- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Team,
What is the max. frequency that can be cutoff by a LPF - My design involves a corner freq. around 500 KHz. &&
How does values of C1, C2, C3, C4 affect the same ? Which one should be preferred for a noise ridden (highly unpredictabe) pulse which rises sharply but fades off way too gradually ! (Butterworth/Chebyshev/2 or 4 pole)
Thanks.
- Labels:
-
PSoC 1
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Ramanathan,
"Don't dream it, be it" (Rocky Horror) or rather do it: Just place an LPF on your design, then right-click on it and select the wizard in the upcoming window to have automatically set all the Cs for you.
Happy filtering
Bob
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
As far as I know, the best (only) way to get rid of noise is to average the samples. The characteristics for your questioned filters are here http://en.wikipedia.org/wiki/Electronic_filter
Bob
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Playing around with the Filter shows:
Oversampling should be about 10 to give acceptable results.
at 500kHz cut-off frequ. would result in 5MHz sampling frequency.
Would result in a clock frequency of 20MHz (4 times sample frequ.) which is more than the listed absolute allowable maximum of 8MHz.(I could only manage to use 6MHz)
Seemingly that will be difficult to perform on a PSoC1
Bob
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Averaging is but one method of signal noise suppression, and has
rapidly diminishing returns.en.wikipedia.org/wiki/Signal_averaging
Some basic filter stuff - www.cypress.com/
Limitation of filtering with PSOC SC blocks is, for large signal, slew rate limited.
The slew rate of the analog column output buffer is 0.65 V/µsec. For operation at high power, this results in a lower maximum signal limit than the filter itself.
Further you have issue with pole/zero accuracy need to oversample. So 500 Khz would need
to be oversampeld by 10 or greater for meaningful response accuracy, PSOC 1 just not capable.
Note an alternative is to do a filter based on fixed gain, like PGA, but 500 Khz for large signal BW
may be a push, you have to look at GBW of PHGA for the gain needed to do a 2 pole LPF.
Note tempting might be to use onboard MAC, buts its speed to do a IIR 2 pole filter inadequate.
Regards, Dana.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Scratch my comments on using PGA in fixed gain filter, it is simply
too low in GBW.
Regards, Dana.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
If you look at this the moving average filter < 20 db in suppression.
n = 16.
http://ptolemy.eecs.berkeley.edu/eecs20/week12/freqResponseRA.html
Regards, Dana.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Bob & Dana,
I'm working on it. My scintillation detector's pulses (analog : 700µs in duration having a sharp rise & a gradual descent) have a dynamic range of 0 - 400mV (from Vss) & are ridden with high-freq noise . But when it is fed to a LPF (which has its reference as AGND(2.5V)), I observe nothing as such @ the LPF's o/p.
I tried amplifying but PGA (gain-8) couldnt amplify as expected. I didnt know why! [input pulse slew rate was not a problem- 60mV/µs].
Anyway to bring the LPF's reference to Vss ? (if this can be done, LPF can be applied directly with/without amplification)
RAM
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Can you provide a screen capture of pulse complete with V, T scale factors,
and ground reference ?
Maybe a project archive as well.
Regards, Dana.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I'll send for sure. But pls think of a way to do this - Anyway to bring the LPF's reference to Vss ?
Thank you - RAM
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Looking at TRM filter is tied to AGND, in turn set by ref mux in global resources.
I do not see any choice for Vss.
From datasheet -
The filter has programmable gain. The filter is biased at AGND, selected in the Global Resources, and the gain is referred to this level.
- Polarity
-
This selects the polarity of the output with respect to the input. Select Inverting to set the output to invert the signal relative to AGND. When Gain is set greater than 1.0, the gain is relative to AGND. When Gain is set to 1.0 and Polarity is set to Non-Inverting, the signal is not referenced to AGND. The parameter applies only to topologies using the A-input.
Regards, Dana.