Since I made my posts about the PowerBook 100 that has “become” mine, I faced a dilemma as to what to do next. Even though I didn’t get too many people chipping in their opinion, it seemed that a full restoration was the route favoured by those who did decide to drop me a line, either directly or via comments.
Because such ventures can take quite a bit of time and become expensive quickly, I’ve decided to try and do things in stages, so that if anything too bad were to happen, hopefully I wouldn’t have already expended too much of my time and effort.
A logical place to start was with the power supply, one of which I already tore down in the previous post and made a capacitor map for. Now it’s time to repair it.
The bad parts were previously determined to be all electrolytic capacitors which have leaked. This is not uncommon for equipment of this vintage, as the electrolytic capacitors have an electrolyte which is chemically active and likes to eventually find its way out or dry out.
As a result, we have to obtain some replacement parts. I decided to go with my local electronics component supplier, element14. Unfortunately, their range of lines locally stocked for next day delivery were limited, so I ended up with some capacitors that had higher voltage ratings (and hence, were physically bigger).
While ideally I would have wanted to replace the capacitors with solid electrolytic capacitors which have a better reputation, unfortunately, none of them were locally stocked. Neither were solid tantalum. I didn’t want to wait a week or order overseas with a larger minimum order value, so I decided to continue with better quality electrolytics – if they lasted another 20 years, that would probably be enough for me anyway.
In order to ensure quality, I decided to stick with 105 degrees C rated capacitors with a 10,000 hour lifetime or thereabouts from Japanese manufacturers. This resulted in mostly Panasonic EE and FR series capacitors, and Rubycon YXM and LLE series for smaller values.
Unit 1: On the Operating Table
I started with the unit I had already opened – namely, the same unit I had cut the power lead of and had practically no output. The first step was to unmount all of the capacitors, so out comes the soldering iron and some desoldering braid. Because of the design of the supply, it was necessary to desolder the eight pins to the vertical module to gain access to the capacitor buried inside.
Because all of the electrolyte had leaked around the board, I went over the enclosure and board with a tissue soaked in high purity ethanol, and then mopped up the mess in the holes with cotton buds. The disassembled supply with the capacitors removed and the surfaces cleaned looks like this:
I cleaned up most of the holes on the underside for accepting the new capacitors – I missed two which I came back for later.
There is some brown residue from the no-clean flux from my desoldering braid. It’s cosmetically unsightly, but otherwise, is not a major issue. A potential issue is the frosty looking partly-corroded joints and traces in the top left of the main PCB.
This is where I made a mistake. I decided to try to touch-up those joints, but because of the chemical residue, it decided to oxidise and potentially corrode the pad. When the fluid remnants are exposed to heat, they form this black tarry corrosion-like mess – see the pin on the module on the left. Luckily, I only touched two pads, which I managed to bridge with solder to ensure continuity.
Because of the higher voltage rating of the capacitors that replaced the originals, it was a tight fit requiring some bending of the caps to fit. If you can, matching the voltage rating and the diameter of the capacitors would be a good idea. The result is not pretty but it should be functional.
The soldering job isn’t my best, but it’s not particularly bad either.
To make up for cutting the cable, I decided to join the cable together and wrap it in heatshrink. It’s not as nice as a clean original, but it’s at least functional.
Unit 2: On the Operating Table
Because I had gained experience from working with the first unit, the refurbishing of the second unit went more smoothly and quickly.
To my surprise, cracking open the second unit starting at the cable grommet as I had advised in an earlier post resulted in a clean opening without major damage. Inside, this one was quite clean without the film of capacitor electrolyte over the casing.
This one had a much closer voltage output, a little high, so was it unscathed? Of course not. A look at the screw that secures the PCB to the case shows that it’s begun its corrosion process – a sure sign that something is leaking.
I suppose this should serve as a reminder that visual inspection isn’t enough. Looking at this, you might think everything is fine.
Turn it upside down and everything still looks fine.
Unmount the capacitors and we can see capacitors in various stages of leakiness – ranging from none at all, to “just about to unleash a tidal wave onto the board”. I wonder if the rubber bung itself decomposes, just like the rubber used in the feet of the PowerBook.
Because the capacitors were fractionally raised above the board, the leaking electrolyte didn’t really affect the board. That makes this one a really just-in-time catch.
Because of its better condition, it was also easier to clean the holes to get the capacitors in.
Which leads to a decent soldering job once everything goes back in.
The joints are as good as factory-made, at least, in my opinion. Then again, I always think what I do is good …
So why didn’t this unit leak as badly as the other unit? Maybe it’s a sign that this was the secondary unit which wasn’t used as frequently, and thus, its capacitors didn’t degrade so much due to heating or electrolyte breakdown. Other than that, there’s also component-to-component variation which means that even in a whole batch of units, they won’t all fail at the same time from the same causes.
How Bad Were the Caps?
Often, I might see older equipment and just run it anyway thinking just how bad can it be? When it comes to capacitors, I had no idea just how a slightly leaky cap would perform compared to a fresh capacitor. To quench my thirst for knowledge, as I had an Agilent U1733 LCR meter handy, I decided to test every capacitor that went in and out. Measurements were made at a frequency of 100Hz for simplicity, most appropriate when considering filtration of mains voltage in 50Hz countries, although the secondary capacitors are probably operated at much higher frequencies as it smooths the switching converter output.
Unit 1 capacitors are top, Unit 2 are on the bottom, with the left section showing the removed original capacitor and the right showing the replaced new capacitor value. A slight difference in capacitors originally used in the two units is noted.
Important values to pay attention to include the actual capacitance in series mode (Cs) for larger (uF) sized capacitors. I’ve created a column for “error” that describes the error in the value – normally we expect the value to be within 20%. Many of the leaky capacitors show a marked change of >90% loss of capacitance.
Another value that needs attention is the Dissipation Factor (DF). This is related to the Equivalent Series Resistance (ESR) but basically it describes how non-ideal the capacitor is – a higher DF indicates energy lost in the capacitor, as does higher ESR. Capacitors with high ESR generally don’t perform well, although acceptable ESR values are dependent on the size of the capacitor to some extent. DF values greater than 0.05 are highly suspect, and ESR values over 100 ohms are definitely problematic. From this, the second supply had some “partially” failed capacitors that had adequate capacitance but were beginning to increase in ESR/DF.
Of course, all the new capacitors were good, as you’d expect. Interestingly, the “R” branded small capacitor seemed just fine, as does the primary Elna RE2 capacitors. The failures mostly were Elna Longlife RSH series capacitors, although the Nippon Chemi-Con SXF also failed and the Panasonic SU was marginal in the second supply.
Testing and Alignment
Right off the bat, I decided to carefully and cautiously raise Unit 1 on my variac to see its behaviour. It came back to life with a 7.662v stable output and roughly 0.8W idle consumption at 120V, increasing to 1.5W at 230V. This seemed reasonable and was within range, so I didn’t do anything except plug it into the PowerBook 100 and started it up. To my surprise, it chimed, the screen came on, and the hard drive clicked twice before spinning up to life. It seems its “peak” current supply is probably a little below where it should be, but it was still within “spec”, so I wasn’t going to tinker with it.
Unit 2 was raised in much the same way, but instead, it had a 7.942v output which was a little too high for my liking. As a result, I got my screwdriver and adjusted the VADJ potentiometer. Turning it clockwise reduces the voltage, so I lowered it to about the same voltage as the other one before plugging it in. To my delight, it worked exactly the same as the other unit. I wonder if adjusting the other potentiometer can provide slightly more current on the output, although, I really don’t want to risk blowing up semiconductors …
Unit 2 was sealed up with super-glue on the internal lip, and is cosmetically very much in good condition. Unit 1 was smothered in superglue along the edges to get it to seal due to damage to the casing edges during its teardown, but is otherwise resealed. The gluing of the enclosures is critical to safety, as it forms the functional insulation for the adapter and if it comes apart in use, contact with mains could occur.
That wasn’t particularly difficult, and is just another “routine” soldering job. I like the fact these older devices are made with such “solderable” pads and single layer boards, making servicing a breeze. In the process, I managed to quantify just how bad the leaky caps were. Now that the supplies are alive, I can turn my attention to the laptop itself when time avails.