Electronics > Projects, Designs, and Technical Stuff
Synchronous Buck IC with Internal Power Switches - failing
(1/2) > >>
Systemtek:
Hi,

So I'm working on a switching regulator using the LTC4162. It is a synchronous buck converter IC by linear/analog and it keeps failing and turning into a resistor of a few ohms after running nice and cool for a few minutes with a constant load. My guess is that ringing on the switch node is causing the primary switch to be pushed past it's Vds breakdown. I haven't rebuilt the circuit since I came up with this idea so I can't check the current through the inductor. If it turned out that the inductor did not have a current passing from the switch node I suppose the next possibility is that the low side switch has failed.

I cannot find the breakdown voltage of the internal power switched in the datasheet https://www.mouser.co.uk/datasheet/2/609/LTC4162-L-1398190.pdf. The Absolute maximum ratings listed on page 3 and the electrical characteristics starting on page 4 are pretty light on info it seems for such a highly rated manufacturer. (annoying this sheet gives lots of general switching regulator design advice and of course masses of info about the digital telemetry system but does not specify the anything more than the Rds on for the power switches).

This means I cannot make a spice model of the internal power switches because one important parameter for simulating the ringing on the switch node when the primary/upper power switch/FET turns on is the output capacitance of the lower/secondary/rectification FET.

have been running the circuit with an Vin of 35v only one volt away from the Absolute max but this is clearly how it's designed to run and I check the bench supply using a meter.

Also I have done my best to clean the flux after soldering this WQFN package with out ultrasonic cleaning. I did not manage to specify the solder mask expansion as I wanted it so there is not solder mask between the pads for the IC pins but visual inspections and R measurement said no bridges.

I appreciate there are probably a lot of factors that could go into a failure mode like this but I thought I'd try to outline some of parameters of the design and make some of my own guesses about whats happening and see if more experienced people can offer at least some suggestions of what to try. This is my first serious switching converter design, speak slowly :)

BTW these ICs are pretty expensive for a hobbyist so I'm taking it easy and trying to get some good ideas of what might be causing this before ordering more and stressing myself out trying to solder the bl*?dy things.

Many thanks,

Han
Systemtek:
Here is a model of my design.
Jay_Diddy_B:
Hi Han,

Operating a 36V ABSmax part at 35V  is asking for trouble. Normal (industry) design margins are at least 25% so 24V nominal is probably the maximum input voltage that should be used.

Also read AN-88

Link: https://www.analog.com/media/en/technical-documentation/application-notes/an88f.pdf

This applies to the input and the battery connection in the case of a battery charger.

Regards,
Jay_Diddy_B
ConKbot:

--- Quote from: Systemtek on November 24, 2019, 04:55:31 pm ---

have been running the circuit with an Vin of 35v only one volt away from the Absolute max but this is clearly how it's designed to run and I check the bench supply using a meter.

--- End quote ---

This combined with no electrolytics onboard makes me wary of the input ringing up above the max input voltage.  The wires to the power supply can have quite a few uH of inductance. This also introduces the hazard of turn on transients ringing up very high, but using a bench supply with a controlled turn-on should negate this.  Unless your bench supply just clicks on an output relay, in which case, not so much.

Electrolytic caps provide the "bulk" decoupling which reduces change in current during the switching cycle, as seen by the bench supply and test leads, and also provide damping to the input LC network though their ESR.

If you want to simulate something, stimulate the power supply output(few tens of miliohm ESR), leads (resistance and inductance), input capacitors (remember voltage coefficient of ceramics, even a 1206 body cap will have significant capacity loss at 35V)  and put a switched load on your input network.

Systemtek:
Thanks for your replies.

So lets say I can design an input filter that is optimized for stopping the input voltage ringing/overshooting. Is it still the case that when charging the 8 Li-ion cells it is designed for it might just keep failing?

The fully charged voltage of the stack is 33.6V it doesn't allow me to drop the input voltage much lower at all. There also isn't enough of a margin there to use a TVS diode that wouldn't have too high of a leakage current as far as I can tell. 

I can't find another 8 cell solution as good and feature rich as this one so I'd hate to have to give it up.
Navigation
Message Index
Next page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod