UCC28070: MOSFET broken when DC power jump or DC output short circuit

Hi

Do you have a schematic , bill of material or any test waveforms that you can post?

If you use the calculator tool for this part you will be able to control the amount of power delivered to the load and protect the Mosfet

Regards

John

Thanks for help, I do calculate by the excel sheet from TI, the Vdc is 600V and input is 200Vac. the power limiting is 8500W.

because failure happened unexpected, so no failure waveform, the normal waveform is perfect. 

I am working on to reduce the Dmax, reduce the current peak limit value, and then do the short circuit test.

is there any reference document to describe when Boost PFC UCC28070 is shorten at output?

Thanks again 

Hi,

If the normal operating waveforms are acceptable then something is happening at startup.

I suggest that you look at the inrush protection and soft start of your circuit.
8500W is a significant amount of power and you will need some circuit to slowly pre charge the output capacitors. During this time the bias supply to the pfc controller is in its UVLO state so that UCC28070 is disabled.
You can extend the soft start time and check whether this helps.

Check that the boost inductors have the current rating for the peak current that is expected.
Check that the RIMO resistor is the correct value.
Check that the current transformers are a high frequency type and can handle the required volt*secs expected.

Test at a power level below 8500W and see how the startup waveforms look like.
Look at the bias supply on the UCC28070 and the gate drive waveforms.
It is important to get some waveforms to check for abnormalities in the switching waveforms.

600V is a high output voltage.
Are you using IGBT's ?
If so be aware that the lower frequency limit of the UCC28070 is 30kHz.
This is probably a little on the high side for an 8500W power supply with IGBT's
You may wish to consider using the UCC28070A which has a lower limit of 10kHz

Regards

John

Hi,
My previous reply was based on the assumption that the failure occurred when you powered up with an 8500W load.
But I'm not sure now what test you performed.
Your second reply seems to say you short circuited the pfc output ?
Is this correct ?

Regards
John

Hi John,

I may not describe clearly, the system is following:

200Vac single phase --> PFC (590V/8500W)---> VFD(variable frequency drive)-->Motor.

we never start up PFC with load, because VFD will not run until it get DC power when PFV already into noload stable stage.

I use four MOSFET (ON:FCH060N80-155), 50kHz switching frequency , with current transformer fromTRIAD CST206-2A, 80ARMS/200turns up to 200kHz,  the Boost inductor is 400uH to 110uH, with peak current of 50A, the calculated and tested current is 42A peak (each interleaved branch).

we test it by load bank at rated power output until thermal stable, it is good.

the failure happened when PFC is running with no load and then switch on the load bank with 8kW setting, and then MOSFET failed.

another day, the VFD is damaged (by some other cause) which short the DC bus which is the output of PFC, and then MOSFET failed, this two things should be similar.

means, when DC output drop a lot (say below the 2.8V) another 100uA is added to current reference, then push the duty cycle to the limit (which is 90% right now) then over stress the MOSFET?

I am doing this:

1 try to decrease the Dmax to 80%, but current waveform has glitch from 6kW output power, the PF is still ok (>0.98), and then back to 90%.

2 reduce the output voltage to 540V, hope can mitigate the stress of MOSFET a little bit.

3 set the current limit as low as possible based on the test result.

any comment or suggestion to this, or where should I look at? again, what happen as a sound PFC design UCC28070 when DC output is short circuit?

thanks

Shusun

I did following:

Peak current limit to 2.17V ( 12K/6.8K to 6V reference), and the current sampling resistance is 11.4Ohm, means limit current to 38A (200 transformer turn ratio), which is the tested maximum current when DC output is 7700W( the max power for coming project).

the Dmax is still 90%, for 50Khz switching, it is 18uS max pulse duration, so the max ET (V*uS) is 39VuS, the current transformer has over 1000VuS capability, is this calculation of ET correct?

so, is that means, what ever happened, the MOSFET peak current will be limited to 38A ( for two 58A MOSFET in parallel), even though peak current and Dmax appear at same time, the current transformer still within it linear area, wouldn't saturated?

if the DC output of the PFC is short, the peak current of MOSFET is limited to 38A which just a little bit higher than normal running, until over current trip the AC fast acting fuse, then MOSFET should survive, is this the case?

Thanks

Hi Shusun,

The UCC28070 has a pfc enable function that will shut down the gate drive when the VSENSE pin drops below 0.6V

So you should check that VSENSE is less than 0.6V during this test. Perhaps you have too much filtering capacitance on this pin.
This is the only protection feature that will shut down the pfc for this test . It is normally used as an open loop protective feature and I'm not sure how fast VSENSE will drop from its steady state regulated value (3V) to the disable value (0.6V)

Short circuit protection is not a normal requirement for a boost pfc because there is a direct path between the power source through the inductor to the load.
This really makes the current uncontrollable and you are really depending on a fast acting fuse for protection .

You may need to design a discrete circuit that rapidly detects an overcurrent (rather than a short circuit) on the output and use this circuit to pull VSENSE below 0.6V.
This will cause a soft start of the UCC28070 and protect the pfc.

Regards

John

Hi John,

Understood, I have OVP/UVP for input AC voltage to pull SS pin to zero, if I need add output OCP, I can parallel this protection function to SS pin, rather than pull Vsense lower than 0.6V, does this work?

Thanks

Shusun

Hi shusun,

 Did you make a simulation? how about consider the the differnet pins in ucc28070 model from ti,including IRIPX,F_KHZ ? Now, it make me puzzled.Looking forward to your reply.thank you!