UCC27211 high side damaged

Ondrej Palacky

Other Parts Discussed in Thread: UCC27211

Hi,

we have encountered the following issue:

the UCC27211 gets damaged from time to time when driving a 3-phase BLDC with a MOSFET bridge (f = 20kHz, Ciss = 11nF, Vbridge_supply = 28V, no caps on HI or LI, 3k9 pulldown on HI, bootstrap cap = 100n, gate resistors = 18R). The UCC27211 is driven directly from an MCU. We are not sure of the exact conditions but suspect the motor brake activation might be a factor (MCU signals driving HIs get disconnected by MOSFETs and the LI inputs are driven alternately HIGH/LOW.

After a UCC27211 failure occurred (MOSFETs remained intact!), we tested the driver statically (motor disconnected, supply = 10V) with these results:

- when driving the LI high (3.3V), LO is at 10V, i.e. OK, when driving the HI high, HO is at 9.35V, i.e. damaged

- In both cases HB voltage = HS = about 9.6V

Does this failure mode indicate any typical damage mechanism? Has anyone seen anything similar? The strange thing is the bridge MOSFETs didn't get damaged.

Thanks

Ondrej

over 11 years ago

0 Peter Meaney over 11 years ago

TI__Mastermind 31580 points

Hi Ondrej,

Do you have any voltage waveforms of the HS node of the driver during motor brake activation.

Regards

Peter

0 Ondrej Palacky over 11 years ago in reply to Peter Meaney

Prodigy 70 points

Hi Peter,

thanks for the quick response. I have attached two waveform pics of all the phase voltages:

- brake_5000rpm.png: the motor is running at 5000rpm, then the brake is activated (controlled by the bottom signal)

- brake_1000rpm.png: same as above but at 1000rpm

Ondrej

phase_waveforms.zip

0 Peter Meaney over 11 years ago in reply to Peter Meaney

TI__Mastermind 31580 points

Hi Ondrej,

Its difficult to pick out the voltage levels at this time scale. Can we zoom in so the the screen show 2 to3 switching cycles on the screen. It is strange that the HS driver is being damaged even though the HS fet is driven low during this period of activity.

Regards

Peter

0 Ondrej Palacky over 10 years ago in reply to Peter Meaney

Prodigy 70 points

Hi Peter,

in the mean time we have checked the waveforms in greater detail and have seen peak. They are (probably) caused by layout and when we put zener diodes between the G and S of the MOSFETs, the waveforms improve significantly. We will have to gather more hours' operation to reach a definitive conclusion but for now the problem seems solved.

Ondrej

0 Peter Meaney over 10 years ago in reply to Ondrej Palacky

TI__Mastermind 31580 points

Hi Ondrej,

Thats good news. Let me know if I can offer any more help.

Regards

Peter

Power management

Power management forum

UCC27211 high side damaged