TPA3245: Clipping level

Hi Akio-san,

Is the customer seeing actual clipping from the device or is this based off of the CLIP_OTW LED? And have they been able to test their system without PFFB to verify if its being caused by this part of the system?

Based on what you have described, I don't think they should be experiencing clipping due to PFFB based on the total system gain being only 10.2dB.  We have a very useful PFFB App Note that was recently released that could be of use if the customer hasn't seen it before.

It's also possible that they are experiencing possible premature clipping from having a lower PVDD than 30V.  This would cause the amplifier not able to reach as high power levels.

Regards,

Robert

Robert-san,

Thank you for your reply.
The customer observed only the output waveform by an oscilloscope. I will suggest to him CLIP_OTW terminal also to be observed.
He did not observe the clipping without PFFB by some reasons.
I did not know that new Application Report was published, until I got your reply. Probably I may post some questions about the new Application Report to another thread, before telling him.
Do you think that the prematuer clipping is caused by the low PVDD voltage rather than PFFB?

Best regards,

Akio Ito

Akio-san,

So the customer observed that in the exact same test conditions the amplifier didn't clip when there was no PFFB? 

If possible an easy way to rule out if its the lower power supply is simply applying 30V to PVDD and seeing if there is still clipping (even with PFFB).  If so I would look into the PFFB component values.

Best Regards,

Robert

Robert-san,

The customer has not observed the waveform of their board without PFFB yet.
It seems to be difficult to change the PVDD voltage now.
I can't disclose all parts value on this thread, so I show only resistors value related with PFFB gain.

R_in =2.7kΩ
R_fb =18kΩ
Input series resistor : 100Ω -> input resistance (Rin_TPA3245) : 24kΩ + 100Ω = 24.1kΩ
The precised gain of TPA3245/PFFB taking TPA3245 input resistance into account is as follows. (see 'TPA3251D2' thread)

TPA3245/PFFB_Gain
= (R_fb * Rin_TPA3245 * Gain_TPA3245)/(R_in * (R_fb + Rin_TPA3245 * (Gain_TPA3245 + 1)) + R_fb * Rin_TPA3245)
= (18e3 * (24e3 + 100) * 10^(18/20)) / (2.7e3 * (18e3 + (24e3 + 100) * (10^(18/20) +1)) + 18e3 * (24e3 + 100))
= 3.2375
TPA3245/PFFB_Gain(dB) = 20*log10(3.2375) = 10.2dB

The customer sent us a waveform with CLIP_OTW terminal signal. I show it below.

yellow: PWM output
magenta: /CLIP_OTW
cyan: prestage OP-AMP output (5.4Vp-p)
green: LC filtered output

Best regards,

Akio Ito

Akio-san,

Thanks for the waveform.  Based off of the waveform I am only seeing the noise from the LC filtered output getting clipped and not the actual signal.  

That said it really depends on what the customer considers clipping.  Do you know what THD+N the customer considers to be clipping? If they are basing the clipping on the CLIP_OTW pin, that can start sending pulses around .01% THD.

Best Regards,

Robert

Robert-san,

The problem our customer minds is the noise at CLIP_OTW asserted, is not THD.
They have been using TI full digital solution for a long time.
He compares clipping noise of TPA3245/PFFB with it of TI full digital solution e.g. TAS5086+TAS53xx.

Best regards,

Akio Ito

Akio-san,

So the customer's problem is that the CLIP_OTW comes on too early even though there isn't really any clipping in the device yet?

Regards,

Robert

Robert-san,

When TPA3245 starts clipping, CLIP_OTW starts asserting too.
The customer wants 30-Watts at THD=1% under 6-Ohm load.
However, the output voltage is less than 18x2 Vp-p at PVDD=21V, when it starts clipping.
18^2/2/6 = 27 Watts

Best regards,

Akio Ito

Akio-san

The CLIP_OTW starts turning on at around THD+N=.01%.  If the device has just begun to start clipping at 27W then it will be under THD+N= 1% for 30W.

Best Regards,

Robert

Robert-san,

Thank you for your reply, but I would not agree on your answer.
If it takes account until 19-th harmonic, the output signal level must be 1.0295 FS at 1 percent THD.
The Fourier series at THD=1% is as follows.

a1*cosωt + a3*cos3ωt + a5*cos5ωt + ... + a17*cos17ωt + a19*cos19ωt

a1= 1.0235E+00
a3= -5.6937E-03
a5= -5.1794E-03
a7= -4.4701E-03
a9= -3.6318E-03
a11= -2.7391E-03
a13= -1.8667E-03
a15= -1.0805E-03
a17= -4.3107E-04
a19= 5.1098E-05

(a3)^2= 3.2418E-05
(a5)^2= 2.6826E-05
(a7)^2= 1.9982E-05
(a9)^2= 1.3190E-05
(a11)^2= 7.5029E-06
(a13)^2= 3.4847E-06
(a15)^2= 1.1675E-06
(a17)^2= 1.8582E-07
(a19)^2= 2.6110E-09

Σ(an)^2= 1.0476E-04
√(Σ(an)^2) = 0.010235203

(18V*1.0295)^2/6/2 =28.616 W < 30 W

Now the customer has another problem. It seems more important than this problem. I want this thread to be pending for a while.

Best reagards,

Akio Ito