1
\$\begingroup\$

Update

With the multimeter and a steady hand, I've found the 3843B PWM controller Vcc pin is 12V when the supply is on, and 8.2V when it is plugged in but cold. Datasheet says it needs 8.4V Vcc to start-up so heating it is lowering its startup voltage or increasing the Vcc just enough.

To test the theory I supplied external 9V power to the pin, to "jump-start" the power supply. It worked!

So now, I think I need to find where this 8.2V is coming from so I can work out how to bump it just that little bit more, maybe by replacing a diode or something.

How this "start-up current" is usually generated on power supplies? Maybe from the pictures you can point to some probable locations I should look? What could be the cause of the low start-up voltage? Thanks in advance.


Okay, this is a weird one. First time on this Stack Exchange, work with computer science, and have only basic electrical knowledge, so beware of any possible misunderstandings below.

Some context

I have an "universal" laptop power supply, one you choose the desired output voltage with a switch, it goes from 12 to 24V. Since some time ago it started working only when warmed up. So to start using it, I had to warm it up with a hair-dryer and then couldn't leave it turned off for too long, or it would cool down.

I decided to fix this for good because it was getting worse over time (needed more temperature to start-up), it looks fun, and I am broke. So I opened it up, armed with a multimeter and started probing. It takes the 110/220V input AC, rectifies it and then passes it through a MOSFET (gate, I suppose, connected to some signal generator, the frequency of which maybe determined by the voltage selector) and onto the main coil, which gives our output voltage which is rectified again, filtered and then out. So the problem is, whatever drives this MOSFET wasn't working when cold. I narrowed the problem to 3 ICs and did the logical: warmed them one-by-one with a lighter to find which one was problematic.

TL;DR: The actual question

So I narrowed the problem down to an IC called "3843B" which turns out to be a "high performance current-mode PWM controller". So, heating this IC (or something on its close vincinity) gets the power supply going. After it starts working, then everything is OK. Below is a photo of it.

What could be causing this? Which of the suspect components in the photo could have a failure mode which makes it work only after it's hot? What can I do or which components do I replace?

Close-up of suspect

IC which when heated, makes the power supply work

Board overview

Annotated board overview

\$\endgroup\$
2
\$\begingroup\$

Check the below-pictured resistor out (soldering and value)- just a hunch.

enter image description here

\$\endgroup\$
  • \$\begingroup\$ It looks burnt but that is just flux that was left over it. I will give it a test tomorrow. \$\endgroup\$ – NothingsImpossible Apr 29 '16 at 11:10
  • \$\begingroup\$ I checked and that resistor seems OK! I updated the question with new info, please take a look. This resistor has more than 200V on it, so it's unlikely it relates to what I found. In short, start-up voltage is 0.2V too low. \$\endgroup\$ – NothingsImpossible Apr 30 '16 at 2:50
  • 1
    \$\begingroup\$ The start-up dropping resistor is a common failure mode- the resistance increases from nominal to the point where it barely starts. However since you have removed it and measured that the resistance value is close to nominal you can rule that out. \$\endgroup\$ – Spehro Pefhany Apr 30 '16 at 12:28
  • \$\begingroup\$ Ah, I didn't remove it... I analysed the board some more and found a series of resistors which take 311V from the cap and drop it to the desired voltage. Seems like it! I will try replacing one of them. \$\endgroup\$ – NothingsImpossible Apr 30 '16 at 12:47
  • 1
    \$\begingroup\$ SOLVED! That capacitor to the left where you pointed out, was bad. I tried changing the resistors for lower values but the voltage was always exactly 8.25V. Hmm.. The only other component connected at this pin was that capacitor... Turns out it was leaking too much current and that was "regulating" the output to 8.25V. Thank you very much. \$\endgroup\$ – NothingsImpossible Apr 30 '16 at 16:30
2
\$\begingroup\$

Could be a bad chip, or could be a bad solder joint. Easy to find out if it's a bad joint- just take a soldering iron and reheat all of the joints and leads in that area (including pin 6 on the IC- solder looks a little sparse there). Use a fine tip on your iron, and add a little flux to aid in the reflow (a tiny bit of added solder might help too, but not too much). Use as low a temperature as you can and don't heat a joint to the point that other joints unsolder themselves. Start with the IC, and test the PS after things cool off. Then move on to other joints, one at a time. Moderation!

\$\endgroup\$
  • \$\begingroup\$ While you're at it, reflow the joints on the nearest pin-through components (that electrolytic and the surge suppressor). \$\endgroup\$ – Brian Drummond Apr 29 '16 at 10:17
  • \$\begingroup\$ Nice suggestion. I will do it and come back with results tomorrow. \$\endgroup\$ – NothingsImpossible Apr 29 '16 at 11:13
  • \$\begingroup\$ Also, there seems to be a missing component (empty pads with solder that appears disturbed) in the top-center of the close-up view. Might not have been there to begin with, but it seems strange. \$\endgroup\$ – BobT Apr 29 '16 at 22:16
  • \$\begingroup\$ I checked and It seems the chip and solders are OK! I updated the question with new info, please take a look. In short, start-up voltage is 0.2V too low. \$\endgroup\$ – NothingsImpossible Apr 30 '16 at 2:44
0
\$\begingroup\$

You may follow the good instructions of the answers and comments given, but it nothing solves your problem, most likely the chip has bad gold wire bonding inside it.

It could be stitch crack or poor intermetalic weld on chip bond pads. This happens when wire bonding process was poor during chip assembly. In this case you got to replace the smd chip.

\$\endgroup\$
  • \$\begingroup\$ I thought of that too! Fortunately the IC is OK, I updated the post with some new info, take a look. \$\endgroup\$ – NothingsImpossible Apr 30 '16 at 2:51
0
\$\begingroup\$

It's hard to guess without oscilloscope and the device. There is an optocoupler, four-legged ic right next to the diode bridges, would start checking it

\$\endgroup\$
  • \$\begingroup\$ Yes, unfortunately oscilloscopes are so expensive T_T That four legged IC was one of my suspects, but heating it did not help it turn on. \$\endgroup\$ – NothingsImpossible Apr 29 '16 at 11:12
  • \$\begingroup\$ By the way. Maybe it's NTC. Try looking for a large component that behaves as a resistor, but it's resistance depends on temperature \$\endgroup\$ – Gregory Kornblum Apr 29 '16 at 11:16
  • \$\begingroup\$ I updated the question with new info, please take a look. \$\endgroup\$ – NothingsImpossible Apr 30 '16 at 2:45
  • \$\begingroup\$ If you are sure about that, just replace the ic. Probably it's internal mechanical issue. \$\endgroup\$ – Gregory Kornblum Apr 30 '16 at 4:06
  • \$\begingroup\$ No no, the IC is fine! Its just not being supplied enough power to function. I am looking into that. \$\endgroup\$ – NothingsImpossible Apr 30 '16 at 6:49
0
\$\begingroup\$

I don't have enough experience but sometimes a NTC resistor is used to limit inrush current and its value only gets down when it's "warm". It may be this component at the top:

...

I wish this is helpful

\$\endgroup\$
  • \$\begingroup\$ That blue thing is labeled "DCF 222M", couldn't find it online. I will make some more controlled tests on the suggestions tomorrow. \$\endgroup\$ – NothingsImpossible Apr 29 '16 at 11:09
  • \$\begingroup\$ Try warming it on specific, good luck. \$\endgroup\$ – iMohaned Apr 29 '16 at 11:47
  • \$\begingroup\$ I think the blue thing is a "Dcf Y1/X1 Safety Capacitor" sold by Zhengli Group. No idea what it's making safe... \$\endgroup\$ – BobT Apr 29 '16 at 22:10
  • \$\begingroup\$ @BobT hah, no idea too. Googled it and it's exactly this one, just a capacitor. But the problem seems not to be NTC too. I updated the question with some new info, please take a look. \$\endgroup\$ – NothingsImpossible Apr 30 '16 at 2:55
0
\$\begingroup\$

probably the black capacitor next to the 2nd rectifier. These tend to have a higher leak current than the specs when they dry. The PWM controller would only produce low duty cycle or interval because the feedback mechanism detect a short circuit in the DC side.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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