Target of "Frequency Stability" plus-minus 50ppm.

Hi Tamu,

In summary first, the +-50 ppm is only needed for the crystal.  Use good components for the R's and C's and you'll be fine.

To expand on this answer let me approach this in a slightly difference way.  In general a crystal is used in three different ways for ADI video devices.

1)  As a reference to generate color burst frequency.  The color burst absolutely requires the +-50ppm spec.  Any variations in the color burst will cause color drift in the output.

2) Used for incoming format measurement such as used in HDMI receivers and their STDI modules.  Absolute timing is not that tight.  Any crystal accuracy will be good enough to measure HDMI formats.

3) For internal digital clock sourcing for modules like HDCP modules.  Absolute timing is not important, it just needs to clocked.

All video devices do not use the crystal frequency for all three applications.  When ADI first started designing video devices they came up with a standard inverter based oscillator design using the 28MHz crystal.  This basic design was just then copied from design to design adjusting it as the process nodes became smaller.

The crystal is the 1st order component driving the frequency accuracy.  Other factors like R, C and temperature are 3rd or 4th order factors.  They affect the overall frequency way less then 0.001% over their tolerance ranges.  The R and C are only used to set the right load condition for the inverter circuit so it can be begin to oscillate.  

if you select the R's and C's to closely match the requirements defined in AN-1260 then the oscillator circuit frequency will be defined by the crystal tolerance and accuracy.  Sensitivity analysis could be done but it's a lot of math.

I hope this helps, if not let me know.

Hi GuenterL,

Thank you for your reply!
I could understand the +-50 ppm is only needed for the crystal.

Please let me ask some questions.
When customers use ADV devices with a XTAL which "Frequency Stability" is +-50 ppm,
how much "Frequency Stability" by the external related circuit (including Rdump and C etc.) is supposed to be allowed?
If you have any criteria of "Frequency Stability" of the external circuit, could you tell us it?
I read the AN-1260, but I couldn't read it.
Also, does it depend on the type of ADV devices? (For example, is it different between video decoders and HDMI receivers?)

My customer uses ADV7480 now.
He uses a XTAL IC keeping the "Frequency Stability" plus-minus 50ppm.
But he says the error/variation of the external related circuit is +25ppm(only plus 25ppm, not plus-minus) in room temperature.
So he wants to know how much criteria/range of "Frequency Stability" of the external circuit is allowed.
If it's difficult to mention about the criteria of the external circuit, could you tell us whether the "+25ppm in room temperature" of the external related circuit for ADV7480 is good or not?

Thank you!
Best regards.
Tamu

Hi Tamu,

To come up with an absolute answer to how the frequency stability is related to external crystal components would require full sensitivity analysis of the equations in AN-1260. I have not done this type of calculations since I was in colledge. Looking at the circuit however I believe the frequency sensitivity for the external R's and C's would be less than 0.1 compared to the internal crystal equivalent LRC's. So for external components with 25ppm would affective sensitivity ppm would be reduced to 2.5ppm.

If you are really worried about frequency accuracy then you have to consider the tolerance of the resistor you are using. Normally we use 1% resistors. This 1% would convert to 1000ppm when applied to the sensitivity calculations. And since I know from experience that 1% parts work just fine it indicates that the frequency sensitivity is a lot less then 0.1 against external components.

There is no criteria for external components other then to use the correctly calculated values and of the highest cost effective tolerance you can get. For resistors this would be SMD 0402 1% and capacitors would be SMD X7R 0402 10%.

On a side note if you go through the capacitor calculations, the calculation might require 45pF where the closest available part would be 47pF 10%. So now you would be using a potential cap with a value of 51.7pF (47pf * (1+10%)) or 14% over the calculated value of 45pF. This doesn't even include the possible temperature variations. And it still works fine.

In conclusion the external componenets are only needed to set the proper loading in the inverter circuit to begin oscillation. Since the sensitivity is so low the external part ppm has very little affect on the stability.

Bottom line from experience I've not seen any problems using good quality external components with the correct crystal on any of our video parts.

A comment about the ADV7480. As stated in the previous post crystal accuracy is most important when generating the color burst frequency. The ADV7480 does not generate any color burst frequency. It is a completely digital device. The MIPI output clock is derived from wither the HDMI clock or LLC pin. The audio clock is derived from the HDMI source.

25ppm external components are good for the oscillator circuit.

Thank you !

Thank you !