Hello,
I am a service technician and have had a number of products come in for repair which use the LM5575 converter. The fault is caused by the circulating diode going short circuit (3A Schottky, 60 volts reverse voltage).
Product 1 has 3 x LM5575MH converters with outputs of 3.3V, 5V, and 12V from a 48V nominal input - failure rate of this has been moderate
Product 2 has 3 x LM5575MH converters with outputs of 3.3V, 5V, and 12V from a 56V nominal input - failure rate of this has been high
The rail which fails is random in each power supply, I have had one instance where two diodes have failed in one unit.
It has been the directive to replace the 60V diode with one which has a reverse voltage of 100V. The LM5575MH datasheet specifies allowing an overhead for this parameter without stating a percentage of overhead. The reference design uses a 100V diode for a 75V maximum input voltage - 25% overhead in that case.
The advise I have been given is to replace the diodes only (even if shorted) and then if the supplies measure OK, then to put them back into service. If the supply doesn't work, then to replace the whole supply. I have a had a few modules where the switch mosfet in the LM5575 has been damaged, and they obviously were rejected.
The assumption is that service technicians are unable to perform the rework required to replace the LM5575 reliably, but this is not a problem in my case - I have no problem with this operation.
Question 1 - should the LM5575MH be replaced if it has been run into a shorted circulating diode for an unknown length of time, and an unknown number of power cycles?
Question 2 - what is the recommended percentage overhead for the reverse voltage of the circulating diode for a given input voltage?
My goal is to ensure long term reliability of the product, and prevent future failures.
Thank you,
Colin Waterhouse.