Restoration & Repair

Enter Your Electronics & Design Project for a chance to win an $200 Shopping cart of product!

Submit an EntrySubmit an Entry  Back to homepage
Project14 Home
Monthly Themes
Monthly Theme Poll

 

In part 1  we had a good look at the power supply internals, figured out the schematic and the functional blocks of the circuit. This gives us a good starting point to troubleshoot the problem, in search for an (hopefully) easy fix. Let's start by describing the problem.

 

 

Power Supply Failure Symptoms

Simply put, the 12V DC output was dead, no power! Plugging in the power supply to the mains did not produce the expected 12V at the output, not even an unregulated voltage of any sort or a ripple...zero, nil! Unfortunately, the failure symptom didn't provide much insights, as there are plenty of possible causes for no output. Let's start troubleshooting then, to try and narrow down the culprit(s).

 

Troubleshooting the failure

Time to narrow down the possible cause by using some strategy to check the different functional block of the supply. As I only do this kind of repair jobs in my spare time, mostly for fun, I don't really have a proven and tested troubleshooting strategy for SMPS,  I would just use my everyday problem solving approach to try and track down the root of the problem.

 

Step 1 - Visual check

Before firing up the board and go poking around with probes, it was worth visually checking the components for any sign of stress/breakage which might hint to more serious problems. The first candidate was obviously the fuse, just in case it had blown. No sign of damage there. Then a quick glance to the power transistor Q1, the transistor Q2, the dual diode DP1, the optocoupler PC1 and the precision reference IC1. All good there. A quick check at all the electrolytic capacitors, to look for any bulging or leaking, but no sign of damage there either. All the 1/4W resistors (the smaller ones) seemed fine, with the exception of R9, which looked a bit dark and with a small crack, but it might mean nothing, as the only way to be sure is to desolder it from the board and check with the tester (noted down, for later). R4 and R5, both 3W resistors, looked a little "tired" as well, so they need checking too. All the diodes seemed fine, and so the rest of the capacitors. Last, but not least, the transformer looked OK too.

 

Step 2 - Check Protection, Filter and Rectifier circuits

Time to power up the SMPS, to check the problem was not due to a failure of the components placed between the power line feed and the rectifier output. The quickest check to perform was to read the voltage at the output of the diode bridge, as a good reading there means all the components upstream of the rectifier were working properly. Measuring the DC voltage I got a reading of 326V, which was the value I expected to read, so the problem must be further down the circuit.

 

Step 3 - Check Transformer

Typically, failure of the transformer is not really a common occurrence, as it tends to be quite a reliable component. With this in mind, I assumed the transformer was OK (beforehand, when I was still tracing the circuit, I did measure the resistance of the windings, to map them correctly, the result is shown in the photo), therefore, I thought it made sense now to measure the windings voltages. The idea was to understand if the problem was on the hot side or on the cold side of the transformer. Across the primary, I could read a DC voltage, and I got some AC voltage on the auxiliary too, while the secondary did not have any voltage across it. Not having tested anything on the secondary side yet, I could not exclude there was a fault there (some shorted caps perhaps) causing the zero reading. So, I decided to check that first.

 

Step 4 - Check for shorts on the secondary side

To test for shorts, I set the multimeter on the resistance reading with beeper (so I didn't have to constantly look at the multimeter screen), and with one probe fixed on the cold ground, I moved the other probe, testing all the pads not directly connected to the ground. As expected, I heard a beep only when I touched the other end of the secondary winding, so looked like no shorts on the secondary side of the power supply. This left only one place to look for fault: the switching transistor with the PWM control circuit.

 

Step 5 - Remove components from PCB for testing

Testing the components in that switching and PWM control part of circuit couldn't be done without removing them from the board, so that is what I did next.

 

Q1 - the switching power transistor

The first component I decided to check was Q1 (2SC4236, NPN ,TO-247 package). Cranking up the solder iron temperature to 350C and with the help of a solder pump, I managed to remove the transistor without too much fuss. It looked in good condition, although the marking was barely readable (see the photo gallery). To check the BJT transistor was good, I relied on the fact that NPN transistors can be seen as 2 diodes (N-P junction) connected back to back on the anodes. Setting the multimeter on diode testing mode, it will read the voltage drop across the P-N junction. As can be seen in the photo gallery, both junctions did show an healthy voltage drop (the difference in the reading for the junctions is due to the construction geometry of the junctions themselves: they have different sizes, and the voltage drop depends on the size of the junction). Q1 was good, so I needed to look for a fault elsewhere.

 

{gallery} Q1 - NPN power transistor 2SC4236 testing

Q1: The transistor removed from the board

Q1: Testing the Base-Emitter junction

Q1:Testing the Base-Collector P-N Junction

 

 

Q2 - The control transistor

Since Q1 was good, I needed to keep going with removing components. Next, was the turn of Q2 (2SC1383, NPN, TO-92L package). Removing this transistor was a lot easier than Q1, as it had less thermal mass. Again, visually the transistor looked in good condition. Once out, I repeated the same junction test performed on Q1, as detailed in the photo gallery. The first surprise: the Base-Emitter junction looked like an open circuit, not looking good! The confirmation was the continuous beep I heard once connected the probes to the Base-Collector junction: it was shorted. Definitely this transistor was bad, and hopefully the sole culprit for the power supply failure, as its duty was to help turning Q1 on and off!

 

{gallery} Q2 - NPN control transistor 2SC1383 testing

Q2 : The transistor removed from the board

Q2:Testing the Base-Emitter P-N Junction

Q2:Testing the Base-Collector P-N Junction

 

 

Step 6 - Replace components

Finally there was a hope that I could bring the power supply back to life. All I needed is to get a replacement for Q2. The marking on the transistor was "C1383", and after searching on the internet, I found the datasheet. Unfortunately none of the distributors I usually buy components from seemed to have such part in stock, so I had 2 alternatives: order from China (and wait for a month for the delivery), or find a suitable replacement part. I decided to go for the latter, and after comparing a few transistors, I finally found a decent substitute: KSC2383YTAKSC2383YTA. All the maximum ratings were met or exceeded, with the only exception of the power dissipation, where the original was 1W, while the replacement was 0.9W (something to keep in mind, especially if the collector current, with the power supply in full load condition, gets closer to the maximum rating).

 

Whilst replacing Q2, I decided to replace Q1 as well, just in case with Q2 failing, the transistor had been subject to stress too. So, again, went on to search for the part 2SC4236, and again, nothing came up. For this transistor though, I have a possible replacement already, as some time ago I ordered some 2SC4237, which turns out to be just a "beefed-up" version of the C4236, so I decided to use that. Since I had to order Q2, I also decided that, while I was replacing some of the part, it made sense to order also a replacement for all the electrolytic capacitors (the bulky filter capacitors probably didn't need changing, as they usually keep in good state a lot longer than the smaller caps, but I needed to reach the threshold for free next day delivery ). I also decided to buy a replacement for the ultra fast switching diode pair DP1, which had marking NDL 020-10F. I wanted to replace it regardless if it was working or not, as its packaging (TO-220) didn't match the holes and the space on the board (TO-247). Looking for this part, I found nothing at all. The only information I managed to get was that it was a 10Ax2, 100V dual diode. As replacement for this, I found the STTH20W02CWSTTH20W02CW, 10A x 2, 200V, and comes in TO-247! I place the order for next day delivery, so the next day I was ready to put the shiny new Q2 in! I did not remove any of the extra components marked for replacement from the board yet, except for Q1 and Q2 that were already out, as I wanted to try the fix before messing around with that.

 

Step 7 - Test the repair

Once the components arrived, I soldered the new Q1 and Q2 on the board, cleaned up and connected the power supply to the mains and placed the probes on the 12V and GND pads, switched on the supply and... it worked! Q2 was indeed what caused the power supply to shut down. Well, while testing, I also noticed the small LED indicator wasn't working, so another component to add to the list of things to replace...

 

 

With renewed confidence and optimism, I decide then to "up-cycle" the power supply, giving it brand new capacitors and also, while the iron was still hot, replace R4, R5 and R9. The first 2 were 3W resistors, which I didn't have available (didn't think of replacing them when I placed the order the day before), so I went for a 5W for R4 and the only 0.22 Omh resistor I had was a 7W for R5 (R5 is the sense resistor on the emitter of Q1). R9 was 1/4W, and I had a replacement for that (bigger tolerance though).

 

 

{gallery} Replaced components

1: Replaced transistors, dual diode and caps

2: R9 showing signs of "fatigue", but still working

 

After the desoldering and soldering session, in the photo you can see the final result (I have adjusted the trimmer to get as close to 12V with no load as possible).

 

 

The final test was to put the power supply under load conditions and see how it behaved. Ideally, this test should have been performed using a programmable load, to verify how the output voltage changes with the current drawn, but I didn't have one, so I had to make the most of what I had, which was a bunch of 10 Ohms 5W resistors and a 1 Ohm 50W one. I arranged the 10 Ohm resistors in groups of 5 in parallel, to get roughly 2 Ohms from each, then put all in series, to get about 5 Ohms. With this load, the supply should provide about 2.4A of current. Below you can see the setup.

 

 

The test performed is shown in the video. Once under load (reading from the multimeters, the voltage was 12.02V and the current 2.29A, for a total power of about 27.5W, so nearly half load), everything seemed quite stable, the only noticeable thing was ah hi-pitched hiss coming from the transformer. Not sure if this was due to some mechanical problem (transformer getting a bit "loose" as it is ageing) or if the replaced components play a part in it, as they are not a perfect match. Anyway, the power supply seems to be OK now and, most of all, this fix saved it from ending up in the bin, while at the same time showing to my wife that I eventually do fix things that I keep .

 

 

Final remarks

Fixing the power supply has been quite interesting, and very educational although, I must admit, it hasn't been very economically savvy. I mean, if I stuck with replacing only the bad transistor, then economically would have been advantageous, as the transistor itself costed peanuts. But when replacing also all the electrolytic caps, especially the bulky filter ones, it started becoming costly, to the point that, without including the time I spent doing it, I ended up spending about £20, and on Amazon a new unit, same model, would cost you £17.58 including shipping! But, as I don't do this as my day job, it still made sense to do it, and I have enjoyed it. One thing for sure, I definitely need to learn a lot about the fascinating world of power supplies!