DP83867E: Using the Phy with no access to programming headers?

Hi Hillman,

The polarity is swapped on the MDI Side.

Also, is Mode 4 on the LED_0 not also SGMII mode?

Hi Anand,

Sorry for the confusion. you are right. SGMII can be enable through mode 2 and 4. 867 PHY will automatically swap the polarity on the MDI side, so the polarity swap in the on the MDI lines will not cause an issue with the Link Driver.

--

Sincerely,

Hillman Lin

I dont have any access to read the registers. Right now I have such that LED_0 is strapped to mode 3, RX_CTRL is strapped to mode 3, LED1 is strapped to mode 3.

Is it required for RX_D0 to RX_D3 to have dummy straps as indicated in the datasheet to establish a link from my MAC to the PHY? 

I know I have a polarity swap going into the PHY from the RJ45. Should I enable mirror mode to fix it?

Are the clock skew pins dont cares in SGMII mode?

Hi Anand,

1. you need to strap the LED_0 to mode 2 and 4 to enable the SGMII

2. Yes, You need a dummy straps for SGMII mode

3. No, it is not required enable the mirror mode. 867 PHY will automatically swap the polarity on MDI side. To know more about the mirror mode, kindly refer to the table 8.4.6.6.

4. Clock Skew pins does not matters on SGMII mode since it passes through a differential signals

--

Regards,

Hillman Lin

Hi Hillman,

Unfortunately it seems when I add the strap resistors to the LED0 pin, the pin appears to look like an open circuit. There's no current flowing into it. It measures 3.3V at the input which is the same as the supply voltage.

Can you verify if the 9k internal pulldown is always present on power on?

Hi Hillman,

One more thing. If I want to drive an LED in mode 2 on the LED0 pin, what is the recommended schematic to do so?

Hi Anand,

• 9k only get connected in the beginning of the power up sequence to latch the strap value. In the functional mode, there is no 9k pull down and pin is an output pin.
• What do you mean by mode 2 in LED. Can you help me to point me to the correction section of datasheet

--

Regards,

Hillman Lin

LED0 I want to set the strap config to mode 2 and drive an LED. What is the recommended schematic to do that?

Hi Anand,

Please refer to session 8.5.3 in the datasheet for recommended schematic to drive an LED.

--

Regards,

Hillman Lin

Despite that I cannot establish a link.

I have confirmed RX_CTRL is in mode 3 and LED0 is in mode 2 when probing those pin on startup. However there is still no link that can be established. Since we're trying to run 1000BaseT, AutoNeg_sel=1 is a don't care correct? 

Further, we found that on the input pins to the PHY from the RJ45, the polarity is flipped, i.e + from the jack is going to - on the PHY. But when we tried a mirror cable where + and - were flipped no link could be established. What else should be checked on the PHY to eliminate that as being the problem?

Hi Anand,

I need some clarification:

• How did you check for link up? 
• Can I have the schematic review for your design?
• Are you able to access the Register now?
• Yes, AutoNeg_sel does not matter if you are running 1000mbs
• Polarity on MDI side shouldn't be the issue since 867 PHY will auto correct the polarity.

--

Regards,

Hillman Lin

1. I checked for LED0 being driven high, which according to the datasheet happens when there is a successful link. Is this true?

2. I will try and get a similar schematic to what we have. The application is sensitive. I can tell you that the SGMII inputs are being driven by/SGMII outputs are driving a pair of LMH1297 in Equalizer mode and Driver mode respectively. I have confirmed that the LMH1297s are running in the modes that we want but I have not confirmed the signal integrity of the Ethernet at that end. Might have something by Monday.

3. No, I am not able to access the registers on this board.

Hi Anand,

Yes you are correct, the LED0 should be high when you see the link up if you did not change the value of LEDCR1 register (0x0018).

--

Regards,

Hillman Lin

Hillman, here is the extent of the schematic I can show you. SGMII In and Out connect to Line Equalizer/Driver LMH1297.

PHY_LEDs connect to Resistor divider to set strap voltages on startup.

Hi Anand,

I look through the schematic and I have couple concerns:

• MDIO and MDC is not connected to the MAC?
• We recommend to coupling capacitors on the MDI side for the link up issue.
• Could you also show me the schematics on magnetic components on the connector side?
• After you change your schematic on LEDs. Could you please send me the updated schematic on LED_0 pin?

--

Regards,

Hillman Lin

LED 0 to 2                                             RJ45 MagJack

Let R57=500Ohm, R53=2.4kOhm in the LED schematic.

MDIO and MDC are not connected to the MAC side, no. 

Hi Anand,

I need a clarification on your strapping for LED_0:

• Is your R57 500 ohms? Because it does not meet our strapping requirement.

Did you remove the coupling capacitors on the MDI side for the link up issue?

--

Regards,

Hillman Lin

R57 is 500 ohms yes. When measured with an Oscope, the voltage level was .561V. VDDIO is 3.3V in the application, so this voltage will strap LED0 to mode 2 according to the datasheet.

Remove coupling capacitors? I am not sure if you mean we should remove them, but no, we have not removed any caps.

Hi Anand,

1. If you want to set mode 2 in the LED_0, please follow the datasheet recommendation. (Rhi 10kohms and Rlo 2.49k ohms)
2. Yes, the coupling capacitors should be removed. This might cause an issue on the link up in the MDI side.

--

Regards,

Hillman Lin

Hillman,

WIll follow up with results within 3 days.

Hi Anand,

Feel free to reach out to me if you have other issue.

--

Regards,

Hillman Lin

Hillman:

We have done as you suggested, but there is no change in behavior. I have verified that LED0 is strapped to mode 2. I have also verified that there are packets going into the Phy. 

For a little more detail as to where it is connected, see below.

This is as much detail as I can give you.

Hi Anand,

We discussed the ongoing issue internally with the team and as there are many variable here, to make any further progress we suggest to do the following  : 

1. Correct the schematic for the required mode with components (straps, LEDs) selected as per the datasheet recommendation.

2. Share the updated schematic + desired mode of operation of PHY for us to review and provide feedback.

3.  Enable the access to MDC/MDIO of PHY so that we can debug using register reads.

4.  Share the detailed block diagram of signal path indicating the placement of LMH1297 and its corresponding connection with PHY etc

Let me know when you have above available for further remote debug.

--

Regards,

Hillman Lin

Hillman,

It will be some cycles before I can get all of that together but I will try to have an updated schematic by end of week. Anything else I find I will share,

Thanks

Hi Anand,

I will wait for your result.

--

Regards,

Hillman Lin

Here are the updated schematic sections. Please take a look and comment. 

Hi Anand,

I am not able to follow why did you add LMH1297 on the MDI side for 867 PHY because we have never done that before. Could you give me a bigger picture on the application on why you add LMH197 for future debug?

I look at the schematic, Could you make LED0 to mode 4 so that it can enable the Mirror mode and LED2 to mode 1. Other part of the schematics looks fine. Keep in mind that in mode 4 LED 0 is active low. Please follow the schematic on the datasheet to design the LED session from the following figures: 

--

Regards,

Hillman Lin

The line drivers and equalizers are connected to run coax over ethernet. That is the most I can tell you on the application.

I will change the schematic as recommended, but why do I want to enable mirror mode?

Hi Anand,

Mirror mode is not required. We recommend it to enable in case of some orientation swapping on MDI side. Please refer to the session 8.4.6.6 in the datasheet for more information.

--

Regards,

Hillman Lin

Hillman,

I have a launchpad board to talk to the Phy over USB, but I am not getting a detect on the Phy when I search for it with application. Could you verify if the board is wired up correctly?

Hi Anand,

The top pin you are connecting on Microcontroller is MDIO pin and the bottom pin are MDC pin. Make you you connect the ground pin as well.

--

Regards,

Hillman Lin

I have done as you suggested but I have encountered a problem in the app shown in this picture. I am using the app to scan for PHYs but I am returning them all as available which should not be possible, as the phy I am probing this way has its address set to mode 3, as seen on the schematic.

Do you know why this issue may come up? I cannot read any of the status registers as they all come back with 0 which should not be true.

Here is a picture of the behavior on the data line. The data as read by the USB 2 MDIO app is coming back as all zeroes, and we have verified looking at the voltage levels on startup that the Mode on LED0 is Mode 4 LED1 is mode 3. Therefore, register 006E = 8822, correct? At least, the SGMII enable bit should be 1. Why would this be happening?

Hi Anand,

I will test the microcontroller in the lab later this week. Meanwhile, Could you summarize all the issues you are currently facing we can discuss this internally through emails or calls.

--

Regards

Hillman Lin