UCC2895: UCC2895 VCC pin burn

Hello Darren

I reviewed your schematic and it looks correct.

The controller is probably being damaged by a voltage over stress. The most likely thing is that the VCC1 has a large voltage spike at turn-on but you didn't include the part of the schematic showing how VCC1 is generated. It is also possible that the controller is being damaged because VCC1 collapses more rapidly than the voltage at one or more of the other device pins as the system is powered down, but of course you won't notice this until you attempt to re-start.

The first thing to do would be to try using a small external DC PSU to deliver VCC1. This would help identify if the VCC1 supply is responsible.

After doing that it would be worthwhile to look at the voltages on the other pins of the device at start-up and shut-down.

Regards

Colin

Hi colin
we do it according to the method your provide, Vcc also still occur short to ground, pls help check whether has other possibility will cause it happen burn and give more suggestion how to solve it，tks！


Referring to TI's method, VCC will still be shorted to ground.

Hi Darren

OK - so we can assume that the VCC rail is not the cause of the problem.

I would also check that the connection between the GND and PGND pins is short - if you are in doubt about this then I'd simply add a short wire link between pins 5 and 16

Please check that the capacitor on the REF pin is between 0.1uF and 1uF - this is needed for stability purposes. The data sheet also specifies that the capacitance at the VCC pin be at least 10 times that at the REF pin. I'm not sure how this could cause the problem you are seeing but you should double check this.

The next thing I would do is to monitor each of the pins of the controller and compare it to VCC during start-up and power down. That means you can look at 3 pins and VCC at the same time for voltage spikes. Even very short overvoltages can damage the controller so be sure to look for events as short as 1us as well as for longer events. The device AbsMax ratings are relative to GND and to VCC so be sure to look for negative going events as well as for positive going events.

One other possibility is that you are getting a voltage flashover from some part of the power stage. This could happen because of some PCB tracks being too close together or running under some metal at a high voltage. One example I came across was a track on the top side of the PCB which ran directly under a grounded heatsink - the solder mask broke down and shorted the track.

It is usually possible to run the power stage at a much lower voltage than its specification (eg 40V instead of 400V). The output voltage will be too low of course and the controller will operate at Dmax but you can see if the controller survives.

Regards
Colin

Hi Tengfeng

sorry delay feedback for you, beause we and customer are Celebrating the chinese new year now, when the customer is on work we will  check the suggestions you provide whether is good to solve this issue,tks!

Hi colin
According to your suggestion, the current test IC is not damaged, but the MOS tube temperature is high, pls help check it and give some suggestions, tks!

Hello Darren

Please confirm that the original problem where the controller went short circuit at startup is now fixed ?

If the MOS temperature is high then there are several possible causes, I'm going to assume that you are using Silicon based MOSFETs

Please double check the actual temperatures - MOSFETs are often specified for junction temmperatures up to 150C 

You should already be able to estimate the power lost in the MOSFETs and the expected thermal resistance of the MOSFET/Heatsink assembly - are the MOSFETs significantly hotter than you expect

Are the MOSFETs achieving ZVS ? - the easiest way to do this is to look at the gate drive waveform. If ZVS is lost then you will see a Miller Plateau on the gate waveform as the Vds reduces. There are some waveforms showing this in the UCC28950 data sheet - Fig 59 and Fig 61 - I know that this DS is for a different Phase Shift Full Bridge controller but the Vgs waveforms for ZVS / non-ZVS will be the same.

The gate drive is not strong enough - normally a MOSFET needs a gate drive of about 10V to ensure it stays fully on. Please check against the MOSFET manufacturer's recommendation

The gate should be turned on quickly - a gate resistor is used to control the speed of turn-on - normally a 10 Ohm resistor is used. I would also measure the resistance - sometimes the resistance of small gate drive resistors can increase

It would be worthwhile double checking that the MOSFET timing is correct, especially between the QA/QB and QC/QD pairs. If QA turns on before QB has turned off then you will get some shoot-through current which will cause extra dissipation and temperature rise. Please note that even if the switching pattern (OUTA,OUTB,OUTC, OUTD) is correct at the UCC3895 controller there can be different propagation delays in the path between the TI controller and the MOSFET.

Are the MOSFETs correctly fitted to their heatsink ?

Are all the MOSFETs getting too hot or just some of them ?

Please let me know how you get on - I would be happy to review your schematic - if you don't want to post it in this public forum then you can send it to me directly at colingillmor@ti.com

Regards
Colin