LMH1228: EOM accumulation time

Kawai

Guru 19775 points

Part Number: LMH1228

Hi Team,

Could you please tell us about the EOM function ?

Q1). How long the device wait and monitor errors for each cell ?

Q2). Does this time differs by data rate ?

Q3). How fast is this error count sampled ?

Q4). Does this sample time differs by data rate ?

Best Regards,

Kawai

over 5 years ago

0 Nasser Mohammadi over 5 years ago

TI__Guru* 88895 points

Hi Kawai-San,

In general, our retimers hit count capture time is specified by retimer register page reg 0x2A. Further, this is data rate dependent. Below is a formula showing each cell capture time based on data rate and reg 0x2A content:
Each cell capture time = 256 * reg 0x2A(decimal) * (32/data rate).
For example, at 11.88Gbps default value for reg 0x2A is equal to 0x30 = 48(decimal).
Duration to count each cell hit count = 256 * 48 * (32/11.88E9) = ~ 33uS

Regards,,,nasser

0 Kawai over 5 years ago in reply to Nasser Mohammadi

Guru 19775 points

Hi Nasser-san,

Thank you for the calculation formula.

So the error hit count accumulating time is about 33us per cell for 12G.

Since 1UI for 12G is 84.175ps, does that mean the device will check errors for 392 times ? (33us / 84.175ps = 392)

If this is not correct, could you please tell me how the device counts error ?

Best Regards,
Kawai

0 Nasser Mohammadi over 5 years ago in reply to Kawai

TI__Guru* 88895 points

Hi Kawai-San,

Your understanding is correct.

Regards,,nasser

0 Kawai over 5 years ago in reply to Nasser Mohammadi

Guru 19775 points

Hi Nasser-san,

Thanks for your support!

Best Regards,
Kawai

0 Kawai over 5 years ago in reply to Kawai

Guru 19775 points

Hi Nasser-san,

I have one question.

When monitoring 12G-SDI, do you mean that the maximum error hit count for each cell is only 392 ?

I believe the actual error hit count has much higher value.

How can each cell have those high hit counts ?

Best Regards,
Kawai

0 Nasser Mohammadi over 5 years ago in reply to Kawai

TI__Guru* 88895 points

Hi Kawai-San,

I went over your calculation: (33E-6)/(84E-12) =~ 392857

Regards,,nasser

0 Kawai over 5 years ago in reply to Nasser Mohammadi

Guru 19775 points

Hi Nasser-san,

Sorry that I had a miss calculation.

Best Regards,
Kawai

0 Kawai over 5 years ago in reply to Kawai

Guru 19775 points

Hi Nasser-san,

From the Eye monitor error count data, cells didn't have error count such as 0xFFFF = 65536.

If each cell is checking error counts for all 392857 waveforms, many cells shoud have error count of 65536 (maximum), but we only see around 4000 as a maximum hit counts in 64x64 cells.

It seems error count is only sampled once out of tenth or hundreds waveforms. What is the correct sampling rate for the error counts ?

Best Regards,
Kawai

0 Nasser Mohammadi over 5 years ago in reply to Kawai

TI__Guru* 88895 points

Hi Kawai-San,

Each tick of hit count corresponds to 32 hits. So the hit count you get, should be multiplied by 32 to get the actual hit counts.

Regards,,,nasser

Interface

Interface forum

LMH1228: EOM accumulation time