DP83TD510E: Calculated SNR values from DP83TD510E seem high

Part Number: DP83TD510E

Hello,

We are using the DP83TD510E and have been reading the 0x0A85 register then converting it to SNR SNLA364A.  Here is the code that we are using:

               taskENTER_CRITICAL();

               HAL_ETH_WritePHYRegister(&heth, 0x0D, 0x001F);

               HAL_ETH_WritePHYRegister(&heth, 0x0E, 0x0A85);

               HAL_ETH_WritePHYRegister(&heth, 0x0D, 0x401F);

               HAL_ETH_ReadPHYRegister(&heth, 0x0E, &mse);       // Read the PHY dsp_reg_72 register.

               taskEXIT_CRITICAL();

 

               Sig2Noise = -10.0 * log10((float)mse/131072.0) - 1.76;

But the SNR we get seems to high.  For example, we get the following results from one test:

From DP83TD510E register read and calculation, SNR = 28.5dB

A scope capture of the signals looks like this:

So we estimate the signal is actually about 1.2V (half of the 2.39V seen in the capture because of full duplex transmission), and noise is about 10mV.

By our estimate, SNR ~ 10 log (1.2/0.01) = 20.8dB

Why is our estimate so different from the SNR reported by the DP83TD510E?  The best SNR we expected was for a 2.4V signal = 23.8dB.

Thanks for your help,

Cyrus

  • Hello Cyrus

    Thank you for the Query.

    I will review the inputs and come back you you by mid of next week.

    Regards,

    Sreenivasa

  • Hello Cyrus

    I am checking with the team and is taking some time due to the US holiday.

    Regards,

    Sreenivasa

  • Hello Cyrus

    Thank you for your patience.

    Please refer below reply

    SNR value is the value that our PHY’s core predict after its attempt to filter the noise and conditioning of the incoming signal. Hence measurement of SNR through scope capture at input of PHY and intentional noise addition may not correlate with final SNR seen by the PHY.

    Regards,

    Sreenivasa

  • Hello Sreenivasa,

    We are trying to use the SNR measurement to identify problems with the 10BASE-T1L link.  We try increasing the length of cable we are using and monitor the SNR.  With a 300m cable SNR is 29.3.  With a 400m cable SNR is 28.5.  With a 600m we cannot establish 10BASE-T1L communications.  

    SNR >20.38 is supposed to be good.  We are far from that at 400m and signal level is about 1.15Vpp (overall the signal is about 2.3Vpp and we expect about half of the amplitude is TX from one PHY and half TX from the other PHY).  However, we have no communications at all at 600m. 

    Can you provide any guidance on this?  Does it make sense the SNR is that high?  Is there some other way we can anticipate the link is near failing?

    Thanks,

    Cyrus

  • Hello Cyrus

    Thank you for the inputs.

    What is the cable type you are using for testing. Have you tried with the other voltage.

    Regards,

    Sreenivasa

  • Hello Sreenivasa,

    I am using Belden 3076F fieldbus cable.  I don't expect it to reach 1km.  I'm just trying to help my customers by letting them know how well the cable connected to my product is working if they try reusing their existing cable.  I am hoping SNR will help me do this.

    I ran the test with a 1Vpp signal.  Here are the results along with the 2.4Vpp signal results:

    Cable Length (m) Signal Level SNR
    2 2.4V 30.1
    2 1V 30.6
    100 2.4V 30.2
    100 1V 30.4
    300 2.4V 29.3
    300 1V 28.4
    400 2.4V 28.5
    400 1V 27.9
    600 2.4V No Link
    600 1V No Link

    Cyrus

  • Hello Cyrus, 

    Thank you for the measurements.

    Would you be able to check with a cable length between 400 to 600 m ?

    Regards,

    Sreenivasa