TPA3118D2: TPA3118D2/TPA3116D2 SNR

Hello Shawn-san,

Thank you for your reply.

The test condition is the same except for the gain setting.
I am using AP 2722.
Gain setting is a default condition of 26 dB because it is not possible to change EVM resistance constant.

Did you enable the BW from <10Hz to 20kHz please? 

⇒Yes I did.

And did you enable A-Weighting filter in the measurement setting? 

⇒Yes I did.

Please make sure the GNDing between the board and the instrument is good enough. 

⇒I think the Gnd connection may not be enough.

　I think that it is hard to strengthen the GND quite enough depending on the measurement environment.

　Please tell me how much cable and GND strengthening is necessary for reference.

How much noise level did you get during the measurement?

⇒I do not know the noise level during measurement.

　For example, does the noise level mean the value of the AP when the EVM is powered on?

 

Best Regards,

Y.Hasebe

 

 

Hello Shawn-san,

Thank you for your reply.
Measurement method of SNR is as follows

＊Hardware
Connect EVM and AudioPrecision (AP)
Input is connected by pin cable at Unbal - GND.
The output is connected to AUX - 0025 via a load resistor and it is connected from AUX - 0025 to the AP.

＊Software (AP)
The input level is set to the maximum value of the distortion level (measured value: 9.8 V) while monitoring the output waveform.
At that time, set the dB value of AP Analyzer to 0 dB.
Turn OFF the output of AP Generator and read the dB value of Analyzer.

I also noticed some points.
In our AP, A_weighted filter and AES 17 Filter can not be turned on at the same time
When 20 kHz _ AES 17 filter is turned ON, it does not become 102 dB, but it is more than A-weighted
The measured value approaches.

Does your measurement have 20 kHz_AES 17 filter turned ON

Best Regards,
Y.Hasebe

Have question about circuit, go to https://www.allaboutcircuits.com/

Buy electronic product, go to  http://www.kynix.com  or http://www.ti.com/

About robot or project ,go to https://www.instructables.com/

About arduino,go to https://forum.arduino.cc/

About signal,system,control ,go to http://archives.math.utk.edu/

Electronic components industry news,follow: https://twitter.com/Kynixsemi

learn electronic knowledge, go to https://facebookcomshaynechen79.wordpress.com/

About circuit,go to http://www.discovercircuits.com/list.html

Any question about general electronics, go to: http://forum.electronicsforu.com/

Hello Shawn-san,

Thank you for your reply.

I found a one-point mistake.
Since the distortion occurred during EVM setting without distortion output, the input level was controlled to 9.8 Vrms.
However, we found that the load resistance derating was about 25 W.
I will try measuring again.
However, A-Weighted and FILTER of AES 17 can not be used at the same time.
So I will measure using AES 17 only.

Best Regards,
Y. Hasebe

Hello Shawn-san,

I measured again.
Turn off the input signal of the AP output and measure the noise level of the load by the AP.
⇒ I confirmed.
NoiseLevel = 145 μV.

For the load resistance, the output waveform was monitored by raising the derating to 50W.
However, the monitor output waveform gradually clips and ShutDown will occur.
That is when you monitor FaultPIN it goes LOW. I will return later.
It will become its symptom repeatedly.
So I turn off input once. (To protect the device)

So, I think that something is being protected from this symptom
I think that OCP or OTP is working from the symptom that the output waveform of this EVM starts to clip.
Will not this happen with EVM?
By the way, the EVM used is TPA 3118 D 2 EVM REV.C.

Best Regards,
Y.Hasebe

Hello Shawn-san,

Thank you for your reply.

I measured ambient temperature.

Ta=25℃

The device is in the cold condition.　

And I measured 1ch only.(1 load connection)

Then I tied to measure noise level.

1.differential input

  AP⇒ON、PowerSupply⇒On, No input,Input Configuration⇒Bal-Gnd

  Noise Level≒500uVrms

２．Single end input

  AP⇒ON、Power Supply ON, No input,Input Configuration⇒Unbal-Gnd

  Noise Level≒500uVrms

If you get a more than 12V RMS signal and less than 90uV noise, the SNR could be larger than 102dB easily

⇒ So I turned off the EVM power as an experiment.

　Then the output noise was about 100 uVrms.

　I know your suggestions.
　However, I can not exclude the basic noise floor including EVM.

　Also, I connected both loads, I pulled out the input cable and measured it.
　The result was 75 uV rms, 85 uV rms.
　When turning on the power supply under this condition, it was 220 uVrms, 440 uVrms.
　I think that the variation between CH is a problem of measuring instruments and measurement conditions.
　In any case it will not be 90uVrms.

   

Best Regards,

Y.Hasebe

Hello Shawn-san,

Thank you for your reply.

The noise floor is around 100uV even when the power supply is off?

⇒No The noise floor around 0uV when the power supply is off.

    I confirmed it today and it was 0v.

　I'm sorry for the confusion.

We have a lot of materials about audio performance measurement on TI website.

⇒Thank you for your information.

    I already knew both App Notes.

However, I believe that this problem is due to measurement system (environment) rather than measurement procedure.
Because there is no description of AES 17 Fliter in the data sheet, it is measured only by A-weighted.
However, since A-weighted alone does not become the value of the data sheet, we think that AES 17 filter is important.
Is this correct?

Best Regards,

Y.Hasebe