Electronics > Projects, Designs, and Technical Stuff
Buck-Boost controller fails but power switch survives
(1/1)
luigi:
Hi everyone,
I am using the buck-boost controller LM25118, to generate 12V DC (7.5A max.) out from a 10-30Vin battery.
This power supply feeds a H-Bridge that drives then a DC motor which will demand 7.5A in stall at 12V.
The problem is that I am having many failures where the power supply chip fails without apparent reason.
The failure consists of the LM25118 stop working and no voltage is measured in the output. There is no PWM in the gate of the MOSFETs.
Replacing the chip (and only the chip) makes the circuit come back to normal and operate normally.
Initially I considered that the failure could be caused by defective IC. But after having multiple boards from different PCB suppliers failing, I ruled that out.
Sometimes at motor start-up the Voltage between the H-Bridge and the power supply drops down to Zero as there is little capacitance at the power supply output. But the power supply output seems to ramp-up back to the nominal voltage normally and will not enter any short circuit protection mode.
After discarding IC malfunction, I thought that the overload condition could be the reason behind the failures, but in such case the failure point would be the power diodes or MOSFETs, not the controller.
The only possibility left in my mind is a transient in the VIN that supplies the chip.
The schematic for the LM25118 circuit is below:
The design was generated by TI's Webench online tool, but I have made some changes.
I've reduced the main inductor to one with Isat of 11A.
In what circumstances, could a buck-boost controller with the topology as in the schematic above, fail while the power switches remain intact?
georges80:
Long wires from the power supply to your switcher can cause inductive transients, especially since it appears you only have ceramics as input caps to your switcher.
Add some higher ESR electrolytic caps (say in the 100uF range) in parallel with your C20 cap. If it's input voltage transients killing the switcher, the the electro cap can 'dampen' them.
You could probe with a scope (use a spring 'ground' clip on the probe end) and likely see some nasty high voltage transients when you connect/disconnect your power supply.
Anyhow, this also assumes you have a decent layout and good ground planes, i.e. the design is otherwise good and within safe operating limits.
cheers,
george.
jbb:
It’s quite hard to say. Have you checked all of the pin ratings? They’re probably OK, but...
Also, the compensation network connections look a bit odd. Some chips are very sensitive on the feedback and compensation pins.
uer166:
Also check for negative transients on switch node (HS pin of IC), it's rated to -4V, but inductance in switching loop can make that lower.
Navigation
[0] Message Index
Go to full version