Repairing SMPSs

Discussion in 'Electronics & Electrics' started by Thalyn, Jul 23, 2018.

  1. Thalyn

    Thalyn Member

    Joined:
    Jun 2, 2005
    Messages:
    429
    As I'm having no luck at all with searching myself (maybe I'm using the wrong terms, but everyone seems more interested in going into depth about how inductors work rather than repairing the PSU itself) I'm hoping someone here can offer me some insight or point me in the right direction.

    Basically, my brother and I are repairing old computers as something of a hobby. He intends to use them, I just like fixing them. We've run across two PSUs, however, which neither one of us knows enough about to be able to fix: one is from an Apple IIe (an Astec +AA11040-B "gold box" unit) and the other from a Sega Saturn ("long" 5-pin unit), representing basically opposite ends of the spectrum for flyback SMPS units.

    The Apple PSU is completely non-functional. Plug it in, turn it on and there isn't even a blip of output on any of the DC lines. We've replaced all of the electrolytic capacitors (often cited as the first step, since they're 35+ years old), the DC rectifier diodes, and the X-type polypropylene cap which looked to have split its case on both sides. We also pulled every transistor and hooked them to one of those $20 transistor testers and it identified them all correctly (which at least means they're not completely dead), and checked what I believe to be the start-up resistors (R2 and R3, both 100Ko) out-of-circuit.

    This unit hasn't worked as long as we've had it (since the mid 90s). The previous owner seems to have abused the system (most the screws were missing, many of the ICs had rusted legs, and the PSU fuse had been replaced with aluminium foil) so we can't be entirely sure that something nasty hasn't happened to it along the way. But "everyone" (search results) seems to insist it should be fixable and that instructions exist on how to do it - just that I've not been able to find any.

    The Sega PSU, on the other hand, has only recently failed. In isolation (so the console is not at fault) it will run perfectly for around 15 seconds before the output voltage on all three rails rises and the unit cuts out with the power light blinking. When the outputs rise the feedback from the DC-side shunt regulator falls, which we assume is causing the rise (we've replaced the regulator without success). Once it does this, the unit has to be left for around 2 minutes to "reset" before the process can be repeated.

    Unlike the Apple, my brother has owned this since new in the mid-90s (just a coincidence), so we know its usage and service history (light-to-moderate and none respectively). It seems to have had a slow failure over the past few months where it once cut out after an hour or so, before it would cut out during a game intro (Torico for those playing at home) and now only lasts for the start-up period on the TOP202 control chip. As such there aren't any obviously damaged components like you'd expect from sudden failure, so we need a better means of diagnostics.

    If anyone can offer some advice on how one (or both) of these units might be repaired or at least diagnosed, I'm all ears. And if you can point me to at least a moderately comprehensive guide to repairing SMPSs, which doesn't get bogged down on inductors, I'd also be appreciative as these will doubtless become more common as we progress to repairing 32-bit systems.

    Cheers.
     
  2. Dice

    Dice Member

    Joined:
    Feb 18, 2002
    Messages:
    164
    Location:
    Altona
    Mate of mine repaired the same Apple power supply a few months back - he found the transformer had an open circuit on the primary side.
    Luckily, its a primary centre tapped transformer so he was able to use the other tapping. Worth a shot.

    As far as the sega - have never looked inside, but would imagine its either a capacitor, dry joint, or temperature issue? Can only suggest freeze spray from jaycar or similar
     
  3. OP
    OP
    Thalyn

    Thalyn Member

    Joined:
    Jun 2, 2005
    Messages:
    429
    I appreciate the reply.

    Yesterday I was actually able to isolate and meter the transformer. Found a diagram in Sam's guide showing expected pinouts and resistances for the predecessor, which is basically the same save the feedback regulator (used to be two op-amps, now just one IC). I don't have access to a milliohmeter but all the pins which looked like they should be connected were, the ones that shouldn't be connected weren't and the resistances were "fairly close" (at least measuring tenths of an ohm at 1x precision). Not a fun job trying to put it back into the board, though, but at least that's somewhat comforting to know.

    Will have to investigate those options on the Saturn. I've gone across the joints previously and re-touched them but I could have missed one (or just done it badly), and the caps are a definite possibility (my brother hates that suggestion, but seems to be warming to it now that caps fixed his Game Gear). I would imagine if it's temperature related than a sacrificial finger could probably ID the culprit pretty quickly, since it cuts out after only a few seconds (something overheating that quick should be noticeable, right?) so I'll check into that as well next time I get the chance.
     
  4. Technics

    Technics Member

    Joined:
    Apr 29, 2002
    Messages:
    1,737
    Location:
    Brisbane, AU
    It's quite unusual for the transformer to open up in an SMPS from my experience. The cheap, shitty caps used in most consumer electronics for the last couple of decades is another matter. The slow failure you report for the Sega does go some way to backing your suspicion there.

    The likely reason you can't find a generic guide is probably because there are so many SMPS topologies and variations on them. People often refer to anything with a trafo as a flyback when it may be a forward converter or other design. Then the individual implementations also vary so much that it would be hard to describe one procedure. The other thing that makes it difficult is the same thing that makes diagnostics of any closed loop system hard. A failure anywhere affects the operation of the entire system. The final reason is that most people don't have proper isolation transformers and aren't familiar with safely working on the equipment so people are reluctant to encourage attempting these repairs.

    That said they do tend to all look pretty similar when you get down to it. A data sheet for the controller chip used is often a pretty good starting point.

    For the Apple PSU. The first place to start would be the "bootstrap" for the switching controller. The controller chips typically need to be powered up before the switching element can supply pulses to the the primary winding. This is often achieved using a resistor and a cap. Once the supply is running the controller may then be powered from its own auxillary winding. There are some other methods employed for bootstrap but they seem to be less common. If that is fine then look at the switching element (MOSFET or BJT on some older designs) or the controller and surrounding components. With no signs of life at all it's less likely to be a feedback problem but if all else fails you will need to look there too. Often they use optoisolators for the feedback though sometimes a winding on the main or a separate transformer is used.
     
  5. OP
    OP
    Thalyn

    Thalyn Member

    Joined:
    Jun 2, 2005
    Messages:
    429
    Definitely leaning towards caps on the Sega. I'll have to try and convince my brother again, or just test them while he's busy with something else. I've only got one of those ATMega-based testers for ESR (or capacitance over 200uF), but at least it's better than no idea at all.

    On the Apple... there actually isn't any kind of switching controller; at least not an individual IC. The design is from 1982 and the only IC on the board is used for the reference voltage - even the feedback loop is by way of a small, secondary transformer instead of an opto! To be honest, I'm not even entirely sure where the bootstrap on the thing is, though I suspect it may be R2 (a 150k which I tested at ~100k) and C24 (47uF, which has been replaced); though C7 (220uF, also replaced) is often cited as a reason these things fail, which is in series with R3 (150K, tested as ~100K) leading back to AC neutral, and R4 (27, untested) before hitting the transformer, so perhaps there's something there to be investigated. C24 and R4 are also between the oscillator transistor (Q2) and the transformer.

    BTW, it turns out I was mistaken on the model number. It's not an AA11040-B - it's an AA11042-C.
     
  6. Technics

    Technics Member

    Joined:
    Apr 29, 2002
    Messages:
    1,737
    Location:
    Brisbane, AU
    I did find a schematic for the B version but they are pretty similar apparently. It's quite hard to follow because it doesnt show the direction of any of the windings on the feedback trafo. You are looking in the correct area though. The four 47uF caps are the main primary side DC bulk capacitance. It appears that it is the charging of C7 via R3 and R4 that turns Q2 on initially. Note that a voltage divider is formed with R5 so it needs to be okay as well. The feedback transformer is also part of that circuit. It won't be doing anything initially but an open winding would prevent bootstrapping. It would be difficult to verify the overall circuit operation without a scope and isolation transformer. If say D1 or C26 was to fail short it would never charge so it could be these or other surrounding components as well the diode bridge DB1, Q1 and Q2 being obviously critical. A lot of the other parts on the primary side are snubbing and clamping. Without being able to use a scope you will just need to test each component in this section. A power supply that bootstraps properly but then falls over often makes a ticking noise so it's not an unreasonable assumption that it isn't getting that far unless proven otherwise. I'm assuming the fuse and switch have been checked too.
     
  7. Technics

    Technics Member

    Joined:
    Apr 29, 2002
    Messages:
    1,737
    Location:
    Brisbane, AU
    I also forgot to mention R1 which is a NTC thermistor used for inrush current limiting. It initially acts as a resistor to limit the current that charges the caps but its resistance drops as it heats up and reaches an equilibrium. They can fail from the thermal cycling/stress. Once Q2 switches on the current in the main trafo primary winding will rise until Q1 is switched on and cuts off Q2. The feedback trafo should start to affect the amount of peak current in the main trafo primary winding before Q2 switches off once the supply gets past this phase of start up.
     
  8. OP
    OP
    Thalyn

    Thalyn Member

    Joined:
    Jun 2, 2005
    Messages:
    429
    I appreciate it, Technics. I've taken a note of everything you've mentioned and next chance I get I'll start removing and testing components. I'd like to get a scope on it, too, but have no isolation transformer and my portable scope likely isn't sufficient (it's a Uni-T UT81B, so could handle the frequency but not really fast enough updating).

    We did check the switch early on in the piece (always start with the obvious, right?), and the fuse is definitely new. To give an even further indication of the abuse this machine suffered before we got it, it originally had a piece of alfoil instead of a fuse... comforting, right?
     

Share This Page

Advertisement: