Problem with MP103

We have copied the reference design as closely as possible. The most notable difference is the transistor that is specified is generally not available in the west (RS, Digikey, Mouser). We compared the datasheets and selected FJPF5027OTU (from ON semiconductor).

If we have no load we get a regulated voltage output however as soon as we put a load of 5-7ma or higher then it runs for a few seconds and turns off for about 3 seconds before trying again.

Hi Tom:

Welcome to our MPS Community.

Thank you for using our MP103. Can you please let me know the output voltage and current of your system and the capacitor value on the VB pin. Also please let us know if you have used the full bridge rectifier diode at the input or half wave rectifier.


Thanks, followed the reference design as far as possible. So yes full bridge rectifier (MB6S I think as per Ref).

The VB pin has two caps C1 is 470 uf electrolytic. We have tried two, lower quality and a high quality with low ESR no dif. The cap between DR and VB we have tried 4.7uf and 10 uf.

We have also replaced the tran and MP103 no luck.

We have also noted it takes between 5 and 10 seconds for it to start up. Which does not seem to be consistent with DS and RefDesign.

(Our target output voltage is between 5.0. and 5.4v we have turned it up and down. What was odd from the beginning is that we got 6V out on a voltage calculated with the formula in the RefDes. )

On the current front we will take as much as it can supply realistically. Around 100ma plus will great. As mentioned with dummy loads above 5ma it turns on for about 2.5 seconds before turning off.

Hi Tom:
Based on the fact that it takes 2.5 seconds before turning off, it seems that the part may be reaching the thermal shutdown limit. Underneath the MP103 is an exposed pad that needs to thermally conduct to the PCB to keep itself cool. On the page 15 of the datasheet we have some recommendations for the layout. We recommend to connect the exposed pad with GND to a large copper area for thermal performance and reliability. I would recommend a pad of at east 120 square mm area. On page 16, we have the sample layout recommendation too. Also if your line voltage is 230V then at 5V Output voltage, the part can deliver 350mW of output power which is 70mA of load current.


Hi Tom:

If you need 100mA output current the other option is to use the MP172A in CCM. The MP172A datasheet has an application schematic for 5V at 120mA on page 17 of the DS.


Thanks, we have connected it and it has much copper around it checked the NTC and its voltage remains above the threshold. Will check it again (as we replaced the chip by hand)

We have noticed on the scope that the DR pin has a pulse every 10ms once it turns on. It starts at 25v and gradually drops. If you monitor the VB pin the voltage drops until it reaches about 10v before cutting out and this at 7ma current draw.

Hi Tom:

The thermal conductivity is best when the MP103 is assemble with solder paste in a re-flow oven. So it is critical that thermal conductivity is not compromised.
Can you please check if the VB pin reaches 15.25V before the 5V Output turns ON. Also for now, please try removing the cap between DR and VB pin as it slows down the switching process of the Q1 transistor. If that does not work, can you please send me the schematic?


That transistor has low current gain. Just some guy on the internet here but I would look for a transistor with higher Beta. It doesn’t need to switch quickly, it doesn’t need huge safe area as all the switching action occurs with low voltage across the transistor.

PBHV8560Z for example has a gain of 70 vs 10 for the transistor you grabbed.

Hi Tom:

The external bipolar transistor behaves like a switched Linear regulator that is switched ON and OFF every 10msecs. But while it is switched ON it behaves like a linear regulator that needs to charge the cap on the VB pin. Like any linear regulator it needs to have enough thermal mass. That is why we have recommended a TO-220 package transistor in our EVM for 5V at 60mA system load.
Since you measure 25V on the DRV pin and that is a diode drop higher than the VB pin, the Cap C1 is charged to 24.3V. At 24.3V, the 470uF cap has enough energy stored for 10msecs and should have a very small drop in the voltage if load is only 7mA at 5V output. The fact you see it drop to 10V and cut off the output is indicating some leakage path on the VB cap.

Thanks tried to match it as closely to RefDesign as the tran on it is not readily available to me and even the datasheet I found for the specified tran gives another supplier name.