TPA3251d2

TI has released TPA3251d2 'AKITA" higher power analog in class D chip with higher order modulator similar to Hypex NCore but with pre filter feedback. First heard about it at CES 3+ years ago. Can be configured as 1,2,3 or 4 channel with 36 volt supply for output stage.
 
I'm really curious how this chip sounds. Pretty impressive description:

"TPA3251D2 is a high performance class-D power amplifier that enables true premium sound quality with class-D efficiency. It features an advanced integrated feedback design and proprietary high-speed gate driver error correction (PurePath™ Ultra-HD). This technology allows ultra low distortion across the audio band and superior audio quality."

I hope the chinese manufacturers will take the first steps soon.
 

Attachments

  • Image1.png
    Image1.png
    101.9 KB · Views: 9,151
  • Image2.png
    Image2.png
    121.2 KB · Views: 8,970
Some short calculation for the required heatsink, worst case:

Rds_on (high/low side): 0.1R
Imax: 14A
Tambient: 30°C
Rth_jc(Top): 0.36°C/W

Power dissipation per channel:

Pd_channel = I²R = Imax²*2*Rds_on = 14A²*2*0.1R = 196A*0.2R = 39.2W

Power dissipation total:

Pd_total = 2*Pd_channel = 2*39.2W = 78.4W

Tcase = Tjunction - Rth_jc * Pd_Total = 125°C - 0.36°C/W * 78.4W = 125°C - 28.224°C = 96.776°C

Rth_heatsink = (Tcase - Tambient) / Pd_total = (96.776°C - 30°C) / 78.4W = 0.85°C/W

Rth_heatsink includes thermal conductivity of heatsink grease.

--------------

Lets do the calculation again for Irms (sinus power):

Rds_on (high/low side): 0.1R
Imax: 14A

Tambient: 30°C
Rth_jc(Top): 0.36°C/W

Power dissipation per channel:

Irms = Imax*sqrt(2) = 14A*0.707 = 9.9A

Pd_channel = Irms²R = Irms²*2*Rds_on = 9.9A²*2*0.1R = 98A*0.2R = 19.6W

Power dissipation total:

Pd_total = 2*Pd_channel = 2*19.6W = 39.2W

Tcase = Tjunction - Rth_jc * Pd_Total = 125°C - 0.36°C/W * 39.2W = 125°C - 14.112°C = 110.888°C

Rth_heatsink = (Tcase - Tambient) / Pd_total = (110.888°C - 30°C) / 39.2W = 2.1°C/W

Rth_heatsink includes thermal conductivity of heatsink grease.

Much better. :D

As music is normally not pure fixed frequency sine wave the maximum temperature will likely not reached even at full power.

----------------------------------------------------------------------
Lastly, for normal use, with an estimated programm factor of 0.5 there will be:

Rds_on (high/low side): 0.1R
Imax: 14A
Music programm factor (mpf): 0.5
Allowed max junction temperatur: 90% from Tj = 0.9*125°C = 113°C

Tambient: 35°C
Rth_jc(Top): 0.36°C/W

Power dissipation per channel:

Irms = Imax*sqrt(2) = 14A*0.707 = 9.9A
Iprog = Irms * mpf = 4.45A

Pd_channel = Iprog²R = Iprog²*2*Rds_on = 4.45A²*2*0.1R = 19.8A*0.2R = 3.96W

Power dissipation total:

Pd_total = 2*Pd_channel = 2*3.96W = 7.92W

Tcase = Tjunction - Rth_jc * Pd_Total = 113°C - 0.36°C/W * 7.92W = 113°C - 2.85°C = 110.15°C

Rth_heatsink = (Tcase - Tambient) / Pd_total = (110.15°C - 35°C) / 7.92W = 9.5°C/W
 
Please let me know what heatsink solution you choose, I'm still unsure. BGA heatsinks are quite expensive.
Haven't made my mind up yet.

Heatsinks are a mechanical pain in the ***. I might do something like mount the card to an aluminum plate with the TPA on the bottom side - I'd have to "swiss cheese" the plate a bit to allow clearance for component leads, I'll play around with the idea over the next few days.