SMU : RMS vs Average Confusion

[gitansh95]

@QuazarTech/support We are confused about whether the SMU is an RMS meter (returns the square root of the mean of squares of the datapoints) or a DC meter (returns simple mean of datapoints). We are using filterLength = 1, to intentionally bypass any averaging in the software/firmware. But what we are interested to know is how the hardware alters the signal.

Let us say we are sampling voltage at a rate of 1000Hz. Both the Average and RMS of a pure DC voltage is equal to the DC voltage value. But if there are fluctuations in the voltage, what happens to those fluctuations?

Also, let us say that fluctuations range from very low frequency (~0 Hz) to very high frequency (~1e+14 Hz). Will we be able to see all of these fluctuations (sampling at 1000 datapoints per second)? To rephrase, upto what frequency fluctuations are visible on the SMU?

[kchat79]

The hardware consists of a preamp, a 1st order low pass filter, and an ADC.

Among these, the low pass filter is the strongest. It tries to smoothen out any fluctuations.

The ADC has a comb filter with notches at integral multiple of line frequency.

In no place, RMS is recorded.

KC

On Sep 27, 2017 7:29 PM, "gitansh95" notifications@github.com wrote:

@QuazarTech/support https://github.com/orgs/QuazarTech/teams/support We are confused about whether the SMU is an RMS meter (returns the square root of the mean of squares of the datapoints) or a DC meter (returns simple mean of datapoints). We are using filterLength = 1, to intentionally bypass any averaging in the software/firmware. But what we are interested to know is how the hardware alters the signal.

Let us say we are sampling at a rate of 1000Hz. Both the Average and RMS of a pure DC voltage is equal to the DC voltage value. But if there are fluctuations in the voltage, what happens to those fluctuations?

— You are receiving this because you are on a team that was mentioned. Reply to this email directly, view it on GitHub https://github.com/QuazarTech/Qrius/issues/4, or mute the thread https://github.com/notifications/unsubscribe-auth/AZkf6Gb3n4IFkkln0enaS123asqVvxVxks5smlS4gaJpZM4Pl0Dd .

[mchandra]

• Could you further elucidate the difference between the ADC sampling rate and the bandwidth set by the low pass filter?
• I'm assuming that the ADC sampling rate is 10^3 samples / sec. So you are sampling a signal at 1000 Hz. Does that imply that the max freq one can sample is ~ 1000 / 10 = 100 Hz (assuming 10 points to accurately reconstruct the signal) ?
• How does the low-pass filter figure into this? Why does the instrument need a low-pass filter? Doesn't the max sampling rate of the ADC already act as a cutoff?

Thanks

[nishant90git]

Hi,

The inputs of all ADCs need a low pass filter. The measurement of ADCs is not immune to frequencies beyond the sampling rate, and can be corrupted by them because of the "aliasing effect". You can look up the term on internet for more info. This low pass filter is also called "anti-aliasing" filter.

The bottom line is that you will see a "anti-aliasing" filter with the cutoff frequency (-3dB or half power point) of roughly 5 to 10 times the sampling rate or lower (in case you are interested only in even lower frequency range) on all ADC inputs. This is true for all kinds of measurements, AC or DC. Also this filter is normally over and above other filters that the preceeding analog path may have.

Also, I have a feeling that you are trying to use XSMU for noise analysis. I would strongly recommend caution because the circuit for noise analysis is very different from that of a DC measurement, if not completely opposite.

In the latter case you are interested in the DC value the signal has notwithstanding the noise present.That is, even if there is noise present in the signal that you are measuring you try to remove it before you sense it using an ADC.

On the other hand, noise measurement is inherently an AC measurement. In the first stage you infact remove any DC bias the signal may have and then amplify the noise. It becomes a DC signal only if you take RMS of the signal after removing the DC voltage. I can elucidate on that if needed.

To illustrate the point simply, a DC measurement should not be affected in any manner (ideally) by the amount of noise the measured signal has. While the noise measurement should not be affected in any manner (again, ideally) by the DC bias the measured signal has.

If you really need something to measure the noise level of "your" signal, then a new product may need to be thought of. Some work on that already happened last summer. The circuit is simple but the real challenge will come from the level (magnitude and frequency) of signals that you are interested in.

On the other hand, if you are simply interested in understanding the noise levels of the XSMU circuit then it is fine.

Also, all of the above is also true for nano Volt meter too. Conceptually, it is not different from a normal DC volt meter apart from the measurement range.

Nishant.

On Fri 29 Sep, 2017, 3:52 AM Mani Chandra notifications@github.com wrote:

• Could you further elucidate the difference between the ADC sampling rate and the bandwidth set by the low pass filter?
• I'm assuming that the ADC sampling rate is 10^3 samples / sec. So you are sampling a signal at 1000 Hz. Does that imply that the max freq one can sample is ~ 1000 / 10 = 100 Hz (assuming 10 points to accurately reconstruct the signal) ?
• How does the low-pass filter figure into this? Why does the instrument need a low-pass filter? Doesn't the max sampling rate of the ADC already act as a cutoff?

Thanks

— You are receiving this because you are on a team that was mentioned. Reply to this email directly, view it on GitHub https://github.com/QuazarTech/Qrius/issues/4#issuecomment-332980245, or mute the thread https://github.com/notifications/unsubscribe-auth/AceUe_YU1jXYd46WoZNrgMBpYWJq05VNks5snBw-gaJpZM4Pl0Dd .