Tpa3255 running in BD modulation at 600khz.

[Hayk]

I will be traveling around for 1 and half month, so away from experimenting. Traveling gives lot of time for reflection, and I am getting new ideas . First I don't like the internal oscillator which can have high level of jitter, hence noise. I will use the two chips both in slave mode and drive with an external low jitter oscillator.
Then, two different combinations are possible with BD modulation. Either 2X4xSE to bridge A1 with A2 and D1 with D2 so that each bridge have symmetrical pins.
Or, to have each chip in PBTL and bridge A1+C1 with A2+C2 and D1+B1 with D2+B2. I need to experiment what happens to B+D if in PBTL mode, when only A input is supplied, normally no any sound should be, but is it the case?
Another mystery in the DS concerning the power dissipation with Pout=0.
In 4XSE is 2W and PBTL or BTL is 2.5W, why?
The extra power dissipation of the chip due to load, is mainly the 2xRon of Mosfets that come in series with the load, which is 2X0.085 or 0.170 ohms. For 8 ohms load and 125w/channel, the chip needs to dissipate 2.5w/channel. For 4 ohms load and 250w/channel, it is 10w/channel. With PBTL mode and using 2 chips, the values get halved to be 2.5+5=7.5w per chip for 4 ohms 250w/channel.

 

[Hayk]

I did remark something that concerns the power dissipation.
It is in function to Ron for conducting and the switching power is in function of Drain-Source capacitance plus the body diode reverse current. The switching current is in proportion with the frequency.
The body diode reverse current occurs after the dead time when the diode continues the inductor current then the opposite mosfet switches on without waiting the reverse recovery time. The two will conduct high pulse current to evacuate the accumulated charge in the diode which is in function of current.
In TPA3110, the graph shows the efficiency to be 90% from 10w to 30w. This means the reverse recovery time is applied in the dead time so the the efficiency is not dependent upon the current.
With 3255, it is not the case. To use lesser dead time hence more linear, the dissipation is secrified which is 3 times more than the Ron at 450khz. With 600 khz it will be 50% higher that is 90w for 350w@2x4ohms
But, there is a solution. To apply a pair of Schottky diodes on each output to parallel the body diodes. These diodes have lower on voltage and very fast recovery time, but 7A only for PDS760, afterwards the body diodes take the relay. See example bellow.
By using a pair of PBTL, the total dissipation even without diodes is halved to be 45w for 350w@2x4 ohms, for each. I don't know what will be with diodes.