IMHO, it;'s a very good idea not to just plug things inot the mains nsee what happens. What I do is : 1) Take ebverything apaart that it is reasonable to dismantle. Obviously I don;'t open up HDAs, But I would remvoe the logic board from the outside of a hard disk unit at htis point. Make lots of notes as to what oes where, and how everything is conencted. 2) A complete visual inspection. Look for damaged or burnt parts, odd screws, etc that are waitring to find somethign to short out, things like that. Check that any fuses are the correct rating, and are not wrapped in aluminium foil or similar (!). 3) Cheack the mains earthing (preferebly at a fairly high test current) and, if you have an insulation tester, check the earth leakage. 4) The bit I get flaemd for soemtimes. Connect a summy load to the PSU (car bulbs -- 6V and 12V -- are useful for this) and run the power supply on its onw. Check the output voltages, Check for excessinve ripple if you have a 'scope. 5) Reassamble, power up, and see what needs fixing. At this point you may have genuine faults to find, but you won't have done any more daamge to the machine. -tony

There are some who just feel it's fun to flame, I guess (or at least that's the only exxplanation I can find...). But htere have been more serious reasons for not doing it : It takes time (but not half as much time as rebuidlign a machine where the PSU has gone crazy!), you have to get some parts (the dummy load bulbs nd a multimeter, at least), and there's the possibility of damaging the machine by taking it apart (all I can say to that is that I've dismantled a couple of hundred classic computers in my time, and haven't don't permeanent damage to any of them). Probably most fo the time it _is_ safe to just power up and see what happens, just as mosto f the itme you don't need dual-circuit brakes in a car, you don't need seat belts, you don't need a tension band or an implopsion screeen on a CRT, etc. It's the other times that you're glad of them. And yes I have had PSUs that were faulty. Most recently was one of the cable PSUs in an HP11305 disk cotnroller, the 723 regualtor was defective and the output was sittign at about 12V. I don;t know what would have happened if tyhe crowbar had been connected (it's on a separate PCB, along with some of the logic), and I certainly wasn't goign to find out. Most likely it would have just been another burnt resisotr and a blown mains fuse. But it could have been worse... THe 2 reasons being (a) I haev a 'scope and (b) I know the trick anyway? Sure, I do that soemtimes, it's quite handy if you are working on a large machien and can't eaily get the 'scope to it. What you need is that the impedance of the capacitor at the mains frequency is small compared ot the input resistnace of the meter. Most DMMs are 10M Ohms (but soem are less o nthe AC ranges -- check the manual). Aanlogue meters are considerably lower. A 1uF capacitor should be fine for most cases. Even if you cna't get an accrate measure of the ripply, you cna at least see if it's higher than you might expect. One advantage the 'scope has though, is that it lets yo see the ripple frequency. On linear PSUs, ripple at twice the mains frequency is likely to be dried-up smoothing capacitors (jsut about all PSUs have full-wave rectifiers). Ripple at the mains frequency often means a defective diode in the rectifier. For SMPUSs, ripple at twice mains frequency leads you to the smoothing capcitors o nthe mians side. Ripplye at the stiching frequency (10's of kHz), particulalt sharp spikes, leads you to the output capacitors. -tony

True. However, the impedance of a capacitor decreases with freqeucny, so if it's low enough at a given freqeucy, it'll be even better at higher frequencies. The lowest freqeucny ripple you are going to get from any normal PSU is at the mains freqeuncy. A linear PSU may well have ripple at twice the mains freqeucny, an SMPSU at the switching frequency or multiples thereof. So having a low enough impedance at mains frequency should be enough. I would be _very_ suprised if any SMPU was deisgned to switch at a frequency lower than mains. One advantage of SMPSus is that, since they run at higherr freqeuencies, you can use smaller transformers and smoothing capacitors. Adding complexxity (by making it a switcher) and then having to use a larger transformer and smoothing capacitors than if it was a simple mains transformer-based supply (possibly with swithcing-type voltage regualtors to impove the efficiency over a purely linear supply) would make no sense. [There are, of course, choppers that run at lest then mains freqeuncy to generate lower-frequency _AC_ outputs from the supply, perhaps to run an indcution motor at a lower speed, or for telephone ringing, But that's not applicable here] IIIRC, power line losses increas with frequency, so this would be unusual to say the least. At one time various areas of England had their own power stations, not linked by the national grid. And they were most certainly nor all 50Hz. I think the highest was 100Hz, though, the last place to have that was Lynton/Lynmouth in North Devon. I would think making an adjustable (active) filter would eb simpler here. But if you are serious enough about electroncis to want to do soemthing like that, you are also serious enough to want to get a 'scope I think. -tony

If an SMPSU ithat should be running at several 10's of kHz is actually running aobut 1000 times slower, you would know about it. By the chopper transisotrs hittlign the ceiling for a start. The transformer would certianly saturate, the drive currents would go sky-high. OK, you might be right. It's _possible_ there's one (I certainly don't know everythign, and it's entirely possible somebody di do this for a special reason that none of us have thought of). However, 1 uF capcitror has reactance of about 160k Ohms at 1Hz (Xc = 1/(2*\pi*f*C). Use that as a blocking capacitor for a 10M Ohm input impedance DVM, and you would still detect the ripple at close ot the correct level. and if the ripple freqeuncy isd slower than 1Hz, you could see the needle flickering on an analogue voltmeter ;-). I don;t think any supply would get a slow as that. And in any case, we've put a dummy load on it, right? Aircraft supplies are often 400Hz since that leads to smaller/ligher power trnasformers running direectly off the supply, as was done before the common use of SMPSUs. Sure. What I would do in my workshop (where I have decent test gear) is differnet to what I'd do at a remote site (given only the instrueltns I'd brought with me). Whether I would build an active filte would also depend on waht components I had available (which depends on whether there would be time to do an order to Farnell, say, whether there's a local electroncis shop (or company stores) that at least has common R's and C's and op-amps, or whether I have to manage with what I've got or could 'borrow' fro mother parts of the device). However, I think the OP was planning on doing this repair at home, and that time weas ntot too critical. In which case I think my comment applies that while you can produce workarounds (and while it's certainly good to understnad enough to think up said methods), a 'scope would probalhy be a sensible thing to obtain if you anre going to take electronics that seriously. -tony

With some SMPSUs you get a burst of oscillations driving the chopper (at the normal switching freqeuncy) for each click. In many cases, the reason why the PSU shuts down with no load is that you get a spike on the output which causes the crowbar to fire, effectively short-circuitng the supply. It then shuts down and tries again. _Never_ disable the crowbar or overcurrent protection circuits in any PSU unless you really know what you are doing. The results caan be exxpensive an unpleasat. I rememrbr disabling the overcurrent protection in a DEC H754 -15V regulator brick (which is a switching regualtor, of course), only for find that the crowbar was firing. WIthout the overcurrent proteciton, I ended up with most of the transsitors dead (several of them blown off the board), resistors burnt out, the 723 useless, and so on. Yes, I did get it going again... -tony

On 12/20/11 4:53 PM, "Christian Kennedy" <chris at mainecoon.com> wrote: Really?!? That's very cool. Was there enough Mac software in the world at the time to fill the drive? :-)