LMX2572: N-divider when using SYNC function

John Willison

Part Number: LMX2572

I intend to sync multiple LMX2572's in a test equipment application. The reference frequency will be 98.125 MHz and the output is a non-integer multiple of the reference frequency (hence a Category 3 Sync). I have arranged an external SYNC pulse which meets the setup and hold requirements with respect to the reference input. I understand that when PHASE_SYNC_EN=1, part of the channel divider will be included in the feedback path requiring N be reduced by 2.

From Table 137 in the spec sheet:

2. If neither OUTA_MUX nor OUTB_MUX is equal to 0 (Channel divider output), program N divider as usual.

3. If one of the OUTA_MUX or OUTB_MUX is equal to 1, set N = N' / 2, where N' is the normal N divider (integer + fraction) value.

My question is: In (2), above, should the word "neither" have been "either"?

Thanks for your help,

John Willison

over 5 years ago

0 Noel Fung over 5 years ago

TI__Guru** 107890 points

Hi John,

Neither is correct.

For example, you use RFoutA only, RFoutB is powered down. if OUTA_MUX = 0; OUTB_MUX = don't care, we need to set N = N' / 2. However, even if OUTA_MUX = 1 but OUTB_MUX = 0, we still need to set N = N' / 2.

0 John Willison over 5 years ago in reply to Noel Fung

Prodigy 140 points

Noel,

Thank you for your quick reply.

So, if just one of the OUTA_MUX and OUTB_MUX is 1 the N-divider should be divided by two. But if both OUTA_MUX and OUTB_MUX are 1 (i.e., neither are 0) the normal N-divider is to be used. Correct?

John

0 Dean Banerjee over 5 years ago in reply to John Willison

TI__Mastermind 28011 points

John,

The TICSPro software shows it correctly (but you have to toggle the VCO_PHASE_SYNC bit to refresh after changes).

If either OUTA_MUX>0 or OUTB_MUX>0, then the "IncludedDivide" is equal to 2 and therefore you should program N=N'/2

So for statement 2, I think that it sould be "both OUTA_MUX=OUTB_MUX=0 …"

REgards,
Dean

0 John Willison over 5 years ago in reply to Dean Banerjee

Prodigy 140 points

Dean,

I hadn't thought to peek under the hood in TICSPro. Excellent reality check for the Boolean linguistics!

Thanks for your reply.

John

Clock & timing

Clock & timing forum

LMX2572: N-divider when using SYNC function