Wednesday, May 15, 2013

Repairing a Telecom AV-830 switching power supply (or maybe upgrading it)

I have an auxiliary Telecom AV-830 switching power supply I use from time to time. Recently while testing a circuit with it, the power supply explodes. The internal fuse was disintegrated and the main ICP was activated. That usually means something very bad is happening with the power supply.

After opening the case nothing burned was found, except for the rests of the fuse: the glass got broken into million of tiny glass pieces. But except that, no further damage where visible.

Testing with the ohmmeter I discovered everything after the diode bridge was in shortcircuit. After some minutes, I figured out the two main switching transistors were damaged.

This power supply uses a pair of 2SK3469 power mosfet transistors to do the switching. The 2SK3469 transistor has a TO-220 plastic case, so they was mounted directly to the heatsink, without thermal grease or insulating material. After desoldering them, I found both transistors were in total shortcircuit: from any pin to any other pin they had zero ohms, even from the gate to drain and source! Something really catastrophic happened with the transistors.

Desoldering the transistors, I noticed the PCB has a double layout for these transistors. There was the TO-220 holes, but there was other holes, suggesting a larger transistor could be used. The extra holes seems to indicate a TO-247 transistor could be used also.

Then I remembered I had an spare PC ATX power supply, so I opened it to take a look. This ATX switching power supply had a pair of 2SK2698 power mosfet transistors, in TO-247 package. What a nice coincidence!

After studying both transistors datasheets I concluded the larger 2SK2698 could be a substitute of the smaller 2SK3469, so I decided to unsolder the 2SK2698 transistors from the ATX PC power supply and install them into the Telecom AV-830. It should work...

The 2SK2698 transistors fitted into the heatsink

Interestingly the new transistors had a small ferrite bead on the gate pin, and I left them in position: It should prevent high frequency oscillations. Because the new transistor cases are not 100% plastic, I installed insulating sheets and added a bit of thermal grease.

To prevent another explosion, I tested the power supply connecting in series with the mains a 40W light bulb ( a classic one, not a low energy one! ) After turning on the power supply, the bulb briefly turned on, and then the power supply was outputting 13.8 volts. All went nicely, so I removed the bulb and connected it directly to the mains, and tested it with a some amps load. Everything worked just as expected.

Now the only issue I have to solve with this power supply is the inrush current. It is so large that I have replaced several times the on/off switch. Any ideas?


  1. Hello, Put a smaller bulb (or resistor) in series with the primary being (the bulb or resistor) short circuited with a relay triggered from the DC output, the relay delay time should be enough.

    73, de Ricardo, CT2GQV

    1. Yes, it is a good idea. The problem is the power supply goes from 4 up to 15 volts, so a really low voltage relay should be used and maybe add a current limiting for the coil.

      Another posibility is to use the internal controller voltage, the one used to light the voltmeter and ampmeter. It should work...

      Thanks for the idea!