thcondra42 wrote:I am looking to sample a noisy pulse input signal from a sensor. The input signal will have a maximum frequency of 50kHz, so conforming to the Nyquist Rate requires atleast 100 kSam/sec. The original idea was to use a MCP3001 ADC to convert the analog signal to a digital signal and be able to process it within the Pi (filter the signal, perform an FFT). It also involved using the Pi's clock for the ADC.
However, I've read that a 100 kSam/sec sample rate is basically impossible using this method. Is this true? And if so, what is the best way to get a digitized representation of the signal inside the Pi at that sample rate?
My apologies if this post is in the wrong place. I'm quite new to a lot of this, including the Pi itself.
thcondra42 wrote:So a microcontroller using ethernet communication would serve the same purpose as the proposed ADC, only with an increased sample rate? Is that the right way to look at it, I'm slightly confused on the actions of the microcontroller that will be communicated.
PeterO wrote:96kHz sampling is easily achived with an I2S ADC. I have 96kHz,24bit stereo input working using a simple I2S codec on a breakout board.
Higher sampling rates may be possible but the codec I'm using (PCM1803A) maxes out at 96kHz.
PeterO
That is a meaningless example that doesn't do anythign useful.mung wrote:The link seems to suggest 42MHz if you max out cpu.
Th OP said "The original idea was to use a MCP3001 ADC to convert the analog signal to a digital signal". Seems pretty clear to me !joan wrote:I'm not convinced the OP knows whether the signal source is digital or analogue. All that seems to be known is the source is a "physical sensor" which in itself is meaningless.
As a matter of principle I disbelieve at least 50% of all posts. I'm not often disappointed.PeterO wrote:Th OP said "The original idea was to use a MCP3001 ADC to convert the analog signal to a digital signal". Seems pretty clear to me !joan wrote:I'm not convinced the OP knows whether the signal source is digital or analogue. All that seems to be known is the source is a "physical sensor" which in itself is meaningless.
PeterO
What exactly do you mean by the "signal will have a maximum frequency"? Do you mean that the "worst case fundamental" will be no more than 50KHz or do you mean that there will be no frequency components over 50KHz? These are very different things.thcondra42 wrote:I am looking to sample a noisy pulse input signal from a sensor. The input signal will have a maximum frequency of 50kHz, so conforming to the Nyquist Rate requires atleast 100 kSam/sec.
plugwash wrote:What exactly do you mean by the "signal will have a maximum frequency"? Do you mean that the "worst case fundamental" will be no more than 50KHz or do you mean that there will be no frequency components over 50KHz? These are very different things.thcondra42 wrote:I am looking to sample a noisy pulse input signal from a sensor. The input signal will have a maximum frequency of 50kHz, so conforming to the Nyquist Rate requires atleast 100 kSam/sec.
To correctly sample a signal the highest frequency present in the input must be BELOW the nyquist frequency (half the sampling frequency). Signals at exactly the nyquist frequency have somewhat strange behaviour where the results depend on the phase relationship between the samping clock and the input signal. Signals above the nyqist frequency will be aliased (effectively converted to lower frequencies) and will therefore interfere with signals at lower frequencies.You will usually want some form of analog filtering to reduce (you can never eliminate them completely) the level of above-nyquist signals and noise that will be sampled and hence aliased.
Remember a square wave or pulse train has many harmonics. For good results you really want to capture the third harmonic and ideally the fifth harmonic too.
I've got a USB dongle that does something very similar but not up to radar frequencies. This is an SDR (Funcube dongle) that provides the I and Q samples over USB by acting as a 192Khz audio device. You might find some inspiration from how this works, I think there's a fair bit of technical documentation around. Starting point should probably be http://www.funcubedongle.com/terilars wrote: My intent is to sample I and Q signals from a radar module, output range is 40 Hz to 15 kHz. Do I need to go the I2S way [1], or look for simpler solution like mentioned earlier in this thread?
I haven't worked on my project for over a year. I did get a different I2S codec to play with but I could never get a device tree configuration to work so I stopped. Maybe one day I'll get back to it....ronnyek42 wrote:Seems like even the mbed audio codec board is somewhat difficult to get your hands on these days...
Peter, I read your I2S page, and Im intrigued. I wondered if there were new recommendations for i2s breakouts... or perhaps better ways if the pi3 brought anything new with it.
I'd like to get an ADC on order... found lots that supported i2s, but not really much in the form factor of a little breakout board.
