I would like to know whether it is possible to program a signal generator such that it produces pulses with the same shape but with a user-defined distribution of amplitudes. The goal is to test our signal processing chain with a known distribution of input signals to see how well it can reproduce a predicted spectrum. Has anyone done that? Is there signal generator with the capability to vary amplitudes pulse-by-pulse? Thanks for any feedback. -Stephan
Solved! Go to Solution.
It depends on the specific model of a signal generator that you have. If you have an Arbitrary Waveform Generator, then you can create whatever waveform you want in software, download it to the instrument, and generate it at whatever amplitude and sample rate as you need as long as they fall within the specifications of the instrument at hand.
As far as varying amplitudes, you can change the analog gain of the instrument as needed (may glitch) or you can create waveforms in software with a smaller amplitude and download those, leaving the analog gain alone on the instrument. It depends on what the appropriate tradeoffs are for your application.
I would like to know whether it is possible to program a signal generator such that it produces pulses with the same shape but with a user-defined distribution of amplitudes.
Keep in mind, that the pulse shape (time domain) and amplitude (and phase, in the frequency domain) distribution are coupled .. one could say they are the same 🙂
Depending on the samplerates you use for your aquisition an AWG or an analog output from a multi-IO-card can do the job. From my experience it's wise to use about double the samplerate for generating than for capturing. Depending on how close you want to look at the results 🙂
We had 'jumping' results due to minor timing issues while using the same samplerate for generation and capturing, depending on what comes first, the ADC or the DAC ...
I like to test with 'analytical' pulse signals, like sin², sin³, gauss-dipole, sync ...
I am (somewhat) familiar with arbitrary function generators and understand that I can change the instrument gain and/or the amplitude of the waveform. But in the past, once I have chosen an amplitude and/or gain, I have then only run the AWG with the same settings.
In an X-ray spectroscopy experiment (which I am trying to simulate), the amplitude of each successive pulse is different, and only the statistical distribution of amplitudes is known. I am concerned that changing the amplitude for each pulse (and then only producing a single pulse with any given amplitude) will slow down the overall signal generation rate dramatically. I would like to produce pulses at 1 kHz or so. Is a change in amplitude at such a fast rate possible? Which AWG would you recommend for that purpose?
See https://www.ni.com/en-us/shop/hardware/products/pxi-waveform-generator.html for NI branded signal generators. 1 KHz is pretty slow so that should not be a problem. You will need one that support Arbitrary Waveforms. Not that these are for the most part PXIe.