TPS40170: HDRV MOSFET Over Current

Bill,

The V62/13607-01XE is the enhanced version of the TPS40170. See TPS40170-EP. These chips should behave the same when it comes to general operation.

The first step would be to swap the ICs from a working board to an over-currenting board and vice versa. That will give us a better idea if it's really the fault of the IC.

The gate capacitor should be used if CGS<CGD. Add a capacitor to increase the effective CGS to be greater than CGD.

-Sam

Sam,

We're planning to swap the other IC onto the board with the issue, but we've also noticed the gate voltage for the high side MOSFET doesn't look like a square wave it looks more like a sine wave.  I didn't take the differential when taking this measurement, but I assume it should look more like a square wave.  The board appears to work for some time with this issue without over currenting, but after a few minutes of being on it starts to over current, and then after that it over currents every time it's powered on. Once the chip start to over current it looks like the high side MOSFET tries to turn off, but when the low side MOSFET turns on it drives the high side gate to turn on also.  I think some damage has happened to the chip at this point.  Any thoughts on what would cause this?  There is one change to the layout and that's the ILIM resistor use to be just one resistor, but was changed to three, which means more parasitic inductance from the traces.  Would this have an effect over current protection?  

Thank you,

Bill 

Bill,

The gate drive waveform would explain why the FET is getting hot. What FET are you using? Check to make sure everything is making proper contact to try to narrow down the cause of this undesired waveform.

Have you tried adding low-side gate capacitance? Has it resolved this issue? If not, we might not be seeing shoot-through. You can also check the SW node. If you get a rise, quick drop, then rise again, you're probably seeing shoot-through.

Please share the schematic and layout and I'll take a look regarding the inductance from the ILIM resistors.

-Sam

Sam,

I can't share the design with you, but I can tell you that it similar to Figure 35. Typical Design Application in the TPS40170 datasheet.  The MOSFETs being used are IRF5NJ540.  We haven't tried adding low-side gate capacitance yet.

-Bill

Bill,

It will be tough to diagnose without seeing the schematic and layout. I've sent a connection request so you can send the schematic/layout in a private message if that's okay with you. Until then, all I can recommend from here is to try the suggestions I gave. Let me know how those go and we'll see where we are.

-Sam

Sam,

We built up a bare board with just the power supply we have been talking about, using all the parts from the original working board, except we used a TPS40170 IC from the batch for the newer board.  So I don't think it's the TPS40170 IC or board layout.  I also used a 5k Ohm resistor as the load with all the same filtering capacitors after the inductor.  I don't think we're having a shoot-through incident as the low-side and high-side MOSFET look to be switching correctly and the SW pin is switching like a square wave.  There is some noise on the output voltage line which look like spikes and are about 110mV above and below the output volt.  There are some other spikes which are higher, but they are not as common.  Could there be something wrong with some of the components?

-Bill     

Bill,

Are you still having issues with the high-side getting hot? Or is the issue now with VOUT noise?

If your VOUT noise is ringing in the hundreds of MHz, that's probably SW ringing passing through the inter-winding capacitance of the inductor. This can be reduced by reducing the SW ringing (smaller loop on the power stage from CIN+ --> HS_FET --> LS_FET --> CIN-) and adding a small ceramic capacitor on the output with a small loop path back to CIN-. You may also be seeing more noise than is real if you're using a scope probe with an alligator clip. Use the tip-and-barrel method or a differential probe for more accurate measurements.

-Sam

Sam,

I didn't leave the board powered for very long.  I basically had most of the measurement equipment setup before powering the chip, took the measurements real quick, and turned it off.  I could leave it on longer to see if the same issue happens, but I would prefer to collect data before it appears to get damaged.  

Does it matter that the output current is not the same as I'm using a much higher load resistance and the output current is in mAs where the circuit with the issue is a couple of amps?  Also, I'm using all the components from a working board, except for a IC which is the enhanced version from a failing board, and the design is on a failing printed circuit board layout.  Is it possible for the feedback portion to get damaged if the right ESR/capacitor isn't used, such as C9 in the example layout?

I am using a probe with an alligator clip that might be the issue.

-Bill

Bill,

The high-side FET overheat issue is probably due to the bad gate drive signal. This heat will be proportional to the output current. This is probably causing the damage so we just need to find out what's causing this bad signal. You can probe the high-side gate at different test conditions (different IOUT, different VIN) to see if you can see a pattern of when the signal looks good and bad.

The output cap shouldn't cause issues as long as you're following the guidelines in the datasheet.

-Sam

Sam,

The MOSFET over current is due to the bad gate driver.  The gate driver doesn't start off having this issue it happens overtime after the chip has been left on.  Also, the gate signal for the high-side MOSFET looks more like a sine wave than a square wave for a chip that is going to get damaged.  I'm asking what could cause damage to the high-side gate driver and what could cause the high-side gate driver to look like a sine wave and not a square wave?  I'm trying to get more information to check if the design guidelines are correct.

-Bill 

Bill,

"Also, the gate signal for the high-side MOSFET looks more like a sine wave than a square wave for a chip that is going to get damaged."

If the waveform looks like a sine wave before it gets damaged, the sine waveform is probably not caused by damage.

Damage to the high-side driver can be caused by exceeding abs. max specs in the datasheet. It might also be caused by improper boot capacitance/connection or improper VBP capacitance/connection.

-Sam

Sam,

Ok, good to know.  We are using 0.1uF capacitor there, but when I did the calculation I got a 0.22uF cap.  So I probably caught what's happening to the high-side gate driver.  I powered the board several times for only a few seconds with a IC from a bad board, but all the parts from a good board and never noticed anything out of the ordinary.  I swapped out the C9 filtering capacitors for ones used from a failing board and powered the board three times.  The first time everything looked normal, but I had spike on the power supply out put which happened on every fourth switch from the low to high-side MOSFET.  I though it was odd so I checked it again and the high-side MOSFET was switching from 40V to zero volts.  I checked this again just to make sure I was on the gate pin and got the same result.  Could a change to the output capacitance/ESR effect the high-side MOSFET? It had been doing something more reasonable all the other times I had checked it, probably around an 8 volt switch.

-Bill

Bill,

I don't have your schematic so I'm not sure what C9 is referring to. Please share your schematic privately through a private message so I have a better idea of what we're referring to. Please also share your layout.

The gate pin was switching from 40V to GND? Is this true for all boards?

"probably around an 8 volt switch." The switch node was 8V or the gate was 8V above SW?

Any reasonable value of output capacitance should not cause damage to the high-side FET. Unreasonable values would cause different issues which would be more apparent.

-Sam

Marshall,

No, my issue isn't resolved.  I'm talking to Sam through email to solve this issue.

Thank you,

Bill