When I was over in Oxfordshire to collect my Model 800, the gentleman who’d sold it to me advised me not to plug it into mains directly, without some current limiting device in series. A filament bulb, for example — to reform the electrolytic capacitors, as he explained.
With my lack of experience powering up vintage electronics, I wouldn’t have thought of that on my own. Curious, I read up a little on capacitor reforming. Essentially, electrolytics that have not seen electricity in many years need to get used to it again gradually.
04-MAR-2024
Before I begin, do I need any load on the DC side? As I learnt from a recent video by Usagi Electric on YouTube on restoring a PDP-11 power supply, his model H780 was designed to run just fine with no load; it would simply switch to linear mode.
This is further illustrated on the 5 and 12 volt regulators’ circuit diagram on the next page of the document quoted above. Those days, schematics were part of the regular documentation. What a time to have been alive!
Even though the Alpha’s H7883-YA PSU is slightly newer, I reckoned it would’ve been just as well engineered as that PDP-11 unit from the video, and there shouldn’t be any problem running it without a load.
The suggested course of action was to start with a 40 watt filament lamp in series with the line and leave things running for a few hours, to reform the caps. Then repeat with a 100 watt lamp. Now, in this day and age one doesn’t get to choose one’s incandescent light bulbs — the only one immediately available was 15 W.
I placed the voltmeter to monitor the AC input after the lamp. The camera is rolling in case something goes awry as I turn the power on for the first time in twenty years.
As the Nippon Chemi-Con caps charge up, the input voltage gradually settles on 47 V. (The video shows 42 V, but it was measured at the 600 VAC limit, so slightly inaccurate.) DC remains off. I let the PSU soak up the electrons for an hour or so, and call it a day.
05-MAR-2024
For the next step I’ve pulled a nondescript halogen bulb out of a light fixture. I have no idea what power it is, but let’s say roughly 40 watts.
This bulb is not mounted in a lamp, so I’m going to use the unit’s own power switch. Since the switch is wired after the EMI filter, once the power cord is plugged into the mains, a little current passes through the filter and the bulb, making it glow orange.
Let’s turn it on and have a look at how the input voltage and two DC outputs, 5.1 and 12 volt rails, will behave. (The nominal voltage on the “5 V” output is actually 5.1 volts, in common with many other DEC power supplies.)
The input voltage eventually rises high enough to allow DC to turn on, after which it promptly collapses due to the increased load. AC then keeps oscillating between 75 and 85 volts with a period of around 4 seconds. I wasn’t sure how healthy this was for the power supply, though, so I didn’t leave it running for too long. The caps have survived and hopefully have used this opportunity to reform a little more.
The former owner of this machine seemed quite keen to see it up and running again, so I updated him on my progress. He said they called the lamp in series trick the “dim bulb tester”. I’d never heard of that and enjoyed the somewhat self-deprecating pun.
06-MAR-2024
Combining 40+15 watt bulbs makes the caps charge faster, but is still not enough to prevent undervoltage and oscillations when the DC rails switch on.
I've pulled out another halogen bulb from the same fixture and put the two of them together. At 80 watts total dim bulb power, the 85 VAC threshold gets negotiated successfully. About 3 seconds later, the PSU detects the lack of tacho signals from the fans, shuts down DC, and illuminates the red ‘fan failure’ indicator just under the power switch. This reduces the load, and the input voltage continues to climb slowly until settling on 165 VAC.
With all three bulbs attached, the AC input gradually reaches 181 volts, and then, some time later, jumps all the way up to 215 V, as if the unit had decided to give up and turned off all internal circuits. This is accompanied by some suspicious random ticking sounds similar to those of electric discharge, and I turn the power off.
10-MAR-2024
Finally, I decided to connect the system fans up and see what happens under this light load. I slid the power supply half way into the chassis so that the fan assembly plug could reach its socket on the bottom of the unit. The camera was positioned to look inside the exposed half of the PSU.
The fans spun up! But, as expected, the bulbs’ impedance was too high even for this load and the supply went into the familiar under-voltage oscillations. After this quick test I call it a day.
On that note, I consider the dim bulb testing phase complete. Out of an abundance of caution, I have decided to borrow a 110 volt transformer and run the system off it whilst I investigate those ticking sounds that occur above 200 volts.
