Problem with LM3150

Hello,

If you can decouple the ground from the circuit and the oscilloscope, measure the voltage across the high side MOSFET (with the oscilloscope ground connected to the drain of the MOSFET). This will help you to figure out whether conduction or switchting losses are dominant.

Do you have sufficient supply voltage (as undervoltage dropout can be below 5V)?

When conduction loss is not dominant, you may parallel M2 with a schottky rectifier as reverse recovery times of many MOSFETs are too long for operation in the 300 kHz range. With a slow freewheel diode, M1 is subjected to high peak power dissipation. Using the schottky rectifier avoids that the body diode inside M2 will conduct.

Best regards,

Wim PA3DJS

Post your schematic?

John

I am concerned that the gate/source voltage of Q2 is too low. However it is governed by the LM3150.

Could the fet be fried? Maybe add a small series gate resistor and see if there's any gate current?

John

Thanks for the responses everyone. Looks like I have enough to do on Monday morning.

After looking at that chip lay out, it appears that all you're going to get is no more than 6 Volts to be driven to the FET on the high side. Maybe a different choice of fet?

You could perform a bypass via a diode from a higher supply voltage to the BST pin. Essentially, you'd be able to supply the driver and Cbst cap with more voltage than what the rest of the chip has.

A basic SI diode with cathode to BST and anode from a higher source. maybe the Vin pin could be used as the source.. It would then be a simply by pass ..

The internal diode that resides there now, will block the voltage from back tracking..

Just a thought.

to

Correct. The high-side drive is bootstrapped by (Vcc - a diode drop). Since Vcc is internally set to +6v, Vgs(hi side) will be ~5v, max.

Yes.

Not allowed. V(BST to SW) must be under 7v, absolute max.

-- Cheers, James Arthur

You are correct. The bootstrap supply is a regulated 5V. You must use mosfets designed for this kind of drive level. The FDP060AN08 is not.

RL

...a 9V gate drive would guarantee a decent Rds(on) with a cold (-55C) FET or a hot (+125C) FET.

Hello,

I am afraid, Legg and James are right; the gate drive voltage of the LM3150 isn't compatible with your MOSFETs.

I overlooked the fixed 6 V regulator for generating the boost voltage. This limits your MOSFET selection to "low gate drive" devices only.

Regarding the body diode, it must be very fast (or bypassed with a schottky rectifier). I have some bad experience with similar type of step down converters (even with evaluation boards directly from the manufacturer). They broke down under short circuit condition. Current glitches (because of diode reverse recovery) were responsible.

Best regards,

Wim PA3DJS

I don't think its a good idea to depend on the body diode for protection of the device. I prefer to use a schottky diode in its place.

Hello Jamie,

I forgot to mention that I ment the diode inside the low-side switch (the active rectifier).

Just before the high-side switch turns on, all load current flows through the body diode of the lower side switch. Depending on how hard the high-side switch is driven, huge current may flow through the body diode of the low-side switch (in reverse direction).

Many modern MOSFETS have negative temperature coefficient in there Id versus Vgs characteristic up to high current (so I think they are not as robust as the older ones).

Best regards,

Wim PA3DJS

Following the suggestions here we've changed the FET to a logic level one, SPP80N06S2L-07 which was the only one of the right characterisitcs that farnell had in stock.

We have also added a schottky in parallel with the lower transistor.

The switching FET still gets too hot, dissipating around 3 Watts. The gate drive rise and fall times are too slow. We have added a schottky diode and transistor to the gate drive to speed up switch off which helps.

So far it looks like the LM3150 doesn't have the drive capability to drive FETs directly.

Hello,

Did you do hand calculations regarding the switching loss based on your measurements (to make sure this is the problem)?

Your MOSFET needs about 6 times more gate charge then the one used by National (see page 15 chapter 7 for details about gate charge limitation versus frequency), so maybe you are right about the gate drive capability of the LM3150 in your application.

I hope you will find the right answer why your circuit behaves differently then expected.

Best regards,

Wim PA3DJS

Cheers. You made me look at the right part of the spec.

The problem is gate charge. The LM3150 needs a lower Qgd FET than the one I'm using.

Just ordered something better from Digikey.