I have built a battery charger using the LTC4002.
I am having problems with the P- MOSFET and diodes overheating and eventually desoldering themselves.
I basically followed the schematic from their demo board...
http://cds.linear.com/docs/en/demo-board-manual/dc551A-A.pdf
I have attached my schematic, and PCB drawings
I did add two N-Channel MOSFETS to "enable" the circuit because 'batt+' can see up to 60v. **The N-FETs do not overheat.
One of my chargers on my circuit does not contain the N FETs section and it experiences the same overheating problem.
I thought that maybe the gate wasn't getting pulled down enough.
My scope sucks but you can see in the pics that it appears to be working. (volts/div = 5v)
The input is 12v and draws roughly 1.5A.

I am stumped on what is causing the problem because I selected components rated for higher current (13.5 A MOSFET, 5A diode) than used in their demo board.

 

Your Schottky diodes have a 60V reverse voltage rating, whereas the demo-board ones are 30V. That may have a bearing on the overheating.
Is your circuit on a breadboard or a pcb?

 

The only thing I can see is that your chip Vcc is connected to the anode of D402. This might prevent the Mosfet from turning completely off. Try connecting Vcc to the cathode.

 

Your Schottky diodes have a 60V reverse voltage rating, whereas the demo-board ones are 30V. That may have a bearing on the overheating.
Is your circuit on a breadboard or a pcb?

Thanks for the reply. My circuit is on the PCB in the pic. I need to use the 60v diode to block up to 60v. I could try the 30v and see if it makes a difference but technically the b560 should perform better than the b330 right?

 

The only thing I can see is that your chip Vcc is connected to the anode of D402. This might prevent the Mosfet from turning completely off. Try connecting Vcc to the cathode.

Thanks for the reply. Crap sorry I drew it incorrectly on the schematic. But on the PCB i wired it the correct way with vcc to the cathode...

 

Here is the updated schematic. This is currently how my circuit is on the PCB. I am still not sure where the problem is coming from.

 

What exactly is the scope supposed to be showing?
What oscillation frequency do you get?

 

The scope is showing the voltage on the gate of the transistor. I initially thought the heat came from the MOSFET not turning on fully.
The gate is pulled high to 12v internally by the LTC4002 chip and down to 6v. I never checked the frequency. I will check when I get home from work. The datasheet says it should have a frequency of 500K, so i will check what it is coming out as.
The charger does work though, as in... 4.25V comes out and charges the battery at ~3A. Not sure if i mentioned it. Thanks again for the help, I'm excited to get it to work.

 

What exactly is the scope supposed to be showing?
What oscillation frequency do you get?

Attached is the switching frequency on the gate. The time divisions are 5x10^-6 seconds. My scope sucks so its hard to tell, but it looks like about 3 cycles per div... => about 600Khz... so its probably the 500K like the datasheet says. *Disregard the volts/div*

I am still leaning toward the MOSFET as the main cause of all the heat. I am thinking that it could be the gate charge and rise/fall time, leaving the the FET in the transition zone for most of the switching cycle when the resistance is higher. Which would cause it to heat up and probably transfer the heat through the traces to the other components.

I originally chose this transistor -
https://www.fairchildsemi.com/datasheets/FD/FDS6673BZ.pdf
gate charge - 88 nC

And so I tried this transistor with 8x less gate charge and it improved slightly but it eventually desolders itself as well.
http://www.nxp.com/documents/data_sheet/PMK50XP.pdf
gate charge - 10 nC

So I am still stuck... what do you guys think? Am I looking in the right direction?

 

Attached is the switching frequency on the gate. The time divisions are 5x10^-6 seconds. My scope sucks so its hard to tell, but it looks like about 3 cycles per div... => about 600Khz... so its probably the 500K like the datasheet says. *Disregard the volts/div*

I am still leaning toward the MOSFET as the main cause of all the heat. I am thinking that it could be the gate charge and rise/fall time, leaving the the FET in the transition zone for most of the switching cycle when the resistance is higher. Which would cause it to heat up and probably transfer the heat through the traces to the other components.

I originally chose this transistor -
https://www.fairchildsemi.com/datasheets/FD/FDS6673BZ.pdf
gate charge - 88 nC

And so I tried this transistor with 8x less gate charge and it improved slightly but it eventually desolders itself as well.
http://www.nxp.com/documents/data_sheet/PMK50XP.pdf
gate charge - 10 nC

So I am still stuck... what do you guys think? Am I looking in the right direction?

Look at page 16 of the datasheet. There is a slightly different version of a charging circuit. The /charge pin is better defined (automatically turning off current flow to battery when charged using the charge pin.

Your circuit is a bit off from the one on the datasheet (and there is likely an error on the datasheet). Your LED should not be connected to battery+. It should be connected to Vcc instead. I think the LED current is messing up your SENSE current.

 

It looks like the problem was the gate charge. And I needed to dissipate more heat with larger traces and thermal vias. Thanks for the help everyone.