DP83TD510E-EVM: How can the PRBS checker be reliably locked?

Hi Gerrit,

You can try to use the following script and check

001F 8000
0119 055F
001F 4000

This should make PRBS lock signal 1.

--
Regards,
Gokul.

Hi Gokul,

thank you for your answer! We have used your sequence now and we read the PRBS_STATUS_4 at the end:

begin
001F 8000  //Hard reset
0119 055F  //PRBS_CFG_1: Enable packet generation towards cable and packet checker coming back from the cable
001F 4000  //Soft reset
011F 0001  //PRBS_STATUS_4: Lock & Clear signal for the PRBS counters
011F       //Result is 0B00 on success and 0800 when failed
end

The result is more or less the same - sometimes it works, sometimes not. The chance is about 50%.The same problem is on the DP83TD510E-EVM using a Windows PC, as well as on our custom-made hardware.

Do you have another idea?

Thank you,

Gerrit.

Hi Gerrit,

Can you please share a block diagram of the whole setup with details on where the loopback of the data is happening?

Is the link between TI-PHY to TI-PHY?

--
Regards,
Gokul.

Here is a small diagram of our test setup. It is really very simple: We have two DP83TD510E coupled together via a cable, both are connected by USB to a PC running USB-2-MDIO. One of the DP83TD510E is in reverse loopback mode, the other one generates packets with the PRBS generator, configured as described above.

           TI-PHY1                                      TI-PHY2
  PRBS Packet Generator Config                 Reverse Loopback Config
    +---------+------------+                  +------------+---------+
    | DP83822 | DP83TD510E +~~~~~~~~~~~~~~~~~~+ DP83TD510E | DP83822 |
    +---------+------+-----+      Cable       +------+-----+---------+
                     |                               |
                     |                               |
                 USB-2-MDIO                      USB-2-MDIO

We configure the DP83TD510E in Reverse Loopback Config (TI-PHY2) only once and keep it in this configuration.

The other DP83TD510E (TI-PHY1) is completely reconfigured for every attempt. We find that after a hard reset (001F 8000 //Hard reset), followed by the other configuraton steps described in my previous post, the PRBS checker is not always locked - only in about 66% of the attempts. The phenomenon is clearly related to the hard reset, but we cannot see why it does not work in some cases. It looks like a random phenomenon at the moment. The communication link is always established, but the PRBS checker is not always locked. When PRBS checker is locked, that is in 66% of the attempts, we read 0B00 from register 011F. In all other cases (failured attempts) we read 0800.

We might create a loop and try it again and again if the PRBS checker is not locked after a hard reset (followed by the other configuration steps), but this would not be satisfacotry. Do we need to do more configuration steps, e.g. write to other registers? A more relaxed timing (waiting 1s after hard reset) does not help either.

Hi Gerrit,

Can we try hard reset first and then wait for link-up before configuring PRBS?

This would require you to first write 0x1F=0x8000 and then wait to manually watch LED_0 go up and then program the script above (without 1F 8000) to enable PRBS.

--
Regards,
Gokul.

Hi Gokul,

we have tried this, but it does not help. The PRBS checker is still not locked after the hard reset sometimes, also when the LED_0 is active. The link is established always, but only the PRBS checker is not locked. Is it possible that our reset and initialization sequence is not (see above) is not sufficient?

Thank you

Gerrit

Hi Gerrit,

I will try to recreate this on my end by Monday next week.

I will be using the following sequence.

-------------------------------------------------------------------------------------

Step-1: 0x0001 = 0x8000 //Hard Reset

------------ Wait for at least 2 sec-------------

Step-2: 0x0119 = 0x055F (I'll remove 0x001F = 0x4000, I suggest you also to)

------------Wait for at least 2sec------------

Step-3: Write 0x011F = 0x0001, Read 0x011F

------------------------------------------------------------------------------------

Can you please try this too?

--
Regards,
Gokul.