Hi Team,
I would like to know a suitable EXTERNAL CLOCK SOURCE.
(I need around 16bit resolution and data rate is TBD.)
I can find “Table 7-2. Dynamic Range (Effective Resolution) at TA = 25°C” in the datasheet.
Could you please let us know similar table at fCLKIN=4.096 MHz and 2.048 MHz if you have any information?
Regards,
Hide
Hello Hide,
Thanks for posting on the E2E Forums! Let me take a couple of days to see if I can find a similar table for the lower frequencies.
As for "I would like to know a suitable EXTERNAL CLOCK SOURCE." Are you asking for recommendations of what kind of crystal oscillators we recommend to be an external clock source for 4.096MHz or 2.048MHz or something else? In general, the external clock needs to follow the specifications outlined in the datasheet, as shown below:
For example, we use SIT8924BA-22-33E-8.192000G--which is a 8.192MHz crystal oscillator--on the EVM which is halved by DCU0008A--a positive edge triggered D Flip Flop to step it down to lower frequencies. You don't have to use this set up in your own system but you can do preliminary testing and test the the lower clockin frequencies on the EVM. It would be best to buy a crystal oscillator for the frequency you are interested, or look into generating a digital clock that meets the above requirements.
Best,
-Cole
Hello Hide-san,
This text from the datasheet explains that answer:
So, if you set the same OSR and gain, you'll get the same value in the table. The second (highlighted) column will be calculated to a different value but all the rest is the same.
For example, I can use an OSR = 2048 (the 4th row in the table) with a Gain = 1, with a f_CLKIN = 4.096MHz--which results in a data rate of 1kSPS--you will get 4.25uVrms. According to the table, for f_CLKIN = 8.192MHz, the sample rate would be 2kSPS for the same 4.25uVrms.
So, all you need to do is recalculate the 2nd column for the new f_CLKIN and you're done.
Longer Summary:
The over sampling ratio is what determines the averaging which will change the RMS noise value in the equation for SNR and effective resolution. All other variables in the equation stay the same. Remember, this is the assumption is uncorrelated, gaussian type noise, so our signal averaging theory says the variance will decrease with more averaging.
I my opinion, this is why the highest clock is called "High resolution mode" because it allows for higher sampling rates with a higher OSR, which results in the most averaging and lowest noise.
Best,
-Cole
