LMX2581: spurious at intervals of 10 MHz

Hi Yokouchi-san,

Right, these are the signal coming from fpd and fosc. 

How did you interface the reference clock to LMX2581, could you share the schematic?

Use a smaller charge pump current could reduce these spurs. 

Hi Noel,

Thanks for the advice.
The 100MHz (XO) on the LMX2581EVM has been stopped.
I mounted a resistor so that 10MHz/0dBm (from SG) can be supplied from OSCin on the board to LMX2581 (pin 29).
(R13: 0Ω, R14: not mounted, R15: 51Ω, R16: 0Ω)
I don't think changing (or reducing) the CP current changed the spurious, but I will check again.
By the way, is it possible to say that the LMX2581 is not suitable for use at 280MHz? Or is the frequency setting (usage) not good?

Hi Yokouchi-san,

Make sure R6 is also removed. Also change L3 to 0Ω resistor.

In your application, do you need a standalone 280MHz clock or this 280MHz clock must be synchronous to the 10MHz clock?

If you just need a standalone 280Mhz clock, you may consider our BAW clock generator, it does not need a reference clock. 

https://www.ti.com/product/LMK3H0102

LMX2581 is absolutely fine to generate 280MHz clock, you may also consider LMX2572LP which has similar performance but consumes much less current.

Hi Noel,

The spurious was checked with a CP current of 0.1 mA.
The improvement was not as great as expected (Figure 1).
L3 was changed to 0Ω. By the way, the loop filter is shown in Figure 2.
I turned off 10M and checked the spurious in the unlocked state.
At this time, CP:3.1mA (Fig. 3) and CP:0.1mA (Fig. 4) eliminated spurious.
We believe we have confirmed the effect from the VCO to 10MHz divided by the VCO.
Next, when Fout is powered down, Fout (280MHz) is eliminated, but there is no change in spurious at 10MHz intervals (Figure 5).
If the CP current is set to 0.1 mA at this time, there is no change.
If 10MHz is turned off, the spurious will disappear.
This makes me wonder if it would be difficult to use 10MHz REF.
Is there any other way (is it because the LMX2581 is not properly set up?).

Regards,
Akihiko Yokouchi

fig1

fig2

fig3

fig4

fig5

Hi Yokouchi-san,

I checked this out in the lab, I can see the same spurs. 

These spurs are due to both fpd and fosc, because their frequency is identical. 

I cannot remove them or reduce them to a very low level. 

If your application is 280MHz, these spurs will be there even if you power down the RF output buffer.

I also checked LMX2572LP, it does not have this issue. Would you consider to use LMX2572LP?

Hi Noel,

Thanks for the confirmation.
I understand about the spurious.
We are still in the process of considering whether we will definitely use a new local oscillator for the system we are considering (we may use a conventional one).
However, since there is a good possibility that we will design a PLL with 10MHz as REF in the future, we would very much like to confirm the characteristics using the LMX2572LP.
So, is it possible to receive an evaluation board for the LMX2572LP free of charge?
Incidentally, LMX2581EVM was provided to us free of charge 10 years ago and is still very useful.
Would it be better to consult someone else (from TI Japan) regarding this matter?

Regards,
Akihiko Yokouchi