DP83822H: Cannot find PHY ID issue discussion

Hi Charles,

What is the expected phy id and what are the strap resistors (pull-up and pull-down used to configure this phy id on respective pins) for that phy-id? Are there more than one devices on the mdc-mdio bus : can there be clash of phy IDs? 

--

Regards,

Vikram

Hi Vikram,

The ID=0x03. There are 4 PHY in the bus, the other three ID are 00 01 02 and can be read by AM5726. Only ID=0x03 cannot be read, the pull-down and pull-up resistor value are below,

D0: RH=2.49k, RL=NC

D1: NC, NC

D2: NC, NC

D3: NC, NC

I detected all strp pin voltage and it meet the target in our datatsheets. 

Do you have some other suggestions here to find root cause? Thanks.

BR,

Charles Lin

Hi Charles,

Is power up sequence matching what is recommended in datasheet? Is there a crystal connected to the device or an oscillator? If oscillator is connected then is it stable before the supplies becoming stable? What is the voltage level on rbias pin?

If possible do share the schematic.

--

Regards,

Vikram

Hi Vikram, 

The power-up sequence matches what is recommended in the datasheet. The customer used crystal. You can see the schematic below.

What will happen if the oscillator is not stable before the supplies becoming stable, will RESET pulse solve this issue? Thanks. 

BR,

Charles Lin

Hi Charles,

Yes if we apply reset pulse after stablization of clock, phy should come back to correct state. Is reset helping? Did you measure rbias pin voltage?

--

Regards,

Vikram

Hi Vikram,

It will not help if we apply reset pulse to this PHY. The Rias voltage is about 0.97V, aligned with the EVM. If DP83822 enters some unknow mode, will reset pulse let PHY come out of that mode? 

Any other suggestions can be shared here to identify the root cause? Thanks.

BR,

CL

Hi Charles,

Do you mean that applying reset pulse did not help? I assume following was the sequence : supplies power up => wait for some time (?) => toggle resetn pin low => keep it low for some time (?) => toggle reset pin high => wait for T2 time (?) => start reading the register ( In reset timing diagram of datasheet there is T2 parameter which talks about valid MDC clock after reset). Lets confirm this timing and values of these delays.

We may also check whether MDC/MDIO timing is being met (serial management timing section of datasheet).

From Rbias voltage, it looks like that phy is powered up but somehow not responding to phy id even though the strap resistors for phy id are correct.

Also are other 2 phys also 822 or they are from other vendors?

--

Regards,

Vikram

Hi Vikram,

YES. You are right, the reset pulse cannot solve this issue. I check the reset timing, all met the spec claimed in datasheet except that MDC signal cannot control by AM5726, the MDC is always here when reset pulse is applied to PHY. 

We have another 2 DP83822 in the same MDIO line, they can be read by AM5726. So I do not think T2 will not affect the PHY ID reading. How about your opinion here? Thanks. 

BR,

CL

Hi Charles,

Yes it is strange if other phys are also 822. So other 822s have phy address 1 and 2? (as 0 is not the supported address) . I am not aware of any issue with phy id being 3 but can we do following experiments.

=>  Power down other phys so that they dont interfere on MDC/MDIO lines and then try addressing phy with phy id 3.

--

Regards,

Vikram

Hi Vikram,

PHY ID should be supported by DP83822. The method you recommended here has been tried by us, the result is the same. What other methods you can think here to solve this issue? Thanks.

BR,

CL

Hi Vikram,

PHY ID should be supported by DP83822. The method you recommended here has been tried by us, the result is the same. What other methods you can think here to solve this issue? Thanks.

BR,

CL

Hi CL,

Yes you are right. I assumed default value (without any strap) as the minimum phy id address.

One more thing I can think of is changing phy id of the non-responding phy from 03 to 01 (to see if there is problem with particular strap setting itself). Have you already tried this also?

Also is COL pin connected anywhere other than the strap (for phy id = 3)? Is it different for other phys?

I think that earlier you mentioned that you checked the voltage level of strap pins. What was the voltage level of COL pin?

--

Regards,

Vikram

Hi Vikram,

Thank you for your response here. The issue is resolved by adding a 22ohm resistor in serial in the MDIO line. This is very strange for me to understand, for we change the CLK frequency to a slower speed before but it cannot solve this issue. Would appreciate it if you can share your professional experience here. Thanks.

BR,

CL

Hi Charles,

Yes it is strange. Did the customer share some waveforms of MDC and MDIO with and without 22ohms? Was this 22 ohms recommended by MAC side ?

Is this 22 ohms has been placed only for phy with phy id 3?

It will be very interesting to have a look at the layout of MDIO traces for these 3 phys.

Let us know what you find and it will help us to capture this as possible solution for similar used layouts/MAC. 

--

Regards,

Vikram