DRV8307: Using CSD18540Q5BT MOSFETs with DRV8307

Hi Stephen,

What type of issues are you seeing?

Some overcurrent problems. I've blown several MOSFETs trying to get this board working. Are the gate driver resistor values critical to basic function or should the values I'm currently using be pretty close?

Rick,

What I need right now is some confirmation that:
1) the schematic connections look correct
2) the DRV8307 as connected should drive these MOSFETs reasonably well

i.e. that I haven't made any gross errors in my design.

Steve

Hi Stephen,

Are you able to measure both UGH_GATE and ULS_GATE when these signals transition? the 240 Ohm UGH_GATE resistor may be causing the high-side gate to slew slowly, resulting in some amount of shoot through.

Have you tried replacing the R1 / R3 / R4 with 0 Ohm resistors?

Thanks, I'll try that next.

Phil,

I changed the 240 gate resistors to 0 ohms.  The response if much the same.

Can you please review the attached scope plots and let me know what you think.  I'm trying to understand why load current is higher than expected at relatively low speed of 1200 rpm.

Also, please provide a list of other signals you would like to see.

Steve

Motor Driver Debug Signals.pdf

I guess what I'm really trying to learn is, how to go about optimizing this design to minimize the heat generated in the MOSFETs.

Part Number: DRV8307

I need some help with my custom board design.

I have been using the DRV8307EVM board to drive a 24V motor, rated 8.6A at 6000RPM.  I need to drive this motor at 5000RPM max.

When I use the DRV8307EVM board, I can go up to about 4000 RPM with a loaded motor before reaching the limit of the MOSFETS.  For my custom board, I have tried 2 other MOSFETs with an otherwise identical design.  The problem is that my custom board generates significantly more heat in the new MOSFETs than is generated in the EVM at similar load and RPM.  I'm trying to understand why and what I can do to my custom board to minimize the heat in the MOSFETs.

On my custom board, I started off with the CSD18540Q5BT MOSFET.  I have since tried the CSD18537NQ5AT part, which has similar Gate Charge specs to the CSD88537ND part used on the EVM board.  Both of these MOSFETs are rated at much higher current than the EVM MOSFET, but generate more heat at the same motor load.

Can someone offer some suggestions for changes to my schematic or methods for optimizing the driver circuitry to minimize heat?6886.motor schematic.pdf

Hi Stephen,

Based on your attached oscilloscope plots from the first thread, I have a couple suggestions.

1. Does removing C4, C5, and C6 help with the motor performance / heat in the MOSFETs?
2. Have you been able to measure the GHx and GLx of the same phase in your layout?  This could be measured either Phase U, V, or W.  This will help determine if theres some amount of shoot through at higher speeds.
3. Does your layout provide enough copper area to help dissipate the heat generated in the MOSFETs?

Phil,

Please see the first 3 pages of the attached document for the GHx and GLx signals.  Page 4 shows the top layer of the layout.  I will try the CAP changes you suggested this morning and let you know.

Steve

0508.Motor Driver Debug Signals.pdf

I removed the three capacitors C4,C5,C6 and took some more current measurements of my custom board.  See the first 2 pages of the attached document.

1401.Motor Driver Debug Signals.pdf

Hi Stephen,

Thank you for the detailed information.  The 1.8V spike measured on the ISENSE line is not expected, this sounds like a shoot through type of event.  This should be triggering the OCP Protection on the DRV8307.

Are you able to zoom in on just one transition between the WHS_GATE signal WLS_GATE Signal?  Or monitor this switching on a smaller time scale?  If both FETs on side of the bridge are being turned on at the same time that would explain the large amount of current measured on the ISEN line.  

Also, we typically do not recommend using thermal reliefs on pads for different power components.  I see that your sense resistor as well as bulk and bypass capacitors all have thermal reliefs on their ground returns.  Using direct connections will allow for better current flow and for these components to properly dissipate heat into the surrounding ground pours.

The rise times on your Hall IC's also seems suspicious.  It looks as though it takes several ms for the signal to transition above the input high threshold, the driver may commutating the motor to slowly, limiting the maximum speed the rotor can spin at by inducing system delay.  Is it possible to remove any capacitance on this hall IC line?

Phil,

I zoomed in on one transition as requested.  Also zoomed out showing current vs the two gate signals.

Motor Driver Debug Signals2.pdf

Hi Stephen,

Thank you for the images. These gate transitions look to be free of any shoot-through, have you checked all three of the phases?

On the three high-side MOSFETs, is there any significant copper area connected to the Drain of these devices? The damaging you're seeing may just be from poor layout of the MOSFETs resulting in the devices overheating under normal operating conditions.

Have you used our EVM as a reference for correct thermal via placement and copper area widths?