TLK106: The LED_LINK pin is on but the Link Status bit is 0

Hi yacob,

Are you using a custom board or the TLK106 evm?

Can you share the schematic of your phy if it's a custom board?

Are you reading register 0x1 multiple times?

Best regards,

Hello Rob,

Thank you for your reaspond.

This is a custom board.

Attached please find the schematic diagram of TLK106.

I read the register 0x1 repeatly and receive the same result.

Sincerely,

Yacob.

Yacob,

The Rbias resistor for the tlk106 needs to be 4.87k 1% ohm. This value is critical to operation of the phy. Please replace that resistor with the proper value.

Best regards,

Hello Rob,

Thank you for your fast repsone.

I understand to replace the R324 to 4.87K.

What about C184 (100 pF)?

Sincerely,

Yacob Hassidim.

Hi Yacob,

I would recommend removing C184. I do not see a large impact, but it is unnecessary. Please let me know how your HW change turns out.

Best Regards,

Hello Rob,

Thank you for your suggestions.

I will inform you about the result of HW changes.

By the way the datasheet has not reference diagram for Rbias like Figures 3-1 and 3-2.

Can you please ask them to add figure also Rbias?

Sincerely,

Yacob Hassidim.

Hello Rob,

I replaced the R324 by 4.86 KOhm.

I didn't remove the C184 yet.

Unfortunately the link status bit is 0 (while the LED link is on).

The value of status register is 0x7849.

Can you please advise me?

Sincerely,

Hello Rob,

I replaced the C184 also.

No change.

Can you please advise me?

Sincerely,

Yacob Hassidim.

Hello Rob.

I removed the C184 also.

No change.

Can you please advise me?

Sincerely,

Yacob Hassidim.

Hi Yacob,

OK, that was the first thing I saw was an issue.  I'd like you to check the following items.

1. Look at timing of input reference clock PHY_2_CLK.  If PHY_2_CLK is not stable when the TLK106 power comes up, you may need to apply a RESET pulse to the TLK106 to initialize properly.

2. Measure RBIAS voltage during normal operation (should be approximately 1.2V)

3. Remove L37 and replace with 0 ohm resistor

4. Ensure PFBOUT is ~1.55V +/- 5%

5. Put the TLK106 in digital loopback mode and check for a link up in register 0x1

Best Regards,

Hello Rob,

Thank you for your response..

Regarding with your suggestions:

1. The board has not especial reset for TLK106. There is a system reset button that resets also the CPU and TLK106. After system runs I pushed the reset button and the status bit is 0.

2. RBIAS voltage is 1.26V.

3. L37 has been removed and replaced by a short wire.

4. PFBOUT voltage is 1.56V.

5. Digital loop back causes LED link on and link status bit (in register 0x1) be 1 (after second read).
After reset of digital loop back and using loop back cable: LED link on but link status bit is 0.

Same occurrence with PCS loop back.

I am waiting for your advisement.

Sincerely,

Yacob Hassidim.

Hello Rob,

There was another change that I think helped for the link status bit in digital loop back:

In the first time the SMT resistor R234 (6.49 K) was replaced by Through Hole resistors total of 4.86 K.
Today we used a SMT resistor of 4.87 K instead of the Through Hole resistors.

After all the changed the link status bit with external loop back cable is still 0.

I am waiting for your advisement.

Sincerely,

Hi Yacob,

With the loopback cable in place, can you please provide a register dump of all registers in the datasheet?

Do you have another PCB available with a different TLK106? I want to rule out possible damage to the TLK106.

Best Regards,

Hello Rob,

Thank you for your support.

Attached please find Excel file that contains the dump of of TLK106 registers.

I just read all the registers from 0 to 0x215 as detailed in TLK105/TLK106 data sheet (SLLSEB8C –AUGUST 2012–REVISED APRIL 2016).

The file contains the following sheets:

- Dump of registers for the PCB with all the HW changes of your suggestions (e.g. R324 = 4.87 K)

- Dump of registers for the PCB without the HW changes (e.g. R324 = 6.49 K)

- Compare of the sheets

It seems they have not any meaningful change (most of the changes are in the reserved registers).

By the way I checked  the link status on the PCB without the HW changes and with the digital loop back bit and it was 1.

It means the HW changes till now have not effects on the loop back bit.

Sincerely,

Yacob Hassidim.

6278.TLK106_RegistersDump.xlsx

Hello Rob,

Attached please find a Word file that contains figures of board layout and questions about the design.

Can you please look at the file and advise?

Sincerely,

Yacob Hassidim.

TLK106_Layers.docx

Hi Yacob,

I don't see any problems in your layout besides the possibility of the impedance being incorrect. Did you confirm that the differential characteristic impedance is 100 ohms?

I don't see any significant problems in the register dump you gave me. There is a difference in register 0x10, but that difference is a false positive it appears. The 6.49k RBIAS value indicates a link is present in register 0x10, though no link is present in register 0x1.

One thing that is a problem with your register dump is that your values repeat every multiple of 0x1f, this means your extended register access is not working. Please review datasheet section 4.3.1.3 and 4.3.1.4

At this point I would also try replacing the TLK106 device with another.

Do you have access to a TLK106 EVM?

Best Regards,

Hello Rob,

Thank you for your response.

Regarding with register 0x10, value 0x2 (bit 1) refers to speed 100 MHz and not link status. The link status is bit 2 and it is 0.

In accordance with TLK106 datasheet and TLK1XX Design and Layout Guide we made the following changes:

• Clock input: 50 MHz (from 25 MHz)
• Pulled up the MII_MODE pin (26)
• Removed the capacitors of 22 uF (C169, C167, C165 and C214)

Unfortunately no change: the link status bit of status register is low (and the LED link is on).

Can you please answer to the following questions?

A) Are our changes OK?

B) In accordance with datasheet section 2.5 Clock Interface: "RMII is not supported with a 1.8V reference clock".

    The XI clock of TLK106 is 2.2 Volt.

    Does this voltage cause the link status bit to be low?

C) What are the conditions for the link status bit of status register to be high?

D) Should the link status bit be high if the HW is OK and a loop back cable is connected?

E) What are the differences between the logic of link status bit and LED link?

Sincerely,

Yacob Hassidim.

Hello Rob,

I am waiting for your answers.

Sincerely,

Yacob Hassidim.

Hi Yacob,

Your changes for the RMII mode are correct. You are finding problems with link-up tho, so I expect your interface isn't an issue.

Yes, the reference clock should be 3.3V, and a 2.2V LVCMOS oscillator could cause the TLK106 to have issues.

The selection of RMII vs MII should not have much impact on the establishment of a link, unless the reference clock you are using is incorrect. You have the proper clock frequency listed tho.

Have you tried replacing the TLK106 device?

C) What are the conditions for the link status bit of status register to be high?
When auto-negotiation is completed, or a valid MLT-3 signal is detected using forced mode. The register 0x1 bit[2] will go high on the second read, because this is a latched low bit.

D) Should the link status bit be high if the HW is OK and a loop back cable is connected?
If the HW is OK, the PHY is in auto-negotiation mode, and a loop back cable is connected, the link status bit should go high.

E) What are the differences between the logic of link status bit and LED link?
The link status bit in register 0x1 is latched low, the link LED is derived from register 0x10 bit[0].

Best Regards,