Anti-Aliasing Filter
Hello,
I am having trouble disabling the default anti-aliasing low pass filter in my IEPE channels. (Dewesoft Krypton Hardware) I am fine with the 100kHz hardware filter but need raw vibration data for FFT post-processing. I read in another forum post that the sample rate needs to be set at 20kHz to provide the option, unfortunately that did not resolve the issue. I have my default lp filter set to 'off' in the DewesoftX options settings.
Dear Freek van der Linden,
thank you for reaching out to us.
As specified in the Krypton manual, pages 78 and 79, our Krypton-ACC devices utilize a Hardware filter that cannot be turned OFF in the channel setup, regardless of the sample rate:
Here's what the behaviour looks like in Channel setup:
Devices, such as for example SIRIUS devices, do have a Low-pass filter that can be turned off, as long as the Dynamic acquisition rate is the same as the channel's sample rate:
Best regards,
Dewesoft team
Hi,
I've not felt the need to dive in this deep before, but I wonder how I should interpret this:
Does it mean that the Alias rejection filter has a -3 dB effect at 0.49×fs and the alias rejection of -100 dB means attenuation of the signal with -100 dB at 0.453×fs?
Or how should I else interpret "Alias-free Bandwidth" and Alias Rejection above?
BR,
Hi,
I've not felt the need to dive in this deep before, but I wonder how I should interpret this:
Does it mean that the Alias rejection filter has a -3 dB effect at 0.49×fs and the alias rejection of -100 dB means attenuation of the signal with -100 dB at 0.453×fs?
Or how should I else interpret "Alias-free Bandwidth" and Alias Rejection above?
BR,
Dear Benjamin,
Here's how we define the Bandwidth, Alias-free Bandwidth, and Alias Rejection:
Bandwidth (-3 dB) = 0.49 fs -> This is where the filter has started to noticeably attenuate the signal. At 0.49 fs, the measured amplitude is attenuated by 3 dB, and the frequencies below are measured with less attenuation.
Alias-free bandwidth = DC to 0.453 fs -> This is the frequency range within which the signal is not contaminated from frequencies above Nyquist folding back into the measurement.
Alias rejection = -100 dB (all sample rates) -> Tells us how much the frequencies above the alias-free limit are suppressed so that they don't 'contaminate' the desired signal.
Below is an image of a graphical representation, although the values written on the graphs don't correspond to our Krypton devices:
Here's an actual example that explains these parameters:
Let's say we sample at fs = 20 kHz. We can then determine:
- Bandwidth (-3 dB) = 0.49 fs = 9.8 kHz -> this is the Cut-off frequency
- At the frequency of 9.8 kHz, the filter has reduced the amplitude of a real 9.8 kHz signal by 3 dB. If we had a 1 V signal, we would measure around 0.707 V.
- Alias-free bandwidth = DC to 0.453 fs = 9.06 kHz
- This means that, if we sample at 20 kHz, the frequencies measured up to 9.06 Hz will be measured without significant contamination from aliased signals.
- Alias rejection = -100 dB (all sample rates)
- If we sample at 20 kHz, the Nyquist frequency is 10 kHz. In a real system, a filter cannot cut off instantly at 10 kHz, so we define a stop-band frequency above it. The signals above this stop-band frequency are attenuated by more than 100 dB before sampling, ensuring that when they fold back (alias), their energy is too low to 'contaminate' the desired signal.
Best regards,
Dewesoft Support