A year ago I was involved with an audio design using a SHARC 21479 running from a 24 MHz crystal and the internal clock multiplier set to 8x.
I was amazed that the EMC emissions were very low from the core of the 21479, and virtually no RF emissions out onto circuit board traces and cables! Even near-field probes showed low RF emissions coming from the core and crystal of the 21479.
I've just got my first prototype circuit boards with a SHARC 21489 on it, and it's functioning ok - accepting i2s audio in and sending it back out.
I designed a 6-layer pcb and took care to have 2 internal ground planes to provide good EMI supression.
My question is in regard to the crystal oscillator section of the circuit.
I used the same 24 MHz crystal and same parts layout location from last year's 21479 board.
The main difference I see is that I'm using a 21489 with the internal clock multiplier set to 16x, for a 400 MHz core clock.
I'm noticing that the harmonics of the 24 MHz crystal are much stronger from the core of the SHARC chip on my board than on the boards with the 21479. Both radiated RF emissions and RF emissions viewed with a near-field EMC probe show a higher RF content from the core of the SHARC 21489 chip.
I'm wondering if someone knows what the cause of the difference in RF strength is?
Is the 21489 just a known stronger RF source than the 21479?
Or does setting the clock multiplier to 16x make the difference?
Does the physical placement of the crystal load capacitors has any bearing on how strong the oscillator signal is?
My bottom line goal is to see if I can tone down the strength of the SHARC's crystal's harmonics.
Page 23 (attached) of the SHARC 21489 data sheet shows adding a series resistor between the crystal and the XTAL pin.
Is this a hint from Analog Devices that the 21489 is known to need some crystal drive power control?