• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 33 subscribers
  • Views 799 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LMH1218: EOM update speed

Kawai
Guru 19785 points
Part Number: LMH1218

Hi Team,

Please allow me to ask you about EOM function.

How fast is the EOM data updated ?

The reason why I'm asking is that when user want to draw eye opening themselves, I wonder if SPI interface would be faster to draw eye pattern compared with the SMBus because of the clock speed.

However, if the EOM data update speed is not fast enough, user would need to wait until the data is updated as noted in the datasheet. 

Would customer have merit using SPI when using EOM function ?

Best Regards,

Kawai

  • 0
    TI__Mastermind 29655 points

    **6/30: Edited with correct calculation of time per data point**

    Hi Kawai-san,

    The udpate for the entire EOM can be calculated by taking the number of data points and multiplying by the amount of time spent at each data point.

    When I checked with our design team, the amount of time spent on each data point is approximately 33 us. Since the EOM is a 64 x 64 grid, this means that all the EOM data is ready after 33 us x 4096 = 135 ms.

    When SPI is operating at maximum frequency (20 MHz), each 17-bit transaction takes (1/20MHz) x 17 = 0.85 us, which is less than the amount of time per data point computation (33 us).

    Since SPI transactions can be completed and read-back is faster than SMBus, I believe there is merit in your logic that you will be able to view the EOM pattern quicker with SPI instead of SMBus regarding EOM.

    However, please note that the increased amount of time needed for SMBus to read back EOM data does not result in a significant difference regarding the eye observed. Therefore, I would offer the suggestion that, while SPI is quicker, both control interfaces are suitable for reading back the EOM data.

    Regards,

    Michael

  • 0 in reply to Michael Lu (Santa Clara)
    Guru 19785 points
    Hi Michael-san,

    Thank you for the information. I understood that SPI has merit in reading the EOM data faster.

    I have a little more question.

    Q) When starting EOM data sampling, after fast EOM start, does the device continuously sample 64x64 data even without reading REG 0x25/0x26 ?
    Or, does the device sample data by data after each REG 0x25/0x26 readings.

    I was wondering if user could wait for like 50us or more, then read all 64x64 data at once without waiting time.

    Best Regards,
    Kawai
  • 0 in reply to Kawai
    TI__Mastermind 29655 points
    Hi Kawai-san,

    I double-checked my math on the SPI transaction and realize I made a mathematical error where I took the inverse of a factor in the calculation. Please see my update in the previous edited post. The time per data point transaction is 33 us, not 500 ps. This means that it takes approximately 135 ms to read back all 4096 data points, assuming that you read back the data the moment it is available. Sorry for the confusion.

    The device samples each data point sequentially after each Reg 0x25/0x26 reading. Therefore, you will always have a wait time in SPI mode before the next data point's hit count is available for read back. This is why in the programming guide we emphasize the need to check on the status of Reg 0x24[0] each time before reading back Reg 0x25/0x26 in SPI mode.

    Thanks,

    Michael
  • 0 in reply to Michael Lu (Santa Clara)
    Guru 19785 points
    Hi Michael-san,

    I understand.
    In the point of obtaining the EOM waveform, it seems it may not have a great speed merit using SPI.

    Best Regards,
    Kawai