experience, as well as diagnostic hardware.

When I visited, I had a hard time refraining from taking the cover panels off the VAX, just like days of yore. (In the mid 1980s I essentially had personal use of an 11/780 for a few years. It was fun).

I highly recommend the LCM.

As cool as restoring the power supplies in their original condition is, it might be a lot easier at least temporarily to replace them with modern 5V and 12V units. Also a great deal less damage risk to the electronics on the PCBs which is what's truly rare and irreplaceable. Then you can repair and test the 'old' power supplies at your leisure and put them back into use if full authenticity is required.

This might not be as easy or practical as it sounds. Replacing an existing power supply that is known to match the requirements of the system with one that was cobbled together from modern components is inherently risky. You're either building something brand new and un-tested or hacking an existing supply to fit a job it wasn't designed for. If the original is reasonably intact and can be verified, it would be simpler and safer to repair it and know the supply matches exactly.

The original post shows that it needs +15V, -15V, +5V and -5V. This could be implemented with four separate high quality active PFC modern power supplies tuned to the right voltages. One thing that's unknown is what the load is for the whole thing, so it'd be necessary to somewhat deliberately oversize the power supplies, but it's not hard to find 700W rated power supplies that will do 15V.

Any modern power supply will be safer and more reliable. If there's any doubt about the pinout and cabling arrangement of the old power supplies, it's totally possible to pull the old power supply, repair it as best as possible with new capacitors, power it up on a test bench and probe it to see exactly what its output is. Then match that on a new cable harness with a new power supply.

It's not "hacking" to provide +15V or -15V , that's within the range of many good quality tunable 12V power supplies.

You don't even need to tune a 12v supply. +15v -15v is a reasonably standard part.

True, though there's a much greater selection of really high quality (medical grade, even!) active PFC 12V power supplies that will tune to 15V.

Some of them even come in specific 15V versions like a Meanwell RSP-750-15:


I find it interesting that they only had to replace three capacitors in one power supply. Don't electrolytic capacitors (the most commonly used type) have a limited lifespan because of electrolyte evaporation? Naively, I would have expected that every single capacitor in the system would need to be replaced.

The average lifespan of electrolytic capacitors between 1999 and 2007 was much worse than usual, due to poor industrial espionage: https://en.wikipedia.org/wiki/Capacitor_plague

You could easily get the impression that electrolytics had really lousy lifespans as a result of this.

I work on machines that are usually 20-30 years old if not more and I pretty much have to replace all the capacitors on them due to age. They tend to dry out, crack, or balloon up.

Not a single one built after 1999.

Even when correctly manufactured they do have crappy lifespans, relative to most other things they share the board with. I wouldn't count on most wet capacitors lasting more than 10 years, even when handled with kid gloves. I've got plenty of things running with capacitors well beyond that age, but they were demoted from production to testing or nice-to-haves.

The lifespan of a electrolytic capacitor is strongly influenced by temperature[0]. A 10C increase in temperature can lead to a 50% reduction in lifetime, so its highly likely that the power supply that caps failed was slightly hotter during operation than the others.

[0]: http://www.kemet.com/Lists/ProductCatalog/Attachments/96/F33...

We measured some of the other capacitors and they were okay. I wouldn't be surprised if more bad capacitors turn up, though.

My BBC Micro required three capacitors to be replaced in the power supply and then it was right as rain.



As other posts have mentioned, its mainly due to a bad batch that was produced in the late 90s. I have 4 30-40 year old musical instruments that are on their original caps and work just great.

Thoroughly enjoying following this. Please keep the videos and articles coming.

Disk drive problems with an old computer reminded me of this discussion: https://news.ycombinator.com/item?id=11376711

Awesome, I actually have a platter disc that looks similar but just has a permanent opening in the side, it doesn't pop up on top like this one.

I can't wait to see an original version of smalltalk running on this!

When the hardware is fully documented, I'd love to see on Kickstarter a small board with some USB ports, HDMI output and an SD card slot emulating an Alto (or a Star). Extra points if it came with a keyboard with the same layout as the original (so you don't need to use a PC keyboard).

The Alto hardware is fully documented [1], source code is available [2] and a simulator is available [3], so there's not really anything stopping someone from making an Alto clone.

[1] Complete schematics at http://bitsavers.org/pdf/xerox/alto/schematics

[2] Source code at http://www.computerhistory.org/atchm/xerox-alto-source-code/

[3] Simulator at https://github.com/brainsqueezer/salto_simulator

I don't think it'd be an easy endeavor. I'm almost inclined to think it'd be easier to start from an emulator source towards a hardware implementation abstracting away stuff that's done in microcode hoping the software doesn't look too close.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact