DS92LX1621: CAD PIN ADDRESS DECODER

Masakazu Adachi

Intellectual 2500 points

Part Number: DS92LX1621

Hello team,

They are using following devices.

Serializer　DS92LX1621SQE/NOPB

Deserializer　DS92LX1622SQE/NOPB

The customer was unable to R/W to a register without access via I2C communication during the following conditions.

Do you know the cause?

10kΩ pullup(+1.8V)／ RCAD = 0Ω
Address 7'bit　Ser:h'58　Des:h'60 -> Unable to R/W

Also, the customer is able to R/W under the following conditions.

10kΩ pullup(+1.8V)／ RCAD = 2.0kΩ

Address 7'bit　Ser:h'59　Des:h'61 -> Able to R/W

10kΩ pullup(+1.8V)／ RCAD = 4.7kΩ

Address 7'bit　Ser:h'5A　Des:h'62 -> Able to R/W

Regards,

Masa

over 3 years ago

0 Lysny Woodahl over 3 years ago

TI__Expert 3595 points

Hello,

I have a few clarification questions in order to best help:

1. Can the customer write to just the serializer or just the deserializer locally when RCAD is shorted to ground?

2. Could you show me what the circuit configuration they are using when the customer is unable to write to the device?

Thanks!

Lysny

0 Masakazu Adachi over 3 years ago in reply to Lysny Woodahl

TI__Intellectual 2500 points

Hi Lysny,

I heard to the customer.

1. The customer says they could not write to the Serializer/Deserializer by connecting directly to GND.

2. The circuit configurations as follows,

Main board side: Ser-Des (1 set), accessed from SoC via I2C communication

Module board side: Ser-Des (1 set), accessed via I2C communication from FPGA

The data path is via an inter-board connector.

Main Ser -> Module Des -> Module Ser -> Main Des *Daisy-chain connection.

Regards,

Masa

0 Lysny Woodahl over 3 years ago in reply to Masakazu Adachi

TI__Expert 3595 points

Hey Masa,

You mentioned there are two pairs of serdes in the design. Is this the correct summary of the customer issue? "When the pullup resistor is 10kOhm and the pull down is 0Ohm to ground, neither the serializer or deserializer can be accessed via SoC I2C or FPGA I2C. But, when the pull down resistor is any other value specified in the data sheet, all I2C communication via SoC I2C and FPGA I2C functions properly."

Proactively, if the above description is correct, could you please ask the customer to read address 0 when I2C communication is possible?

Also, is the customer setting both serializers to the same I2C address? Or, both deserializers to the same I2C address?

I am trying to isolate when and where the error is happening. If you could also send me the schematic the customer is using for setting the I2C address, that would be helpful!

Thanks,

Lysny

0 Masakazu Adachi over 3 years ago in reply to Lysny Woodahl

TI__Intellectual 2500 points

Hello Lysny,

Thanks to the support. Your summary is correct.

I have a question.

>>Proactively, if the above description is correct, could you please ask the customer to read address 0 when I2C communication is possible?

Does this mean reading address 0 when RCAD is 2.0kΩ or 4.7kΩ? I would be glad to know what that means. Sorry for my limited knowledge of Ser-Des.

Also, I will confirm address configuration and circuit diagram to the customer.

Regards,

Masa

0 Lysny Woodahl over 3 years ago in reply to Masakazu Adachi

TI__Expert 3595 points

Hi Masa,

The customer can read address 0 when RCAD is any of the working values. This is the table from the datasheet:

I would like the customer to check the value of address 0, when address 0 can be read. The value should match this table.

Thanks,

Lysny

0 Masakazu Adachi over 3 years ago in reply to Lysny Woodahl

TI__Intellectual 2500 points

Hello Lysny,

Thanks for showing me.

I will contact you when I receive a reply from the customer.

Regards,

Masa

0 Lysny Woodahl over 3 years ago in reply to Masakazu Adachi

TI__Expert 3595 points

Hi Masa,

Okay!

Thanks,

Lysny

0 Masakazu Adachi over 3 years ago in reply to Lysny Woodahl

TI__Intellectual 2500 points

Hello Lysny,

Sorry for the delay.

I confirmed address configuration and circuit diagram to the customer.

See attached png file.

And I confirm it now.

>>I would like the customer to check the value of address 0, when address 0 can be read. The value should match this table.

Regards,

Masa

0 Casey McCrea over 3 years ago in reply to Masakazu Adachi

TI__Guru** 103215 points

Masakazu,

The 2 DES devices in this diagram will have virtual I2C connections due to the I2C passthrough functionality. So they can not have conflicting addresses like this. You need the two DES devices to be strapped to different I2C addresses

Best Regards,

Casey

0 Masakazu Adachi over 3 years ago in reply to Casey McCrea

TI__Intellectual 2500 points

Hello Casey,

Do you have an example of what the SER and DES should be addressed as in this configuration?

Regards,

Masa

0 Lysny Woodahl over 3 years ago in reply to Masakazu Adachi

TI__Expert 3595 points

Hey Masa,

One serializer and one deserializer can be left with RCAD = 0 Ohms and one deserializer and one serializer should be changed to RCAD = 2k Ohms. (Any other listed RCAD value would also work for this situation.)

With the listed setup above, you have four distinct I2C addresses.

Let me know if you have any additional questions!

Thanks,
Lysny

0 Masakazu Adachi over 3 years ago in reply to Lysny Woodahl

TI__Intellectual 2500 points

Hello Lysny,

Thanks.

By the way, the I2C bus (SCL/SDA) of the Ser-Des on the main unit side and the module side are not shared but connected separately, but does the I2C passthrough functionality cause virtual I2C connections?

Regards,

Masa

0 Lysny Woodahl over 3 years ago in reply to Masakazu Adachi

TI__Expert 3595 points

Hey Masa,

Yes I2C passthrough connects the devices to the same I2C bus.

Thanks,
Lysny

0 Masakazu Adachi over 3 years ago in reply to Lysny Woodahl

TI__Intellectual 2500 points

Hi Lysny,

In the case of this customer's configuration, the I2C buses of the main and module are separate, so I don't think there is any problem even if the Ser-Des addresses are conflicted.

I assume that the customer wants to use the same address because they want to daisy chain many modules together.

Regards,

Masa

0 Lysny Woodahl over 3 years ago in reply to Masakazu Adachi

TI__Expert 3595 points

Hi Masa,

Could you reconnect with the customer and ask why they want to use the same I2C address? I'm not clear on the design reason to use the same address instead of distinct addresses.

From the looks of the diagram you sent, it appears that the I2C bus on the deserializer is connected to the second serializer. This will connect the all four devices to the same I2C bus.

You can also prove that the same address is the issue here by disconnecting the first deserializer from the second serializer to confirm that communication works when RCAD = 0 Ohms.

Thanks,
Lysny