AD2/EE input sensitivity

[SL01]

The AD2 spec shows a 500µV/DIV sensitivity on the scope analog inputs. Is this a "real analog" circuit with that sensitivity, or is it a circuit with lower sensitivity but software interpolated to reach a higher sensitivity? On the EE, I could not find this sensitivity spec at all - what is it?

 

Thanks.

[attila]

Hello,

The AD2 scope ADC is 100 MHz, 14 bits, two gain steps for 5V / 50V range, ~330uV / ~3.6mV resolution

The EExplorer scope ADC is 40 MHz, 10 bits, with adjustable resolution from ~800 uV to ~40 mV in about 60 gain steps.

 

[SL01]

Hi Attila,

 

Thank you for your reply.

The AD2 datasheet mentions 500µV in the low range - is it in fact lower at 330µV as you wrote?

[attila]

The AD2 resolution at high gain is ~330uV  = 5.4V x (1 + calibration) / 2 ^ 14
The calibration coefficient varies for each device and it is about 1%

[SL01]

Thanks, Attila. My AD2 arrives today and I will test it accordingly..

[SL01]

Hi Attila,

The 330µV value is a calculated according to your formula so that assumes it is devoid from impact from DNL. It will be very interesting to see what the measurements show.

Thanks again.

[SL01]

Hi Attila,

Looking at the AD9648 datasheet, there appears to be a discrepancy between the DNL it listed, and the one listed for the AD2. The AD9648 datasheets specs ±0.5 LSB, whereas the AD2 specs ±0.35 LSB. Moreover, the AD9648 datasheets specs ENOB at 12 bits at F_in at 9.7MHz and lower as the frequency climbs, which means its sensitivity would be at best at 1,318µV. That's quite different than 330µV.

The variable that would change it would of course be the 5.4V or a negative calibration factor.

What am I missing?

 

[attila]

The ~330uV is the resolution not the sensitivity.
The DNL of ±0.35 LSB might be a typo in the reference manual. The AD9648 specs ±0.5 LSB.
The 500uV/div in the reference manual is not specified as the sensitivity but the range option available in the software. "Scope scales: 500uV to 5V/div"

[attila]

Hello,

The twisty curve in FFT could show up when having logarithmic scale and start frequency too far from stop. This is fixed for next software release. For now, to prevent such curve in WaveForms settings/options select curve linear.

For averaging you have multiple options:

In scope interface the under channel option default is Average. This averages the 100MHz ADC values to the current sampling frequency. For instance for 800kHz sampling 125 AD conversion are averaged.
Under time settings you have averaging of acquisitions.
In FFT you have various peak hold, min hold, linear, exponential averaging.

Here R1 is without any averaging, R2 with sample averaging, R3 with acquisition averaging, C1 with FFT min hold.

In Spectrum interface under channel options you can also select sample averaging and you have the FFT averaging options for each trace.
Note: Set the magnitude top to 0dBV or lower to have scope at high gain (5V pk2pk) or adjust the range under channel options.
T1 min hold, T2 rms average, T2 sample average

Here is a noisy 10MHz signal acquisition averaged (I used flying wires) R1 is not averaged.