DP83869EVM: 100Base-Tx amplitude

Hello,

This deviation may be attributed to different corners of silicon, board design, setup variation (such as cabling/probing), as well as component tolerance. IEEE is a very strict standard in and of itself. While tolerances are always tricky to ensure performance is within specification from a HW specification, we also have register configuration which can help. 

I would like to ensure that setup/probing is as optimal as possible before diving into register configuration. Could you please provide the vendor you are using to measure, as well as photo of the setup?

Sincerely,

Gerome

Hello,

here is the figure of test setup. I am doing ECT test about 8 years with already many PHYs so I am convince that the deviation is cause by PHY is self. My the biggest concurs are the so big deviation in the amplitude depending at taken PHY piece and also the amplitude symmetry is on the edge. Can you provide me what is the tolerance of output amplitude or statistic deviations?  What can be adjust by reg?

Hello,

It appears you are using a TDSET3 FW which is similar to what is done in our labs. I just want to confirm that the picture you are using was done for 10Base-Te testing and not for the 100Base-Tx testing you are reporting correct?

To be able to reign in the amplitude symmetry, you may decrease the respective field using Reg 0xA0. Please note this is trimmed and will vary per device. It is advised to read back the default value, make a change based on the corresponding channel/field only, and then retest. This not only will decrease VoD leveling (helping with board 2), but also have an indirect effect on increasing symmetry (board 1).

Sincerely,

Gerome

I am confirming that I am measuring the 100BASE-TX Ethernet compliance. The setting in register is not suitable for us because it must be set differently for each piece of PHY. We are in the phase of developing of product with huge quantitie so this aproach is realy not possible.

we need to know what we can then expect. Can you us the output amplitude specification , what level we can expect? Do you have also some distribution measurement over many sample, in the best case over the works case samples?

Hi Viktor,

I have personally seen run/run variation occur when utilizing TDSET3 software. Could you please try the following:

- Increase sample size from 64 to 256

- Run more than once (I've done 3x times) to get a general range of where the PHY is at performance-wise

Tektronix had been consulted on this in the past and their response was that sometimes sampling isn't optimal and the tradeoff to increase size is time to complete a test. As the performances are very marginal, I would like to see if it was a bad acquisition being root cause of your dataset.

Sincerely,

Gerome

There are my result. Before I repeated measurement I make this

 - waiting 30 min before oscilloscope warm up

- probe compensation was performed

I wanted to point out that the simulate issue we discover with DP83822IF ( DP83822IF: MDI 100BASE-TX voltage tolerance  where the distribution was provided to my colleague. But in case of DP83869 I see even more output deviation

Hi Viktor,

Thanks for providing the data points. It does appear that the violations are consistent. I am looking into why this is the case. Can you please provide the FAE that is assigned to your Siemens team? I would like to also coordinate efforts with them.

Sincerely,

Gerome

Hi Gerome,

sorry but I do not know what is FAE, can you give me more information

Hi Viktor,

Your FAE (Field Apps Engineer) is Simon Forstner. I am in contact with him as well as my inner team to discuss this topic.

Sincerely,

Gerome

Hi Viktor,

As an update, I have tried to measure a 2x EVMs on my side using TekExpress and found no issues so far. I will look to expand my dataset in the chance that this was good picks from our distribution. However, we would like to explore correlation between what is tested from a device standpoint and the EVM test. Please look for next Tuesday for update.

Sincerely,

Gerome

Hello,

As an update we are still looking at correlation behavior between what we have tested and how this exhibits on EVM for compliance testing. We are sending some devices out to our test houses for recollection, but this will take time due to international shipment. Please expect next update next Wednesday.

Sincerely,

Gerome

thank you for feedback. Starting next week I am going to be at holiday for 3 weeks, so take your time :-)

Hi Viktor,

Acknowledged. Have a good holiday!

Sincerely,

Gerome