I thought this Power Supply was fixed. I was WRONG!

Поделиться
HTML-код
  • Опубликовано: 19 ноя 2024

Комментарии • 90

  • @danblankenship5744
    @danblankenship5744 Месяц назад +6

    The best way to isolate the root cause is to swap parts between different supplies. When the symptoms move from one power supply to the other then you are 100% sure. It is so nice to have spare working devices for comparisons.
    I have froze a cotton swab and touched this to the likely components to minimize temperature changes to adjacent components. This allows a can of freeze spray (or inverted keyboard duster) last a little longer.

    • @Tony359_2
      @Tony359_2  Месяц назад +2

      The cotton swap idea is cool thanks! When it comes to parts, this PSU's PCB is not the best quality, it falls apart pretty easily (and components too apparently!) so swapping parts is a bit complicated! But thanks!

  • @InssiAjaton
    @InssiAjaton Месяц назад +2

    I would add a couple more parameters to consider. One is the relative humidity that changes with temperature. The other issue is the leakage resistance of the capacitor, especially by humidity, when the capacitor case is cracked. Indeed, the leakage on tantalum can vary by temperature alone, while the ceramic capacitor leakage is predominantly varying with humidity after the case is cracked. You measured the capacitance, but not leakage of the capacitors. In this application the capacitance does not matter much, but the leakage current can matter. By the way, humidity can affect more than capacitors, including the trimpots, fixed resistors, the PCB or the TL431 epoxy housing…

    • @Tony359_2
      @Tony359_2  Месяц назад

      very good pointers, thank you!

  • @Foobar_The_Fat_Penguin
    @Foobar_The_Fat_Penguin Месяц назад +2

    I agree with other commenters who said that this small drop is probably fine. As I understand it, the relevant standard for "medical grade" (IEC 60601-1) is mostly about safety, not accuracy. So you wouldn't expect a "medical grade" PSU to be any more precise than your standard wall wart.
    I think a somewhat simple way to test the whole regulation circuit at once is to just back-feed voltage from your bench power supply into the regulated output (without mains power, of course). Slowly raise the voltage while monitoring the primary side of the opto-coupler in resistance (or diode) mode. When you reach the threshold voltage, the photo transistor should start conducting. Because the primary side isn't actually running, you have excluded it from the loop. If, at that point, the photo transistor reading isn't stable for a given voltage from your bench PSU, your problem is on the secondary side (or in the opto itself). And if it's rock solid but starts to drift in real operation, your problem is either on the primary side (e.g. unstable PWM controller), or in the interaction between the two (some weird oscillation).

    • @Tony359_2
      @Tony359_2  Месяц назад

      ah interesting thanks. I thought that "medical" also meant that the PSU also had some better features like ripple etc. Thanks for that!

  • @anvz6
    @anvz6 Месяц назад +8

    The reference voltage does not matter at all if it's fixed. Any voltage in range of the divider will provide a perfectly stable output voltage controlled by the trimmer.
    The reference itself is not important. What is important is it's stable.

    • @Tony359_2
      @Tony359_2  Месяц назад +4

      absolutely - I think I got that in the end but I made a little confusion at the beginning! Learning as I go :)

    • @K5HJ
      @K5HJ Месяц назад

      I agree totally. I would also add that the capacitor values or not that critical, they are simply used for decoupling or filtering. If the "blue" one was open, that could have caused the problem.

  • @maxtornogood
    @maxtornogood Месяц назад +4

    Some failures are more subtle than what we are used to seeing. Swapping those pieces out has it a lot more stable now!

    • @Tony359_2
      @Tony359_2  Месяц назад +2

      I hope so! Though I might discourage repairs on this model of PSUs if those capacitors self-destruct so easily.

  • @IanScottJohnston
    @IanScottJohnston Месяц назад +2

    Hi Tony, just found your channel, looking forward to more electronic repairs. Not your issue here, but switchers are great fun to repair and if you can limit the incoming AC current they are much easier to fix in the event of a more catastrophic fault. Good old series incandescent light bulb a good starting point.

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      Hello and welcome! I always use my power limiter when doing such repairs. ruclips.net/video/38jQtb0LAhA/видео.html
      It's under my desk and it has a very convenient ON OFF button :)

    • @IanScottJohnston
      @IanScottJohnston Месяц назад

      @@Tony359_2 Ahhhh, and it's nicer than my one, and I like the name....👍

  • @FireballXL55
    @FireballXL55 Месяц назад +6

    I love watching your videos.
    The data sheet actually says that the Ref voltage has a spread from/to at 25 deg C.
    I am not a fan of changing the cap type you replaced the blue "film" cap with a multilayer ceramic cap, these have poor value with heat and or voltage unless you specify the temperature coefficient when you order them, like NP0 X5R X7R etc.
    But depending on the circuit location it may or may not be a problem.
    All the best Rich

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      oh... I thought they were the same type! That was not intentional for sure. Can you tell me more?

    • @tw11tube
      @tw11tube Месяц назад +5

      At that location, I don't think a multilayer ceramic capacitor (MLCC) will be a problem. The originally blue capacitor is likely the cap in the negative feedback from cathode to ref at the TL431. The ref voltage is ~2.5V, and the cathode will be at around 5V - 1.5V (forward voltage of the IR LED in the optocoupler), which is around 3.5V, so there will be 1V over the cap. This is *way* below the rated voltage, so capacity loss is not dramatic. Furthermore, as long as the circuit does not oscillate, the capacitor does its job just fine. Probably you don't need 100nF to suppress the oscillation, but something like 30nF would work in most circumstances and the remaining capacity is just margin.

    • @Tony359_2
      @Tony359_2  Месяц назад +2

      that's great to hear! So for next time I should get a "film" capacitor instead?

    • @tw11tube
      @tw11tube Месяц назад

      @@Tony359_2 I'm actually unsure whether the original cap was a film cap at all. While the marking is slightly different, the original broken capacitor looks very similar to the Murata RCE5C1H123J1DBH03A capacitor, which is a MLCC. Yet, that specific MLCC is of a "better" type. The cheap MLCC capacitors have a dielectric that is characterized by a "type code" of X5R or X7R. Those capacitors have an extreme drop in capacity if you apply DC voltage to them. The specific Murata capacitor I linked has a dielectric with type code NP0, which is considerably better in keeping capacity over DC voltage, but at the same time, a NP0 capacitor is bigger and more expensive than an X5R or X7R cap.

    • @Tony359_2
      @Tony359_2  Месяц назад

      Thanks, I'll consider that for the next batch!

  • @Constantin314
    @Constantin314 Месяц назад +1

    you're using the warm/cold method pretty much in every video, which is fun :)

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      I know! But it's honestly working every time! Well... if I ever manage to put together a follow up of the 18i20, you might see an instance where it didn't work very well!

  • @menotyou8369
    @menotyou8369 Месяц назад +1

    That blue capacitor appears to be a film capacitor. You replaced it with a ceramic capacitor. The temperature coefficient of those caps are way different.

    • @Tony359_2
      @Tony359_2  Месяц назад

      yes, others have pointed that out - though it doesn't seem to be the culprit. But I'll keep that in mind for future repairs thanks!

  • @GapRecordingsNamibia
    @GapRecordingsNamibia Месяц назад +1

    So, I can confirm as I have had quite a stint in PSU design, temperature is a most certain factor, especially when it comes to the TL431. Another thing to note is that there are different grades of the TL431 with will give better or worse accuracy depending on the grade you have, there is also a temperature controlled version of the TL431 like the "oven" type crystals which keeps the temperature of the TL431 consistent, the price though for average electronics is just not worth it, you then also start to reach the realm of GPS, military and the likes which need that accuracy, sadly, along with that accuracy comes quite a price tag, for what this supply is driving I would say this is more than good enough....

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      Ah, the "oven" design, yes. I think I saw that on some Curiousmarc's video, those were ancient HP analogue devices where, obviously, precision was a requirement. This experiment with the feedback circuit was more a curiosity of mine for fun/future reference. At least now I know 500mV=BAD, 50mV=ok, 12mV=perfect :)
      Thanks for adding those details, I love learning new things!

  • @ivolol
    @ivolol Месяц назад +1

    The datasheet shows that a TL431 can, at 25 C, give you anywhere from ~2.45V to ~2.55V (depending whether 1% or 0.5% grade tolerance etc). A TL431 in 1% grade showing 4.77V when you unboxed it, would be totally within spec. After you take that reading, if it is being fed 10mA the whole time, going from 0-70C, it would typically deviate from that reading by +/- 3mV (or 6mV total), depending on the specified grade again. So very simplistically, varying ~80 microvolts per degree C change. The trimmer is there to allow one to adjust out the tolerance of the TL431 and also the tolerance of the dividing resistors.
    It's possible that the resistors you swapped in might have a lot less tempco of their own, whereas the factory ones are likely as cheap as possible; so their ratio changes less and that also gives you a more stable voltage over temperature. If you are super interested, you can dunk the OG resistors, vs your new resistors, in a small beaker of heated mineral oil, and check out their resistance change as they heat up.

    • @Tony359_2
      @Tony359_2  Месяц назад

      You are absolutely right, I am aware my trail of thoughts is a bit confusing but I was learning too! One thing I am not following is: 4.77V? I'm not sure I understand what you are referring to.
      Yes, maybe the resistors I got are better quality but since another stock PSU does not show that drift, I think they degraded a bit over time - or maybe it was a "worse" batch which would still allow the PSU to work as expected. Who knows. But it was fun to experiment and learn!
      Thanks for helping out!

    • @ivolol
      @ivolol Месяц назад

      @@Tony359_2 whoops! *2.477V

    • @Tony359_2
      @Tony359_2  Месяц назад

      ahhh! It was 2.37V at the beginning, outside the expected range I think.

  • @TheDefpom
    @TheDefpom Месяц назад +1

    @31:20 its the temperature coefficient that influences the drift with resistance, not so much the defined accuracy.

    • @Tony359_2
      @Tony359_2  Месяц назад

      yes, you are correct! I noticed the mistake while editing, I left it there to show my "trail of thoughts", very aware that it might be a bit confusing :) Thanks for helping!

  • @general23cmp
    @general23cmp Месяц назад

    Well done

  • @pepealexandre
    @pepealexandre Месяц назад

    Great explanation and editing also!
    A lot of work, thank you, sir.

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      Thanks! It does take quite some time indeed!

  • @TheOriginalEviltech
    @TheOriginalEviltech Месяц назад

    Hot air is great for localizing the area, but you can use the soldering iron to heat up certain components without heating up the neighbouring components.

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      you're right, I forgot to try with the iron itself, thanks!

  • @TheDefpom
    @TheDefpom Месяц назад +1

    The voltage reference actual value is not that important, what is importance is that it is stable and consistent.

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      indeed - I know the "thought process" of this video is a bit messed up! But now I have learnt something more!
      I think the red flag is more that the voltage into the TL431 was outside of the 431 tolerance and changing with the adjustment when I initially tested. But I instead focussed on the absolute value which - as you and others have pointed out - is not an issue as long as it's stable.
      Thanks for watching!

    • @TheDefpom
      @TheDefpom Месяц назад

      @@Tony359_2 was still a good and interesting repair. That blue capacitor is a dipped ceramic type, ceramic caps drift wildly with temperature by the way, but being cracked was certainly the main issue there.

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      Thank you Sir!
      Yes, I more or less intentionally overthought this one, my curiosity prevailed :)

  • @insanelydigitalvids
    @insanelydigitalvids Месяц назад

    A great video and a great learning experience!

    • @Tony359_2
      @Tony359_2  Месяц назад

      Thank you, I'm happy you enjoyed it!

  • @DustinWatts
    @DustinWatts Месяц назад

    Capacitors are usually 10-20% tolerance... for lower tolerances... you pay! Haha

  • @HV_SMPS
    @HV_SMPS 9 часов назад

    Hello,
    Thank you very much for your great tutorials.
    I have an old power supply for repair. Its output voltage has a low-frequency ripple with an amplitude of about 1 volt. However, if I draw even a very small current from it momentarily, the ripple disappears.
    What do you think the problem could be?

    • @Tony359_2
      @Tony359_2  6 часов назад

      I'd say it could be by design. Very old power supplies just don't like NOT having a load. Most won't start or will "hiccup" or will output much higher/lower voltages.

  • @SidneyCritic
    @SidneyCritic Месяц назад

    I've seen carbon resistors drift but only after decades, but it was surprising metal-film drifted, ie, caps are more drift prone. Maybe the casing shrunk and humidity got in.

  • @rolandturner4891
    @rolandturner4891 Месяц назад +1

    I appreciate the exhaustive exploration in the second half of the video :-) If I may return the favour...
    Your measurements and free-standing tests both assume DC and a static condition, neither of which is true:
    - It's a switching supply, so there's at least some ripple on both the TL431's "supply" and on its input.
    - The system is in dynamic equilibrium, and the TL431 is inside the control loop. Tiny disturbances can readily escalate.
    This is a bit trite, except that the TL431 has its own internal damping, and therefore a zone of control-loop stability that's affected by any capacitance presented by the circuit. See Fig. 6-16 or 6-18 of the datasheet. The device condition must be *above* the relevant curve in order to be stable. Any capacitance presented by the circuit must therefore be either:
    - negligible (sub-nF) so as not to affect the built-in damping at all; or
    - overwhelming (>~5 uF) so as to ensure stability no matter what the TL431 does.
    If the circuit designers have done the latter but the capacitor is faulty then the condition may drift into the zone of instability as the capacitance falls, whether because of a mechanical fault, or just that its not an NP0 capacitor. Quite what would happen if the TL431 turned into an unintended oscillator and therefore began beating against the pulse-width modulator is hard to predict, but I wouldn't rule out a sustained gradual drop in output voltage as the capacitance drifted further and further into the zone of instability.
    www.ti.com/lit/ds/symlink/tl431.pdf

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      Thanks for your kind explanation, I appreciate the time you've spent to share that with me.
      When that tantalum marked in green fails, I see what you mention: the output starts jumping (and the device using that rail crashes!).
      I might want to make sure I get better capacitors for next repairs, I wasn't aware there were different types available.
      Thanks again!

  • @zorrorides1
    @zorrorides1 Месяц назад

    Curious; please check the ESR of the subject capacitor. See if that changes and how much when heated and cooled.

    • @Tony359_2
      @Tony359_2  Месяц назад

      I didn't think about checking ESR, thanks!

  • @LozzTheDev
    @LozzTheDev Месяц назад

    If you get a meter that measure ESR (Equivalent Series Resistance) for capacitors it will probably show an issue with that faulty Tantulum capacitor

    • @Tony359_2
      @Tony359_2  Месяц назад

      the broken one is not a tantalum it's a film one. The one with the green mark would not read as a capacitor anymore when it fails :)

  • @walter7671
    @walter7671 Месяц назад

    I have nothing to add except that it is probably raining in Edinburgh.😅 I enjoyed the analysis, as always. Greetings from Mar del Plata, Argentina!

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      I'm far away from Edinburgh but I suspect you might be right :) Thank you!

    • @PIXscotland
      @PIXscotland Месяц назад +1

      It was more frosty that rainy actually. 3c on the car today so I guess the voltage was 5.16v...

    • @Tony359_2
      @Tony359_2  Месяц назад +2

      the capacitor has definitely gone then! :)

    • @walter7671
      @walter7671 Месяц назад +1

      @@PIXscotland So it's a lie that it always rains in Edinburgh, I was wrong. 😬

  • @K5HJ
    @K5HJ Месяц назад

    The minimum and maximum voltage of the TL431 in the data sheet are at 25 degrees C, not over the temperature range.

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      Indeed, I read it wrong! It's basically telling me that different ICs might have slightly different reference voltages and then there is a coefficient to calculate the voltage drift based on temperature!

  • @davidrobertson1980
    @davidrobertson1980 Месяц назад

    Hmmm with your nice big heat sunk resistive load also changing temperature with the room temp it could be another issue ;)

    • @Tony359_2
      @Tony359_2  Месяц назад

      The tests were done with the device the PSU belongs to. Let's say it's like a computer.

  • @johnsickmann9224
    @johnsickmann9224 Месяц назад

    Hi Tony, Was watching this and noticed I couldn't see where the -Ve (COM) lead of the FLUKE was connected for most of the test. Only at the last test did you have the probs in the connector on the PCB. I caught that you said the test load was several Amps. I have been caught by the PTC nature of the copper in the PCB and your connection wires. At several Amps both the PCB and wires will a measurable voltage drop and this might be affecting your results. Also the Connector will add quite a bit to the reading in both resistance and thermo couple affects of dissimilar metals. Try again with the ground of the Fluke not in the current path of the load when testing the TL431. Also try measure the voltage drop of each wire to the load I was surprised the first time I ever did this.

    • @Tony359_2
      @Tony359_2  Месяц назад

      Thanks for the hint!
      The values you see on the graphs were actually measured at some test points of the device under test. So the whole appliance was in the way. I made sure the device was idle in the same state every time. It's an old device so it's not doing anything in the background which might change the current drawn over time.
      Yes, there will be a voltage drop for sure but as long as the load is the same every time, the drop won't change and can be ignored.
      The issue was that with a load in the way (whether it was the appliance or my dummy load) the voltage would slowly decrease. That should not happen, wherever you are measuring - assuming that
      1. the load is constant
      2. Wiring is good, that is they're not warming up, changing their internal resistance etc.
      Have I understood your points correctly?

    • @johnsickmann9224
      @johnsickmann9224 Месяц назад

      @@Tony359_2 Thanks for the reply yep you have got the points spot on.
      I also made the comment so other readers might think about where and how they measure. Now a side note I once bought some power supply leads and they dropped 350 mili-volts in each wire at 1 Amp ..... Looked good but nearly all insulation and only 0.3mm of wire. They got quite warm to touch.
      Have a great day from New Zealand.

    • @Tony359_2
      @Tony359_2  Месяц назад

      Your comments are very welcome - thanks for contributing!
      Ah yes, I'm familiar with those cables. I guess plastic is cheaper than copper! Hello to all my Kiwi friends! :)

  • @eduardoavila646
    @eduardoavila646 5 дней назад

    For these heating tests, wouldn't it be better to touch a tip of a soldering iron to the component instead of blowing hot air at them, that way the heating of adjacent components would be minimal?

    • @Tony359_2
      @Tony359_2  5 дней назад +1

      indeed - I simply forgot to use the technique :) Thank you!

  • @AnnaVannieuwenhuyse
    @AnnaVannieuwenhuyse Месяц назад

    "Plateau" comes with an "eau". It can be pronounced like "oh" in English. That becomes plat-oh. :)

  • @humidbeing
    @humidbeing Месяц назад

    Most power supplies are +/- 5% at best. 50mV is nothing to worry about. A handheld dmm doesn't even have the tolerance for that really..

    • @Tony359_2
      @Tony359_2  Месяц назад

      Absolutely - it was just a way to learn and explore!

  • @repairitdontreplaceit
    @repairitdontreplaceit Месяц назад

    it may well be that its stable under load !

    • @Tony359_2
      @Tony359_2  Месяц назад

      I'm not sure I am following?

  • @dolall5470
    @dolall5470 Месяц назад

    I wonder why You did not focus on thermal drift of voltage divider resistors more then capacitors?

    • @Tony359_2
      @Tony359_2  Месяц назад

      Because the video was already 45 minutes :)

  • @FUNKLABOR_DL1LEP
    @FUNKLABOR_DL1LEP Месяц назад

    20:40 hi, please what kind are this small black and red testclips placed all the time in this board?

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      Ah, I got them from a local shop here. Try looking for IC GRABBER. They are expensive...

  • @GaRbAllZ
    @GaRbAllZ Месяц назад +1

    45 mV??? Did you take temp delta of the Fluke into account? Just curious...

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      no, but you might be reaching the end of the video soon :)

  • @humidbeing
    @humidbeing Месяц назад

    All tantalums change value with big temp changes.

    • @Tony359_2
      @Tony359_2  Месяц назад

      was that a tantalum?

  • @Dutch-linux
    @Dutch-linux Месяц назад

    timestamp 12:14 there is also a transistor of some sort... is that defective and reacting on temp changes ?

    • @Tony359_2
      @Tony359_2  Месяц назад +1

      that's the TL431 :)

  • @brunonikodemski2420
    @brunonikodemski2420 Месяц назад

    Unless you are a Technician or Engineer, it is NOT worth repairing power supplies. Just buy another one. Once you have any of the electrolytics fail, the rest of them are also dried out, and will soon fail also. It becomes a "wasted time" race, for doing actual repairs. For medical supplies, this could actually be a local "crime", depending on the legal regulations. You need to be a registered repair shop, for anything relating to people, if they die or are damaged as a result of your actions.

    • @Tony359_2
      @Tony359_2  Месяц назад

      You are totally right - if you watch the main channel's video of this repair, I did say that straight away. But the purpose of this channel is to learn and tinker.
      The PSU is "medical grade" but does not live in a medical device. I would never think of fixing a medical device!!
      Thanks for watching!