DP83867IR: 10Base-Te / 100Base-TX / 1000Base-T Compliance Test

Hello Koshi-san,

Kindly share the following :

1. Schematic highlighting the connections and components between phy and the ethernet connector.

2. Connector/Magnetic model number used.

3. Any register configuration tried so far for 10BT and 1000BT compliance results improvements?

--

Regards,

Vikram

Hello , Vikram

I am RYO , a customer of Koshi-san.

I will reply on his behalf.

>1. Schematic highlighting the connections and components between phy and the ethernet connector.

/cfs-file/__key/communityserver-discussions-components-files/138/Schamatic_5F00_Layout.pdf

>2. Connector/Magnetic model number used.

 model number :   LPJK4071BWNL

/cfs-file/__key/communityserver-discussions-components-files/138/_5B00_LINKPP_5D00_LPJK4071BWNL.pdf

>3. Any register configuration tried so far for 10BT and 1000BT compliance results improvements?

No,But at 10BT TEST,

Since the link partner cannot output pseudo random,
the pseudo random signal by the PHY BIST (0x16) function was used as the test signal.

Below, register settings
0x1f 0x8000, PHY RESET
0x00 0x0100, 10Base-T / Te mode
0x10 0x5008, Forced MDI mode
0x16 0xd804,

//----------------

Best Regards,

Ryo Kukita

Hello Ryo Kukita,

Thanks for all the data. I have reviewed the schematics and layout snapshot and I did not find something abnormal. Looking at the validation data, I see register<x00A6>=0x0683 give better margins. I am checking with design the possible reason but do you think you can try this extra register setting out on your board and run the failing tests again?

--

Regards,

Vikram 

Hello Vikram ,

Thank you for your reply.

I could not confirm the contents of register <0x00A6> in the DP83867IR data sheet.

Can you tell me the contents?

Which of the following failed test items is affected by this register setting?

<10Base-Te>
1) TP_IDL Template, with TPM (last bit CD0) : Fail at Load2
2) TP_IDL Template, with TPM (last bit CD1) : Fail at Load2
3) TP_IDL Template, without TPM (last bit CD1) : Fail at Load1
4) Transmitter Return Loss

<1000BASE-T>
1) Difference A,B Peak Output Voltage(w/ Disturbing Signal)

Since the test was conducted at an external testing site, there is no measuring instrument at hand. Therefore, I cannot run the test immediately.

Best Regards,

Ryo Kukita

Hello Ryo Kukita,

Ok if you have to go to external house for debug then lets list down more possible debug steps. I will be able to get back to you with register details and possible debug steps by Wednesday of this week.

--

Regards,

Vikram

Hello Ryo,

Can you please share the voltage swing level of net CLK25MHz_NET1 and the source of it? Is it coming from another 869 phy? 

Is there any difference in schematics of P1, P2, P3 and P4? Clock or power network?

Also please confirm that from DUT to measurement board a short (<2inch) CAT5e cable is used.

--

Regards,

Vikram

Hello Vikram,

>Can you please share the voltage swing level of net CLK25MHz_NET1 and the source of it? Is it coming from another 869 phy? 

  swing level of net CLK25MHz_NET1 : Vh=1.77V , Vl=0.05V

  Soure of it : Vh=3.3V , Vl=0V

 The output of the clock oscillator is distributed by CDCLVC1103 and supplied to each PHY.

>Is there any difference in schematics of P1, P2, P3 and P4? Clock or power network?

　The clock network is as described above.

　The power supply network is divided by TEST, P1, P2 and P3, P4.

See attachment #7 and above 

/cfs-file/__key/communityserver-discussions-components-files/138/Schamatic_5F00_Layout2.pdf

>Also please confirm that from DUT to measurement board a short (<2inch) CAT5e cable is used.

　Part Number : Keysight N5395-61601

    Length : 7inch ( 17.5cm)

   Type Cat5e

Best Regards ,

Ryo Kukita

Hello Ryo Kukita,

I am running late in receiving and compiling debug steps that we want to try. Instead of Wednesday as discussed earlier, I will be able to send you the steps on Friday only.

Sorry for the delay.

--

Regards,

Vikram

Hello Ryo Kukita,

I had further review meetings with design team and it comes out that we will need more inputs  :

1. For 1000BTx tests, were SMA cables used to connect load board to scope? Or was it a differential probe? Document sent earlier by you has a page titled "Measurement system (1000BT)" it shows SMA cables but Keysight part mentioned in the document actually has differential probe point to do %A-B difference test. Kindly confirm.

2. For 10BTe, usually there is a setting in the scope which is different than 10BT measurement. Is that so for the Keysight scope used?  We see that there are measurements for both 10BT and 10BTe in the results section, so just wanted to confirm that while measuring 10BTe results all required changes were made.

3. Is there an option to use shorter cable (2inch) to connect DUT and load board?

4. Do you have any compliance testing facility at your end? It may not be same as the external lab but it will help us to do a few iterations faster.

After review with teams here it came out that we will can try a few software settings to see the improvement of results of 1000BTx. But for 10BTe results we could not isolate the possible reason of these results in your system and we have'nt seen these kind of results for 10BTe.

Things are slow at our end, sorry for the delay and we will take some more time for that : by 27th May. We are looking forward for these inputs from your side and working towards compiling the configurations required to be tested. 

--

Regards,

Vikram

Hello Vikram ,

1.

Measured with a differential probe.

Connect to probing point in section 2 of N5395C.

2.

Measured at 10BTe setting.
Note: 10BT is the result measured in March 2020. Based on these results, we found that DP83867IR was not compatible with 10BT, and in April 2020 we retested with a setting of 10BTe.

3.

There is no option to use short cables (<2 inches).
There is 1 inch in the connector part. Is the length from end to end of the connector on both ends 2 inches? By the way, Tektronix TF-GBE-SIC is 4 inches including the connector part.

4.

I don't have the necessary measuring equipment for compliance testing.

//---------

Best Regards ,

Ryo Kukita

Hello Vikram ,

>1. For 1000BTx tests, were SMA cables used to connect load board to scope? 

There was an error in yesterday's reply.
I am using an SMA cable.
Please see the attached file

/cfs-file/__key/communityserver-discussions-components-files/138/Probing-Setup-for-Test-Mode1.pdf

//----------

Best Regards ,

Ryo Kukita

Hello Ryo Kukita,

What I understood from the picture shared that SMA connection points on the measurement were for return loss and other tests but not for %A-B test. I see that there are differential probe points at bottom right of the image. Can you double check if those were used instead of SMAs for %A-B test (which is failing for 1000BTx)?

We are trying to re-create the failure at our end but we dont see any reason for kind of failure of 10BTe and hence we are not able to share any debug settings at this moment. We are doing a few more tests today and tomorrow and will keep you posted.

--

Regards,

Vikram

Hello,Vikram

Does the probe point in question refer to the probe point in Section 2 of the test fixture?

Best Regards ,

Ryo Kukita

/cfs-file/__key/communityserver-discussions-components-files/138/Probing-Setup-for-Test-Mode1.pdf

Yes section 2. 

--

Regards,

Vikram

Hello,Vikram

The measurement uses Section 1 and Section 11 of the N5395C fixture, and the connection with the oscilloscope uses 2 SMA cables.

Best Regards ,

Ryo Kukita

Hello Ryo,

We have tried different boards but we have not been able to recreate the issue. Following are the differences between Tektronix test setup in TI lab (same as IEEE compliance house : UNH) :

1. Differnetial probe (Tektronix : P6247, 1.7GHz)

2. Short ethernet cable : < 2" 

Does your compliance test center has the option to use similar or Tektronix hardware. As SMA cables are wide bandwidth cables so it can be because of some unwanted noise during the test.

--

Regards,

Vikram 

Hello Vikram,

Try retesting the 1000BTx at another compliance test center with Tektronix instruments.

From the quote below, is additional software configuration required?

Quote:
① Vikram-san Reply, 5/14
Looking at the validation data, I see register<x00A6>=0x0683 give better margins.
② Vikram-san Reply, 5/22
After review with teams here it came out that we will can try a few software settings to see the improvement of results of 1000BTx.

Finally,
For the 10BTe result, are you considering possible reasons for the result and debug settings?

Best Regards,

Ryo Kukita

Hello Mr Ryo,

I have just sent you a friend request. As this issue has been open for long, so do you think we should exchange information faster by doing a phone call. This will help us resolve this faster.

I have sent you the request from E2E and it will have my email ID. Kindly write me a mail about your convenient time and we can have a short call to decide on next steps.

--

Regards,

Vikram

Hello ,Vikram

I have just sent you frend request reply from E2E Frends . 

Since this is the first operation, I'm worried that I could send it.

Did you receive a message?

Best Regards,

Ryo Kukita

Hello Ryo-san,

We have had some break-through in replicating issue similar to yours in one of the boards today. I will compile the debug steps once I receive the final data and share the debug configuration with you and FAE by 23rd June. Then if required we can have a call to align on next steps.

If you have lab time before that, can you please try following configurations :

1. For 1000BTx :

a. Measure P1's channel A with following "extra" configuration :

reg<0025>= 0x0400

2. For 10BTe :

a. Measure TP_IDL on load 2 (where it was failing earlier) with following extra configuration :

reg<0x009F> = 0xCCCC.

--

Regards,

Vikram

Hello Vikram-san,

>>1. For 1000BTx :
I can't try because no measuring instrument.

>>2. For 10BTe :
I don't have a test fixture, but I can try.

Amplitude increased by 0.2V.
Is this the result you expected?

please show a attached file.

/cfs-file/__key/communityserver-discussions-components-files/138/10BTe.pdf

--

Regards,

Ryo Kukita

Hello Vikrams-san.

>We have had some break-through in replicating issue similar to yours in one of the boards today. I will compile the debug steps once I receive the final data and share the debug configuration with you and FAE by 23rd June. Then if required we can have a call to align on next steps.

When can I have a debug configuration?

Did I have something to offer you?

I look forward to hearing from you.

Best Regards,

Ryo Kukita

Hello Ryo Kukita,

Thanks a lot for your patience. We finally have the software/hardware settings which show us improvement on our different boards. Even though our board measurements and your board measurements are not the same but we believe that carrying out these settings should give you improvements also.

Here are the steps :

Change CT1,CT2,CT3,CT4 connected caps to a value10nF each (C3131, C3132,C3133,C3134) and try following experiments for 1000BTx compliance :

1. Experiments with all channels active in parallel :

Try following configurations.

Configuration 1 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239

Check channel B compliance with following configurations :

Configuration 7 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0420}

Configuration 8 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0420} + {reg<0x00CE>  =0xB800}

Configuration 9 :  Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0420} + {reg<0x009F>=0xCCCC}

Check channel C compliance with following configurations :

Configuration 10 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0440}

Configuration 11 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0440} + {reg<0x00CE>  =0xB800}

Configuration 12 :  Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0440} + {reg<0x009F>=0xCCCC}

Check channel D compliance with following configurations :

Configuration 13 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0460}

Configuration 14 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0460} + {reg<0x00CE>  =0xB800}

Configuration 15 :  Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0460} + {reg<0x009F>=0xCCCC}

Also if possible at your end do try these compliance tests on Tektronix test hardware.

--

Regards,

Vikram

Somehow the earlier message did not get posted correctly and some lines were missing so editing the message again :

Hello Ryo Kukita,

Thanks a lot for your patience. We finally have the software/hardware settings which show us improvement on our different boards. Even though our board measurements and your board measurements are not the same but we believe that carrying out these settings should give you improvements also.

Here are the steps :

Change CT1,CT2,CT3,CT4 connected caps to a value10nF each (C3131, C3132,C3133,C3134) and try following experiments for 1000BTx compliance :

1. Experiments with all channels active in parallel :

Try following configurations :

Configuration 1 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239

Configuration 2 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x00CE>  =0xB800}

Configuration 3 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x009F>=0xCCCC}

2. Experiments with one channel active at a time :

Check channel A compliance with following configurations :

Configuration 4 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0400}

Configuration 5 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0400} + {reg<0x00CE>  =0xB800}

Configuration 6 :  Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0400} + {reg<0x009F>=0xCCCC}

Check channel B compliance with following configurations :

Configuration 7 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0420}

Configuration 8 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0420} + {reg<0x00CE>  =0xB800}

Configuration 9 :  Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0420} + {reg<0x009F>=0xCCCC}

Check channel C compliance with following configurations :

Configuration 10 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0440}

Configuration 11 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0440} + {reg<0x00CE>  =0xB800}

Configuration 12 :  Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0440} + {reg<0x009F>=0xCCCC}

Check channel D compliance with following configurations :

Configuration 13 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0460}

Configuration 14 : Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0460} + {reg<0x00CE>  =0xB800}

Configuration 15 :  Default configuration for test mode 1 as in app note http://www.ti.com/lit/pdf/snla239 + {reg<0x0025> changed to 0x0460} + {reg<0x009F>=0xCCCC}

Also if possible at your end do try these compliance tests on Tektronix test hardware.

--

Regards,

Vikram

Hello Vikram.

Thank you show the some of software/hardware setting.
I will try.

I have some of questions following :

1) Could you tell me the contents of Reg<0x00CE> and Reg<0x009F> .
2) Configuration with 10BTe is Reg<0x009F>=0xCCCC only ?

Best Regards,

Ryo Kukita

Hello Ryo,

Were you able to try out the configurations?

--

Regards,

Vikram

Hello Vikram

We plan to try it on an external testing site on July 14.

Best Regards,

Ryo Kukita