5
\$\begingroup\$

We have got a few boards with failed power. Some boards went alive after replacing a bad capacitor (swollen can). Some boards went alive after replacing a zener diode (somehow shorted to ground I think. My colleague threw it away so I do not know what condition it was at). The said capacitor, C101, and zener diode, ZD1, are located in the lower left corner in the SMPS circuit shown below:

enter image description here

These two are neighbours so naturally I think they might fail due to the same source. However, I have no clue why they failed. Are there any design flaws in the circuit? These are of small percentage of the boards out in the field, but I think it must be some design problem that triggers the problem intermittently. Please also note that the capacitor that died in the field is rated 50V.

Here are some measurements at Vtrans, Vdiode, and Vzener. Vtrans (with load at Panel_12V): enter image description here

Vdiode (with load at Panel_12V): enter image description here

Vzener (with load at Panel_12V): enter image description here

Vtrans (without load at Panel_12V): enter image description here

Vdiode (without load at Panel_12V): enter image description here

Vzener (without load at Panel_12V): enter image description here

The waveform before D120 looks like this during a typical load at Panel_12V (It looks terrible but I don't know if it's normal or not...): enter image description here

Zoomed shot of the waveform before D120 looks like this during a typical load at Panel_12V: enter image description here

The board also emits a hissing sound (rather loud, compared to other boards with switching power circuit I have seen). I don't know if it's relevant to the cause of the bad capacitor and zener diode. It is also a problem we want to solve though.

Please find the datasheet of the SMPS IC, FSDM0465RB, here: https://www.fairchildsemi.com/datasheets/FS/FSDM0465RB.pdf

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Probably not the cause of the failure, but C109 has its negative end connected to GND, surely this should be Panel_GND. It looks like you have breached the isolation of the transformer and opto-isolator. \$\endgroup\$
    – Steve G
    Commented Sep 4, 2016 at 14:45
  • \$\begingroup\$ @SteveG VDD_15V shares the same ground as U101. It is being used in another part of the circuit. \$\endgroup\$
    – Brian Wang
    Commented Sep 4, 2016 at 15:47
  • \$\begingroup\$ Even though thisis a low power circuit I also get a bit nervous of the linked input and output grounds. The high voltage side should have a 'neutral' and a protective ground and even after rectifying the AC input (if this is the case) I would still keep the protective ground separate. After a fullwave bridge (required if you cannot tell live from neutral) you will have a diode-drop between the ground reference and the negative rail, making or assuming they are the same is a mistake. \$\endgroup\$
    – KalleMP
    Commented Sep 4, 2016 at 18:12
  • \$\begingroup\$ Fairchild's reference design shows a 10nF coupling cap between input and output ground. So primary insulation design is a key factor. \$\endgroup\$
    – Tony Stewart EE75
    Commented Sep 4, 2016 at 18:41

2 Answers 2

Reset to default
3
\$\begingroup\$

You should put a scope on the node between the ZD1 cathode and the ++ end of C101. Monitor what happens there under all conditions of low and high range of the VCC/340 node voltage range. For each input condition also vary the current loads on the PANEL_12V and the VDD_15V from the minimum expected levels to the maximum. Also at each of the four corners testing switch the supply on and off a few times.

You will likely to find conditions where ZD1 is being asked to dissipate significant power or voltage spikes or voltage levels over the rating of the C101 capacitor.

\$\endgroup\$
1
  • \$\begingroup\$ Does the VDD_15V load affect the SMPS IC's behaviour? I thought only the load at PANEL_12V is the main factor here. I will do more measurements in these different conditions tomorrow. C101 that died in the field is rated 50V (I will edit the post to reflect that). Thank you. \$\endgroup\$
    – Brian Wang
    Commented Sep 4, 2016 at 16:11
3
\$\begingroup\$

Specs indicate 5mA max operating current for U101, thus evaluate your Zener current for excess voltage from winding pin 7 via 100 Ohm.

The peak voltage only needs to be 13V at Vac min. and current 3~5mA avg. thus if you are burning out Zener, winding ratios are too high or input V range too wide?

U101 has an absolute max V of 20V

Hissing is often from lack of insulation protection and ion stream noise(low current corona to a sharp point).

Power Line Transients , PLT can be a potential problem. So insulation needs to considered for 3kV, 10us pulses.

last photo indicates 80Vpp is where? transformer out? If so current is too high.

PLT is an IEC EN61010, power quality stress test. http://www.wernermn.com/assets/files/PDF/Solution%20Expo/Presentations/PQ/PQ03.UsingTestEquipmenttoDetectandMeasurePowerQualityIssues.pdf

Ripple current rating of C101 must exceed design, hence low ESR because turns ratio from primary to zener is only 5:1 from 340Vdc to 13V thru 100OHm R101 is too high current peak 340/5~70Vp/100 Ohm is 700mA ripple current pp or 350mA rms and some 22uF caps are only rated for 20mA rms ripple current.

\$\endgroup\$
12
  • \$\begingroup\$ I have added some scope shots. The voltage at the zener diode is around 13V in typical usage scenario. Or maybe there's some other conditions that need more testing? Could you elaborate more on "ion stream noise" and "Power Line Transients"? Thank you. \$\endgroup\$
    – Brian Wang
    Commented Sep 4, 2016 at 16:17
  • \$\begingroup\$ last photo indicates 80Vpp is where? transformer out? If so current thru R129 does not compute to (80-13)/100 \$\endgroup\$
    – Tony Stewart EE75
    Commented Sep 4, 2016 at 16:32
  • \$\begingroup\$ I have added more measurements to the post. Please have a look at them and see if they are helpful for identifying the problem. \$\endgroup\$
    – Brian Wang
    Commented Sep 5, 2016 at 2:30
  • \$\begingroup\$ The loud hissing sound came from C103, which is a ceramic capacitor. We replaced it with a film capacitor and the noise level came down significantly and became acceptable. \$\endgroup\$
    – Brian Wang
    Commented Sep 7, 2016 at 0:21
  • \$\begingroup\$ We also measured the current flowing through R119 with a multimeter. The reading shows a 8 to 9 mA. This is an RMS reading and does not reflect the huge voltage swing seen on the scope. Is it reasonable to assume the zener diode is safe? \$\endgroup\$
    – Brian Wang
    Commented Sep 7, 2016 at 0:27

Not the answer you're looking for? Browse other questions tagged or ask your own question.