1
\$\begingroup\$

I have been using the Analog Devices AD9553 Eval Board, I am operating it as a clock translator, inserting a 10MHz LVDS differential signal and setting the device to produce a 100MHz LVPECL differential signal. I aim to operate the device in the LBW region (B=0.17KHz) as my signal is not expected to deviate in frequency by drastic amounts.

At first I had issues with phase locking all together, but this was remedied (so may it seem). I can observe the lock signal remain high for the majority of the time, dropping low rarely.

However my main concern is that my system is supposedly locked - my 100MHz LVPECL signal does not stay stationary on the oscilloscope when triggered from the sinusoidal 10MHz reference signal. I am observing the sinusoidal reference signal as I cannot probe the LVDS signals when connected to the PLL (my scope only has 1M and 50R termination modes and this interferes - Scope:TDS5104B). I know there is not a problem with the LVDS conversion stage as all those signals are stationary on the scope.

I have had a play around with the values in the registers but to no avail (N Div etc). I have attempted to use the debug mode to view the PFD signals etc from the LOCK pin but they show nothing on the scope, the only one I can observe is the XTAL clock. I take it that because there are no signals on the PFD div 2, PFD ref div 2 and the PLL feedback div 2 - that there is a problem with my reference signal, values or the device.

Regarding my setup, I am prototyping the system on breadboard, I do not know if this could be a cause of problems (I have kept the wire lengths as small as possible and have a COAX going from the breadboard to the EVAL board) , but I am confused as it says the PLL is locked - but is infact not locked to its reference source. Attached is one of the combinations I tried from the EVAL board wizard software. AD9553 Datasheet: http://www.analog.com/media/en/technical-documentation/data-sheets/AD9553.pdf

Guidance & Advice would be very much appreciated.

Thank you for reading this.

Combinations in the EVAL software

\$\endgroup\$
  • 1
    \$\begingroup\$ 100MHz on a breadboard is definitely asking for problems. \$\endgroup\$ – Brian Drummond Dec 2 '15 at 13:40
  • \$\begingroup\$ :\ , well the 10MHz LVDS is generated on the breadboard and is transferred with COAX to the EVAL board - this has SMA outputs which I am connecting directly to the scope. So 100MHz does not pass through the breadboard. However there is some ringing on the the 10MHz signal - dont know if that could be the cause of this? \$\endgroup\$ – Renegade243 Dec 2 '15 at 13:48
  • \$\begingroup\$ Now that's a good question, worth looking into. \$\endgroup\$ – Brian Drummond Dec 2 '15 at 13:55
  • \$\begingroup\$ If the issue is with ringing (or some other issue) on the input, changing the bandwidth of the phase / frequency detector may show interesting results. \$\endgroup\$ – Peter Smith Dec 2 '15 at 14:28
  • \$\begingroup\$ Yes I have changed the bandwidth of the frequency detector to the next stage (20KHz). The PLL lock indicator is now high permanently. However the LVPECL signals are still not stationary on the scope when triggered with the 10MHz sine wave. \$\endgroup\$ – Renegade243 Dec 2 '15 at 16:30
1
\$\begingroup\$

The solution to this was to make sure the PLL is configured to accept a differential signal ! There is a register which allows this to be toggled.

Additionally the input amplitude of the 10 Meg LVDS needed to be attenuated to be accepted by the PLL, this was done via a 100R resistor across the two differential inputs.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.