Archive
Bad Capacitors Strike Again
I’m working on another Bose 901 Series 1 equalizer. It looks to be the same manufacturing run as the first one I wrote up last July, uses the same components, and those components look like they’ve had the same failure mode.
The rather dubious EM-branded capacitors have the same discoloration on the top middle I saw on the other Bose, as well as on the Farnsworth K-262P which had been repaired once before. I’m curious what the physical failure is, as the discoloration would suggest to me overheating, but that seems very unlikely in this circuit and application. Newly manufactured replacements aren’t likely to have this problem as the film technology has improved significantly in the last 40 years.
Westinghouse LCD Repair
After seeing the repair I made on the Samsung LCD monitor, a friend gave me a few-years-old Westinghouse LCD/TV that had quit working – it wouldn’t power on anymore. It’s a Westinghouse SK-19H210S, 19″ LCD accepting VGA or HDMI up to 1440×900 resolution (somewhat smaller than true 1080P) and can also tune ATSC and NTSC television signals to receive HDTV over the air.
It’s apparently a very known fact this one has a weak power supply – all over the web. I opened it up and grabbed the power board:
Tucked away all in the back is one capacitor that’s visibly failed, which means it’s likely several are bad or will be soon.
New parts arrived from Mouser.com:
Using my trusty Hakko, I replaced six capacitors. 4 caps in total showed signs of leaking from the bottom as well (discolored board below), 2 seemed okay but I replaced anyway because why not. I’m getting better at using the Hakko and doing this kind of PCB rework in general, the entire process from start to finish only took about 15 minutes this time.
100uF 400V
2200uF 10V
1000uF 10V
1000uF 25V x 2
47uF 50V
Interestingly (or maybe not), these bad caps were the same brand as the bad caps from the Samsung: CapXon. Obviously those have reliability problems, or are just the cheapest they could buy.
Reassembled and powered on. The first power-up would come online but drop off immediately and it was making a hissing noise; it turns out I hadn’t firmly connected the backlight leads. After fixing that, I snapped everything back into place. Consumer electronics these days aren’t made to be opened up, so the case doesn’t quite fit back together the way I’d like it to around the control panel on the side, but it’s not visible unless you look for it fortunately.
Another one fixed! This one was about $12 of parts. Looks like this one goes for around $80 these days, so I’m half-way to getting my money’s worth out of that rework station already.
My next TV repair will be somewhat more ambitious. I got this Samsung HL-P4663W, a 46″ DLP (720p) HDTV for free from Craigslist. It needs a new bulb, and some other rework, and it’ll be worth a few hundred dollars after I get it sorted. I don’t intend to keep this one (as I already have a 46″ Samsung LCD that does full HD resolution) but just to repair and sell most likely.
Adventures in Dumpster Diving: Samsung 225BW LCD Repair
I found a Samsung 225BW LCD sitting on top of my apartment’s dumpster, and figured I’d drag it upstairs. It’s a few year old model but it’s better than the current older Dell LCD that I’ve been using (1680×1050 versus 1440×900). A quick check showed that it would power on, sort-of, but the power light would flicker constantly and there was no backlight.
I popped it open, suspecting a problem in the power supply – and turns out that was right. Several capacitors on the board were showing signs of failure. Capacitors are the main component I replace in the vintage radios but cost-cutting OEMs are often known to use caps that fail after only a few years when new to save a few cents on each part that goes out the door on new things as well. In this case their 330uF and 820uF @ 25V caps had failed and the logic board was no longer getting good power.
Modern electrolytic caps fail by bulging and leaking out the top and/or the bottom, it’s easy to see at a glance. The top two are bulging and leaking; the bottom ones are bulging only which is a bit difficult to make out in the photo.
This project is one of the reasons I bought a Hakko 472D desoldering tool. It’s made for reworking through-hole and point-to-point boards, and works by melting the solder and then applying a strong vacuum through the center of the nozzle sucking it out of the way and cleaning the connection. It wasn’t cheap, but I thought it’d be important to have one of these as I do more types of electronics hobby work. I tested it out on an antique radio and it works perfectly for the annoying old joints.
This board is pretty easy to work on, the components are widely spaced and marked.
Even though it’s not bad, I’m replacing the large main filter as well – just in case. It’s the same brand as the failed ones.
Here I’ve depopulated the bad components from the board and have placed the main filter back in position, with the old one above it for comparison.
The new caps are larger than the old ones – for the same ratings, a larger size capacitor is going to be a bit more durable. For example these 330uF 25V models:
Slid the components through the top, spread the leads to hold them in position while soldering and reattaching:
Bad planning on my part meant I forgot to take a photo of the board post-repair, but it only took about 30 minutes to do the entire thing – most of which was spent figuring out how to adjust the Hakko. And for the power-up:
Success! Back to life. This LCD goes for around $150 online even today and I’ve been meaning to add a second monitor to my desk anyway, so I’m about 1/3 of the way to recovering the cost of that desoldering station after the first project. One down, two to go….This project required 3x330uF 25V capacitors, 2x820uF 25V capacitors and 1x150uF 450V capacitor which came out to $9.83.
Cost Engineering in 2011
While the power transformer in my previous post may have lasted for years, it ultimately still succumbed to poor design and the transformer is widely regarded as a weak point in Zenith sets.
Cost engineering to pick the exact right time of failure has gotten considerably more precise since then. The warranty on the array controller in my enterprise backup server expired on Jan. 2, 2011 – just over two weeks ago. Today, a surface mount capacitor decided it had enough, I’m impressed with how close to the end of the warranty it lived.
My last computer was dead-on perfect, though. A voltage regulator at the base of the board decided to end its life in a spectacular fashion while playing an online game, releasing the smell of ozone into my apartment and leaving me staring at a black screen for a minute wondering if my computer really did just explode or I was imagining it.
The time of death? One day outside the warranty period.
Cost Engineering in the 1930s
It looks like cost engineering was already happening back in the 1930s:
In the foreground, the destroyed original power transformer from the 1939 Zenith 7-S-363 radio; the rear, an equivalently rated replacement built in October 2010 by Edcor Transformer. In order to squeeze out a few cents cost savings per chassis, Zenith under-rated their power transformers and ran them right at the limit. In certain modes of tube failure, or just age and components wearing out, the power transformer’s razor-thin safety margin would be exceeded resulting in overheating, shorting and melting. Which is exactly what’s happened on the one I’m repairing at some point in its life.
Burned Up Aiphone
This is what happens when you connect the high level output of an audio amplifier, to an input designed to be hooked up to an MP3 player or other line level audio source. This Aiphone PA amplifier was incorrectly installed by a technician and it made this 1/2W resistor dissipate about 40W. It held up…briefly…but quickly turned into a charred mess. It was somehow managing to pass a terribly distorted signal but didn’t damage any other components. After replacing the resistor on this pictured unit, and another identical one damaged the same way, it was back in service.
It’s not always overly aggressive cost engineering that makes things fry. Sometimes it is actually operator error.
Share this:
Like this: