problem with 5V0.4A AC/DC power converter design with LNK623
Hi PI engineers,
I recently designed a 240VAC - 5VDC power converter below 3W. I've made couple of dozens prototypes without problem. when it comes to production. manufacture have made 100 and reported nearly half failures with output fluctuating between 10V and 0V. all have the same issue.
after I received the faulty boards, I did the following test:
1, checked the output waveform looks like LNK623 entered an auto start mode, but the period is around 1.5s instead of 2.5s (according to datasheet). I attached the screen shot of the o'scope for your reference.
2, swap the LNK623 chips between the good board and faulty board. good boards became faulty, faulty boards became good. this gave me the feeling that something is wrong with the LNK623 chip on the original faulty board. I will address that LNK623 as "faulty" LNK623 in later test
3, made changes on boards with "faulty" LNK623, change R101 from 330K all the way down to 36K, it began to work. but R102 getting hot after a while. did the same change on boards with "good" LNK623, it works as normal without generating extra heat. I attached the schematic as well.
could you please help to analyze what can cause this problem? I appreciate if you can give me some ideas or suggestion.
Thanks,
Scholes
Comments
Hi scholes,
Could you also share us some information regarding the transformer for good and bad unit?
Regards,
Jedidiah
Hi Jedidiah,
Thanks for your reply. the same transformer has been used in good and bad unit. I've attached the spec of the transformer.
Hi scholes,
I would like to understand the difference in the good and bad transformer used. Like the actual leakage inductance and the primary inductance. Based from your data, you said that changing the snubber value R101 solves the problem. Decreasing R101 value decrease the ringing on the Drain to Source voltage and the leakage inductance of the transformer is one of the factor for those ringing. Can you send us also waveform of your Drain to Source voltage of the good and bad unit?
Regards,
Jedidiah
Hi Jedidiah,
thanks for your reply.
the same transformer has been used on all units.
it's a bit confused when we address them as good or bad unit. actually, the fault travels with the chip, not the unit. it's clearer to say the good or bad chip.
according to my test, good chips work with all units with the original design which R101=330K, R102=300, R104=8.25K, R105=3.3K.
bad chips will enter a look-like auto start status, which the output fluctuates between 0-10V as show in the picture attached.
two methods of modification can make the output stable:
1, reduce R101 to 36K, R102 to 200, but R102 got heated pretty quick.
2, increase R105 to 3.9K, but the output will be around 4.5V. manufacture guys reported that this method could only work for some bad chips, not all.
I suspect the bad chips got damaged in some degree, the feedback threshold might have changed. I checked the datasheet, the Vfbth is around 1.86V, any method I can verify that value?
Could you please advise what can cause the damage to the chips? what's the behavioural action would be if the chips were damaged?
I don't have the drain to source waveform so far.
Thanks!
Hi scholes,
If you are changing the snubber values and the result is that the good IC can now regulate at 5V, I think the problem here is the ringing on the Drain to Source voltage. If you could share with us more waveforms like Drain to Source, the waveform on the winding used for FB pin and the waveform across FB will be better. The ringing on the primary winding can be reflected on the winding used by the FB pin. In case you want to know if the IC is really damaged, you may ask some help from your IC supplier whoever the distributor is from their Field Engineer if they can help you check the parameters of the IC or other parameters needed to have more details on the problem.
Regards,
Jedidiah
attached the waveform and schematic.
Thanks!