LMX2594: t1, t2 delay

Taketo Sato

Genius 10325 points

Part Number: LMX2594

Hello team,

I have a question about phase synchronization.

Datahseet figure24 shows a concept of the function, and t1 and t2 delay is mentioned in the figure.

Does both t1 and t2 delay time have dependency on Fosc?

In other words, does LMX2594 shows different t1 and t2 delay time when comparing Fosc=40MHz case and Fosc=50MHz case with exactly same register settings?

There is a description that the delay time is dominated by VCO calibration time. If Fosc changes, state machine clock frequency also changes, so I think LMX2594 shows different t1 and t2 delay time if changing Fosc frequency. Is it correct?

Best regards,

over 6 years ago

0 Hao Z over 6 years ago

TI__Mastermind 18070 points

Hi Taketo-san,

Yes t1 is lock time + MASH_RST_CNT time, where MASH_RST_CNT should be longer than lock time with sufficient margin. So t1 is dependent on OSCin frequency.

t2 is the delay from input to output. Dean has a document on this: LMX2594 Global Delay.pdf

Regards,

Hao

0 Hao Z over 6 years ago in reply to Hao Z

TI__Mastermind 18070 points

The short answer for t2 is that it indeed changes with PFD, VCO frequency and so on, but it does not have a well-characterized relationship as between lock time and state machine clock.

Also, t2 has about 60ps variation from part to part.

Regards,
Hao

0 Taketo Sato over 6 years ago in reply to Hao Z

TI__Genius 10325 points

Hello Hao-san,

When measuring an EVM with 40MHz Fosc and 50MHz Fosc, there is about 500ps difference as t2 delay time. Register settings are same(output divider=192 and N divider=192) in both measurements but only Fosc is different.

is the 500ps time difference expected as t2 delay difference between 50MHz Fosc and 40MHz Fosc?

Best regards,

0 Hao Z over 6 years ago in reply to Taketo Sato

TI__Mastermind 18070 points

Hi Taketo-san,

I haven't measured this myself but 500ps doesn't sound right to me, because this is half a cycle for a 1GHz signal and I wouldn't think that the difference can be this much.

Could you briefly describe your measurement procedures? I assume that the output frequencies are the same with OSCin, which are 40MHz and 50MHz, according to your descriptions. Is that true?

Regards,
Hao

0 Taketo Sato over 6 years ago in reply to Hao Z

TI__Genius 10325 points

Hello Hao-san,

Where does the 1GHz signal information come from? 500ps is half a cycle for a 1GHz, however test measurement condition is 40MHz and 50MHz, not 1GHz.

What time can be expected as t2 delay time difference between 40MHz and 50MHz?

Best regards,

0 Hao Z over 6 years ago in reply to Taketo Sato

TI__Mastermind 18070 points

Hi Taketo-san,

I was only mentioning 1GHz to describe the 500ps. Anyways, I'll measure this in the lab and get back to you later today.

Regards,
Hao

0 Hao Z over 6 years ago in reply to Hao Z

TI__Mastermind 18070 points

Hi Taketo-san,

I see 360ps difference with 40MHz and 50MHz so 500ps should be normal.

Regards,
Hao

Clock & timing

Clock & timing forum

LMX2594: t1, t2 delay