Switch Mode Power Supply

Discussion in 'The Projects Forum' started by HarwoodD, Oct 17, 2012.

  1. HarwoodD

    Thread Starter New Member

    Dec 26, 2011
    7
    0
    Hey everyone,

    I’ve got an 18 bottle wine fridge that quit. I opened it up to find a SMPS with a blown inrush current limiter which I replaced. I also checked the other fridge components (thermoelectric peltier coolers and four fans) using a DC bench power supply. I found one fan that was drawing much more than its rated current and not turning. I figured this is what blew the power supply.

    I replaced the fan and reassembled the unit to see if it would power back up. Nothing. I disconnected the peltier coolers and everything comes to life. Through trial and error I’ve discovered that the power supply will not power up when the peltier coolers are connected. Once I disconnect them the power supply comes to life. Once the power supply is up and running, I can plug the coolers back in and the unit cools perfectly. The coolers seem fine, but the power supply won’t start with the normal load. It shows about 1.9 volts until the heavy load is disconnected. It also has a buzzing/sizzling sound. It’s like it just won’t start with the heavy load on it.

    I also noticed that, once up and running, one of the heat sinked power transistors and its associated 250V 330uF input cap is getting very hot. I checked both input capacitors ESR and found them to be at the datasheet max limit. Not obviously bad, but certainly not good. A capacitance check found them to also be at the lower 20% limit. I’ve ordered replacements but haven’t received them yet. Could this be causing the problem? Anything else I should look at?

    I’m somewhat new to how a SMPS works. I’d like to use this as an opportunity to learn more about them. Can anyone tell me what type of SMPS this is? I’d like to research it more. Key components consist of two power transistors controlled by a TL494 DIP IC, an isolation transformer, and a Shottky barrier diode.

    Photo attached.
    Thanks!
     
    Last edited: Oct 17, 2012
  2. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    Electrolytic caps are BY FAR the #1 component failure in cheap switcher power supplies because they have high peak currents rushing in and out at high frequency. It's also where manufacturers pinch pennies and buy cheap.

    Not sure if that's the only problem, but anytime I have an old P/S that fails, I shotgun all the switcher caps because if one failed the others will soon follow.

    AS to the cooking switch transistor: hard to say, it may have failed as well but probably not or it would not work at all.

    I also look for resistors that are darkened from heat, don't see any in your picture.

    The other parts that get stressed are the snubber R-C components, hard to pick out without a schematic.
     
  3. Dodgydave

    Distinguished Member

    Jun 22, 2012
    4,969
    744
    Tl494 is the classical Atx psu chip, pins 12 + and 7 - are the supply pins, and complimentary outputs are on pins 8 & 11

    http://www.datasheetcatalog.org/datasheet/motorola/TL494.pdf

    check if its got the supply, between 5V to 40V, looking at the pcb the relays are 12V coils, so does it give out a 12 volt a supply to the peltier devices?
     
    Last edited: Oct 17, 2012
  4. HarwoodD

    Thread Starter New Member

    Dec 26, 2011
    7
    0
    bountyhunter - I've got the rest of the caps in my supply bin. As soon as I get the lager caps in I'll change them all just in case. Also, I can't seem to find the datasheet for these caps anymore. I'd like to check the ESR again since the overheat. Any idea what value would be consided a pass/fail? They're 250V 330uF.

    Dodgydave - Yes, it is a 12V output supply. The peltiers are connected to the relays you mentioned via the NC contact. The relays are energized to remove power from the peltiers as necessary for proper cooling. The fans are wired to run all the time. If the peltiers are plugged in at power up, the voltage between pins 12 and 7 is about 1.9V. As soon as the peltiers are disconnected the voltage jumps to 12V and the rest of the unit powers up (fans, LCD, etc...). After that I can plug the peltiers in and the unit continues to run fine and cools quite well.

    I just can't get the power supply to fully power up without manually disconnecting the peltiers then plugging them back in.

    Is this a common problem with a SMPS?
     
  5. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    They MUST be switchmode caps which means they will have guaranteed ESR values stated for 50kHz or 100kHz on their data sheet. I don't know the exact ESR value, lower is always better, but you must never use standard grade aluminum electrolytics anywhere for switch caps because they will die quickly and overheat badly.

    If those are from a standard maker like Sanyo or Nichicon you can look them up.
     
  6. HarwoodD

    Thread Starter New Member

    Dec 26, 2011
    7
    0
    bountyhunter - Whoops! Clearly didn't know that. So ALL caps on the board must be low ESR style switchmode caps? My ESR meter operates at 100kHz, if I find a cap with a published ESR for that freq on the datasheet is that a switchmode cap or is there another key attribute I'm looking for?

    I keep seeing alot if info regarding ripple current, but not ESR. Are these related somehow?

    I found a "High Ripple Current" cap from DigiKey. http://nichicon-us.com/english/products/pdfs/e-gw.pdf

    Thanks for the heads up!
     
    Last edited: Oct 18, 2012
  7. Dodgydave

    Distinguished Member

    Jun 22, 2012
    4,969
    744
    Usually with this type of psu, there will be a delay to get the 12v up working, then it will power the peltiers, there must be a slave smpu on the board to give the ic its 12v.
     
  8. HarwoodD

    Thread Starter New Member

    Dec 26, 2011
    7
    0
    What is an SMPU? Sorry if that's a dumb question. Google didn't help :)

    The IC + pin trace leads back to the smaller transformer.
     
  9. RamaD

    Active Member

    Dec 4, 2009
    254
    33
    It could be that the soft start circuit is not ok. Is there a cap between Pin 14 & Pin 4 of TL494, and if so, could you please check that out?
    BTW, does the 330uF still gets hot, when everything is ok when you connect the peltier loads after sometime?
     
    Last edited: Oct 18, 2012
  10. HarwoodD

    Thread Starter New Member

    Dec 26, 2011
    7
    0
    Yes there is. It's a 50v 1uF. Checked it in circuit with my ESR meter and measured 2.4 Ohms and 0.94uF.

    Yep. I let the unit run for about 30-45 min fully powered up. Then I began hearing the buzzing noise getting louder and found the 330uF cap and the power transistor heat sink to be VERY hot.
     
  11. Dodgydave

    Distinguished Member

    Jun 22, 2012
    4,969
    744
    pin 4 is the start/ stop function, start is 0v and stop is 5v,so see if pin 4 is low at start up
     
  12. RamaD

    Active Member

    Dec 4, 2009
    254
    33
    Yes, if pin 4 is low at startup, then the softstart is not there. But it is kind of difficult to see, as it would go down from Vref to low values in a few tens of ms after power on. I cant think of anything other than a scope to check this. Maybe, add another 1uF in parallel to check the symptoms - it probably woudnt hurt to make it little more 'softer start'. But the cap value seems quite ok.

    Heatsink can get hot, maybe to a max outer limit of about 85 deg C - was the fan working blowing the heat away during then? Above 60-70 deg C is not touchable! Dont ask me how I know!

    The Cap getting hot is not ok - normally due to excessive ripple current or the cap has lost its ripple current capacity. It is better to replace the cap, as a matter of fact both caps (330uF/250V) and check it out - I am assuming there are two of them!
     
  13. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    The ones near the switching devices must be. They supply the current to them. If possible, use low ESR caps everywhere you can.

    By picture, it looks like a half bridge design with the two 330uF caps as the storage caps and the two devices on the heatsinks are the switches. Replace both 330 caps and see if the heating problem goes away. If one switch still runs hot, it could be damaged.
     
  14. Dodgydave

    Distinguished Member

    Jun 22, 2012
    4,969
    744
    The TL494 usally has a slave start up oscillator (osc), that produces its voltage, then the the ic takes over and feeds the start up osc with sync pulses to produce the desired regulated voltage,so your circuit will be similar to this design,without the op amp chip.

    Output pulses on pins 8,11 feed the slave osc transistors 2sc3306, they self oscillate at first pulsing the transformer on the right, which feeds the 12 dc supply back to the ic on pin 12 to take over. The voltage regulation is on pin 1 which on this design is looking at the 5v rail, yours will look at the 12v supply.


    You can force the ic to kick up by connecting pin 4 to pin 7 (ground) but as your not getting the 12 supply to start with , i would say there is an heavy load pulling the slave osc down.

    http://danyk.wz.cz/s_atx01a.png
     
    Last edited: Oct 18, 2012
  15. HarwoodD

    Thread Starter New Member

    Dec 26, 2011
    7
    0
    Ok, sorry it took me a few days to get this check done. I live in Okinawa and we’ve been weathering some storms lately.

    So I’ve finally located some low ESR caps of the right value that I can order. Data sheet attached. Unfortunately, I’m moving back to the USA in a week or two so I’ll be unable to order these and receive them in time to pack up. That means I’ll have to wait until my things are shipped across the Pacific to see if that will be the only fix necessary.

    Update: The power supply now won’t fully power up even with the peltier collers disconnected. It seems to be getting worse. The ESR of the big caps is climbing with every troubleshooting attempt.

    I do have a low budget scope, so I checked out pin 4. Doesn’t look good. I’ve attached screenshots of what I found. Photo one is at 2v/div and 2.5ms/div. Photo two is at 2v/div and 500us/div.

    Does this make sense to anyone? Same result with another 1uF cap in paralell.
     
    • Low ESR.pdf
      File size:
      104.1 KB
      Views:
      24
    • 1.jpg
      1.jpg
      File size:
      50.8 KB
      Views:
      48
    • 2.jpg
      2.jpg
      File size:
      33.7 KB
      Views:
      39
  16. Dodgydave

    Distinguished Member

    Jun 22, 2012
    4,969
    744
    try shorting pin4 to pin 7, then power up and see what voltage is on pin 12,also what wave forms are on pins 8,11
     
Loading...