Electronics > Projects, Designs, and Technical Stuff
ATX PC power supply integrity check... with a very bad choice of tools...
starhawk:
Back in 2014, I put together a PC for my father for Christmas. He is not technically inclined at all any more, essentially having left that part of himself behind with the 1980s when they ended. Recently he contacted me with the information that the rear fan in the power supply of that computer had begun to make considerable noise. This is bad, because he lives 5+hrs from me (each way, and I don't drive) and he has absolutely no suitable backup machine -- and I don't have the financial ability right now to provide him with one, or to replace this system entirely, should it fail unrecoverably.
The supply is a Logisys PS550A-BK "500w" unit. Note quotes. This is a relatively cheap supply -- hey, it's what I could get; the entire build of that machine was a roughly year-long exercise in creative scrounging (etc) to get the absolute best PC I could out of the absolute minimum cash outflow -- and although the parts powered by it will never reach even a third of the supply's ratings (they pull a little under 150w by my estimation), a potentially overheating supply is indeed a cause for grave concern -- inexpensive PC power supplies are extremely well known in the computer enthusiast community (and most of the surrounding population) for going out in remarkably spectacular fashion, such that typically other major components in the computer are destroyed in the process -- in fact, it is not unheard of for the entire box to have to be thrown out and a clean start made, because the supply's death throws were so incredibly severe that they torched absolutely everything inside.
I want to run three tests on that supply. One is a simple voltage-sag test, to see how it's performing under load. I have a Radio Shack autoranging multimeter for that, which I got as a surprise gift from an Internet acquaintance years ago, with whom I lost touch when I left the forum we were on. The next is a visual inspection to check for potentially-failing capacitors (etc) and any other components which look like they may have been chronically overburdened. The last is a ripple and noise check, and that's where the title of this thread comes into proper effect...
The tool for that, as I'm sure everyone here knows, is an oscilloscope. I have one, but it's, er, a Tektronix 422 [sadtrombone.mp3], which for the unfamiliar is a 15MHz-bandwidth 'scope that is literally so old it predates the commonplace use of integrated circuit technology... no, really, it's discrete semiconductors and passives throughout with a very few tubes thrown in for good measure. At least it was new enough that everything's on PCBs rather than tagboard!
The specific example 'scope in question was a treasured gift from another friend with whom I've sadly lost touch (Ray, if you're still out there and you see this... shoot me an email, I want to know how you're doing). He was an engineer in Burroughs' computer division in the 1970s, he told me, and this was his work 'scope, with which he worked out the design of a series of (to me, anyhow) truly impressive electronic masterpieces with the approximate size and form of an executive desk and the kind of sheer computational horsepower that would probably, honestly, seem pretty sad next to even the TI-83+ that put me through high school. The power cord is beginning to crack, and has been missing the ground prong on the wall outlet side (!) for, well, longer than I've had it... I have the print manual, complete with its front and back covers in shiny blue leather-texture-printed cardstock and a plastic comb binding, but this manual identifies itself as being for the 20000+ Serial Number 422s, and based on the labeling on my particular 422, that might not be the correct manual...
I have attached photos of the oscilloscope face (with the serial info and all the controls), and of the cabling and cable attachments I have, to this post.
What I need to know is, having watched precisely two ~30min YouTube videos (one of which was indeed an EEVBlog vid on power supply testing, but it was more oriented towards bench supplies and very modern testing equipment) on 'scope operation, and never really having read the manual -- what is the proper way to operate this 'scope to see what I need to see? The closer you can get me to a set of directions a la Mapquest, the happier I'll be -- I am very much an inexperienced dunce with this particular device, and I need to get this one right. Yes, I know that this is a less-than-ideal 'scope for the task at hand -- please be aware, however, that it is not possible for me, at this time, to get *anything* better or different, both because of cashflow and because of time constraints.
Dad is picking me up Saturday for about a week's visit because Thanksgiving... I'll be taking the o-scope and
multimeter along, plus what I consider a basic tool kit (soldering supplies, spare screws, clippers/pliers, and a PC bit-driver set) along with a replacement supply (musty-smelling P4-era Dell 305w...hey it's known good) and suitable replacement fan, both out of my junkpile... I hope by the time I get there, you folks will have pitched in and told me what I need to know to do this ;)
starhawk:
Aw c'mon guys... I know, words, but I'm friggin tryin to help my own Dad here. If I'd been asked to help you work out how to do something techy for /your/ computer-challenged father, the very least I'd do is make sure I understood what you were asking me, which would honestly probably involve a lot more reading than this.
andy3055:
In the short time you are there, what is the chance you will be able to fix it for good? If you need a part, thee is no place to get it especially considering the holidays. Why not order a supply and get it shipped to your dad's place? You can get one for less than $30.00
https://www.ebay.com/itm/Logisys-550W-Max-Power-Supply-Unit-PS550A-BK/193053563566?_trkparms=ispr%3D1&hash=item2cf2e39aae:g:H4MAAOSwLl9dWDdr&enc=AQAEAAACQBPxNw%2BVj6nta7CKEs3N0qU3TpoWB59ghkMNebZ0K2OGhW%2B7MMs8o%2FJgte4CU4eAdGHXsvRgDsuMEMLkzgLn%2BLhxZzpr9qMLZrQyIZ3D0t8QQidyJNdNdI%2F8sCCu0vJ3EDwRChTtuEV8WMnAqt6s3uEKn%2B8Xy%2BxZKkOxJtzRmPH55HqfiKn4TPrN9poU00KSCKqt1h7nfoA%2BU9Cdw1p21I2xsNsp1ms5rtTIEezum6lwd099sMuJZUQI2CUhY%2BPz8IaqX30WRKWRqqvRl3G3VwtpcF4kkM2lNB9NA36c2%2B%2FBPS7gAhpNknlv3yJUDwnba1ZniQi2hedoNFlYwuck1AbxV0GsDzfJS60HQzNqP8F0XtzGCG2EO5UddOXlstlfQQ4fddbA0zObQPEpTIwX6XQGrNAgoKmzc%2FZe4hbnb8gX0i4yjySHdf0hgsH2Uzjg5gb3cRv5aSONqLs%2FmPLjOmyAjI7cmJKngwF01%2B1LlcL7hkardfZ404m5CX5wFX0Sv7j8dyUw43yIWurKL4VqdAiMjL3oN3MgQHYzq8AtVfhPwFFkAFpFU3Y2rx1ne8Vfd0S0dTEnmcoOhZpNneI4Xn3shBJq8%2BIeh4TKahdPyJa802PFzgfxytokJNTPJfACWIEI%2FQtLf6KW8vGemb9BROrQ%2F%2BIpEDaFwX3u%2F4GLx5hZAD9Fk11%2BlOc%2Fd2e2H%2BYKKPBDzqW6w201x6eCHXUJm%2FrzJzQRHkbVHGlaZ7q8drUCnJurlB5c3WJ8%2B8lYWfYXxg%3D%3D&checksum=19305356356644e265df58594fa3993f2a6dc34a4e69&enc=AQAEAAACQBPxNw%2BVj6nta7CKEs3N0qU3TpoWB59ghkMNebZ0K2OGhW%2B7MMs8o%2FJgte4CU4eAdGHXsvRgDsuMEMLkzgLn%2BLhxZzpr9qMLZrQyIZ3D0t8QQidyJNdNdI%2F8sCCu0vJ3EDwRChTtuEV8WMnAqt6s3uEKn%2B8Xy%2BxZKkOxJtzRmPH55HqfiKn4TPrN9poU00KSCKqt1h7nfoA%2BU9Cdw1p21I2xsNsp1ms5rtTIEezum6lwd099sMuJZUQI2CUhY%2BPz8IaqX30WRKWRqqvRl3G3VwtpcF4kkM2lNB9NA36c2%2B%2FBPS7gAhpNknlv3yJUDwnba1ZniQi2hedoNFlYwuck1AbxV0GsDzfJS60HQzNqP8F0XtzGCG2EO5UddOXlstlfQQ4fddbA0zObQPEpTIwX6XQGrNAgoKmzc%2FZe4hbnb8gX0i4yjySHdf0hgsH2Uzjg5gb3cRv5aSONqLs%2FmPLjOmyAjI7cmJKngwF01%2B1LlcL7hkardfZ404m5CX5wFX0Sv7j8dyUw43yIWurKL4VqdAiMjL3oN3MgQHYzq8AtVfhPwFFkAFpFU3Y2rx1ne8Vfd0S0dTEnmcoOhZpNneI4Xn3shBJq8%2BIeh4TKahdPyJa802PFzgfxytokJNTPJfACWIEI%2FQtLf6KW8vGemb9BROrQ%2F%2BIpEDaFwX3u%2F4GLx5hZAD9Fk11%2BlOc%2Fd2e2H%2BYKKPBDzqW6w201x6eCHXUJm%2FrzJzQRHkbVHGlaZ7q8drUCnJurlB5c3WJ8%2B8lYWfYXxg%3D%3D&checksum=19305356356644e265df58594fa3993f2a6dc34a4e69
starhawk:
Per the first sentence of the final paragraph of my initial post... I have a musty-smelling Pentium 4-era Dell power supply which has been tested and is known good and is coming along with me in case it's needed, along with a suitable replacement fan.
I manually calculated out the power draw of the computer in question and it came to roughly 135-140w. Even allowing a considerable margin for error, we're still talking 200w or less. The supply in question, marked as a fourth-revision Dell "NPS-305BB C", should, therefore, be more than adequate to the task, despite being secondhand (pull from a friend's computer that needed rebuilding due to age) and old enough to not only have the previously-described scent but the wires are actually faded to the point that the yellow leads look off-white in places. (I've never seen that before... and it happened in a relatively light-tight Dell case, too, which I have a hard time understanding as well...)
Now... the point of this thread, as specified previously, is that I need to know how to work that o-scope to check the level of electro-schmutz coming off the old supply... could we get to that part, please...?
SeanB:
You will find the capacitors on the output side are bulging, but probably still providing some capacitance. Same for the main board ones. New power supply , or replace the capacitors, along with the fan, or simply pull the end label off, pop out the tiny little rubber bung and put in a few drops of 3 in 1 oil, or 0W50 engine oil. Place bung back, then clean with some solvent (probably thinners, as it will be a little more aggressive on the plastic) and place a bit of tape across the bung, and put fan back. If it has a speed control board or a thermistor for speed control remove the board, or short the NTC sensor ( typically a green blob on a long set of wires close to the heatsink) so the fan always runs at high speed, not at a low speed, cooking the power supply innards, but saving on noise and nothing else.
Scope easy enough to look at the 12V and 5V rails, using one of the spare Molex connectors for drives, and using AC coupling on the 1V/div range, to start with. Anything over a half volt peak to peak is a concern, but only going to be a worry if more than around 1Vpp, and will typically be just hash, probably with some sort of waveform around twice the switching frequency. Those rails are pretty noisy, but still work with that.
If replacing capacitors you replace all of those aside from the main input ones, especially small ( under 220uF) electrolytics, anywhere in that power supply, as they are notorious for drying up and being silently dead. Look at the standby supply and replace all the electrolytics there, they will be cooked. only issue is getting output capacitors that are the same tiny footprint, but extra long and actually rated to handle the current specified, as those are typically run way outside the ripple current ratings, if all are actually installed. you might have to use slightly fatter ones, spaced off the board a little, just to get them to fit.
Navigation
[0] Message Index
[#] Next page
Go to full version