UCC21750: UCC21750DWR maximum power dissipation

I just saw this here: https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/904742/ucc21750-q1-power-dissipation-inconsistency-maybe

It states that the listed dissiaption is a calculated operating value and not a maximum rating. So half of my questions is solved. The only open question is the safety limit according to point 6.8. Is it correct that the safety is no longer guaranteed if we raise the maximum supply current to 120mA at -4/+15V equall to 2.2W even if the operating junction temperature stays at 115.8°C?

Frank, 

Frank Schoelch said:

Is it possible to operate the driver above the maximum permitted power dissipation of 985mW defined in point 6.5 and the supply current of 61mA defined under point 6.8 as far as the junction temperature stays below 150°C?

Yes, but we dont recommend operating right at the limit of Tj, be sure to leave headroom and consider that ambient temperature changes, I would not suggest designing for a operating point >125C, that is just shaking hands with danger. 

Frank Schoelch said:

As an alternative, we could boost the driver output using a separate driver IC like Ixys IXDN614. The useage of bipolar output transistors as shown in the datasheet may be too slow for our application.

We seen customers using such a solution with IXYS booster, but not generally with UCC217xx.  There is a big issue: this IXYS part inverts the signal because the PMOS is on highside and NMSO on lowisde like a typical cmos inverter. 

Because the logic is now backwards, Miller clamp and Short circuit detection will not work at all with the IXYS part as a booster. If you dont need either of those featuers, fine, but there is another solution. 

Frank Schoelch said:

The useage of bipolar output transistors as shown in the datasheet may be too slow for our application.

Regards

The reason we recommend bipolar is because for the Lowside, a PNP must be used if you need to boost the pulldown current because using a FET will not allow Soft-turn off to work properly.

For pullup part of the booster either NPN or NMOSfet will work without affecting function. 

Since it seems like you are OK with the current rating, just having a concern about thermals, I might even suggest the following

* NMOS/NPN on highside connected to OUTH, and leave OUTL as normal. 

This way you reduce some of the load on, but have no impact to STO feature. 

Of course, you are also able to use a full push pull buffer structure as well. 

Let me know if this helps or if you have additional question. 

Best

Dimitri

Frank Schoelch said:

2*32,9+50=115.8°C

Also, please do not use the junction to board thermal resistance for TJMAX calculation, 

Please use AMBIENT RJTHETA, the value you are estimating is higher because of of this. 

Thanks for your answers - we will modify our current design accordingly. Two more questions:

-In which way does it have a negative impact on the safety isolation barrier in regard to the safety limits (point 6.8 in the datasheet)? If the power supplied to the UCC is limited to -4/+15V at 100mA or 1.9W, the maximum junction temperature is well below 150°C. So the saftey barrier should stay intact even with a failure inside the driver. Is this correct?

-Ixys offers the IXDN614 which is a non-inverting driver. Would it be possible to connect the miller-clamp at the output of the Ixys driver or could this cause current spikes? As an alternative, the miller clamp could be connected at the input of the Ixys or it could be left unconnected. Which method would be recommended by TI?

Regards

Frank Schoelch