LMX2594EVM: close in phase noise worse than simulation

David Kleeberger

Part Number: LMX2594EVM

Tool/software:

Hi,

Im using the LMX2594EVM together with a 125MHz referece clock from Crystek (https://www.crystek.com/specification/clock/CCPD-575.pdf). The simulations results with the original EVM loopfilter yield this performance:

However I do measure this phase noise performance with a FSW with k40 option:

Any hints why the close in phase noise is about 30 dBc worse than the simulation results?

When I use a SMB100B with 125 MHz as reference, the phase noise significnatly improves. So I assume my measurement settings of the FSW are correct.

Cheers David

1 month ago

0 Noel Fung 1 month ago

TI__Guru** 107300 points

Hi David,

How did you connect the LVPECL clock to the EVM?

0 David Kleeberger 30 days ago in reply to Noel Fung

Prodigy 20 points

The clock is a SMD chip an directly soldered to the EVM baord. Is used this configuration for the AC-coupled LVPECL:

R27=R28= 180R

R25=R30 = not fitted

R32 = not fitted

R24b=29b= 0R

0 Noel Fung 29 days ago in reply to David Kleeberger

TI__Guru** 107300 points

Hi David,

You should include R32, it is the termination for the driver.

0 David Kleeberger 29 days ago in reply to Noel Fung

Prodigy 20 points

Hi Noel,

Thanks for the advice. I did implement the 100R termination, unfortunately this did not solve the issue with the bad close in phase noise.

The blue traces shows the updated circuit topology and the purple trace shows the legacy configuration. Both yield the same phase noise. Anything else I could try?

0 Noel Fung 28 days ago in reply to David Kleeberger

TI__Guru** 107300 points

Hi David,

Phase noise is already good, the problem is you have a bump between 10kHz and 50kHz offset.

This is likely due to the power supply or the XO.

Could you verify if the output of the XO is good? if the problem is due to the XO, try different shunt resistance for R27 and R28. usually the shunt resistance is between 150Ω and 240Ω.

Clock & timing

Clock & timing forum

LMX2594EVM: close in phase noise worse than simulation