DP83848Q-Q1: Need schematic review

Hi Toshiyasu,

Do you see that link is established with the link partner by reading register 0x1?

Please send me the datasheet for the transformer. The part number shown has no results when I search for it.

If i understand correctly, the magjack has an integrated transformer and does not require an external transformer.

Datasheet for the magjack is here: https://www.mouser.com/datasheet/2/643/dr-MAG-A829-1J1T-KM-1310190.pdf

I currently don't have the code to read the registers.

I will write the code and post the value when I have it.

I'm currently having another problem.

The oscillator seems to produce an acceptable waveform when it's disconnected from the DP83848Q-Q1.

However, if I connect the oscillator to the DP83848Q-Q1, the waveform no longer looks good.

I thought I fixed this problem before, but it seems to have returned.

It sounds like you do not have link because there is a clock issue then, which makes sense why you see no link LED.

Does your connected link partner show link?

Please probe CLKOUT. Also, please make sure you have a ground pin connected on the scope probe. 

Are you sure that the Kyocera oscillator is enabled?

The amplitude is much too low.

Found and fixed the clock issue - was a bad ECO on the board.

I will try to get back to the original problem and check register 0x1.

This manual has an obvious bug on page 5.

The PHY address mapping table references pins 42-46.

This is a 40 pin chip. It does not have pins 42-46.

I have checked register 0x1 and it is zero.

In fact, I have tried to read all PHY registerse (0x0 - 0x1f) and they are all zero.

I did a sanity check and connected a Saleae logic analyzer to MDC (channel 0) and MDIO (channel 1) with the logic analyzer triggering on the rising edge of MDC.

I don't see MDIO going high ever. Logic analyzer output attached.

The MDIO issue appears to be a bug in the microcontroller vendor libraries.

Going back to the link not being detected issue - I'd like a clarification.

The DP83848Q-Q1 should not require any MDIO initialization to light the LINK LED when a link is detected, correct?

I have fixed the MDIO bug in the vendor-supplied libraries and I now see MDIO toggling properly.

The DP83848Q-Q1 does not seem to respond to MDIO read commands.

The MDIO line appears to be left high.

Saleae logic analyzer screenshot is attached.

I put a signal diode (1N4148) between the MAC and the PHY so I can make sure the MAC isn't holding the line high.

The MDIO line is definitely not being held high by the MAC.

The problem is the DP83848Q-Q1 is not responding to MDIO commands.

I tried addressing the PHY at all possible PHY addresses 0-31.

it doesn't respond to any of the them.

Does the reset pin require an external pull-up resistor?

The device does not require a pull-up since there is an internal one.

Can you please check clk_out to see if there is a 25MHz clock?

Do you see the link led on when you connect a cable?

Register access is not required to achieve link. 

Hi Ross,

They are using the RMII interface and there is a 50MHz clock on CLK_OUT (pin 21). 

The link LED signal does not go high when they connect a cable to an Ethernet switch.

Thanks,
Mitchell

Mitchell said:

Hi Ross,

They are using the RMII interface and there is a 50MHz clock on CLK_OUT (pin 21). 

The link LED signal does not go high when they connect a cable to an Ethernet switch.

Thanks,
Mitchell

Please share the clkout waveform.

Are you sure that the oscillator is behaving properly?