Output from a Hilbert Transform

"Probes" in the development software only work with the "stimulus" blocks, and even then are only useful for plotting simple frequency responses.  I don't think it works with Hilbert (since it's complex), and the "probe" (unfortunately) doesn't ever work with waveforms, just transfer functions.

(A signal probe and full DSP simulator would be really nice though, ADI.)

     Hello Joe,

     You're not doing anything wrong.  The Stimulus-Probe tool doesn't perform a simulation of your selected processor and signal flow -- rather, it predicts the path's response using complex math inside SigmaStudio itself.  Thus it has known limitations.  Some time ago I ran a series of tests to find what it can and cannot do, you can find the results here:

Stimulus-Probe Capability and Limitations

     Apparently we can add to the limitations list that the Hilbert Transform doesn't plot at all -- I suspect that simply, there's no corresponding algorithm in SigmaStudio to predict its response on Stimulus-Probe.  Basically, the block has unity gain and maintains a 90 degree phase shift between its outputs over the audio frequency range, but both outputs are phase-shifted / delayed from the Hilbert's input by a frequency-dependent amount.  See ADAU1446 Hilbert transform response for an example measurement of this phase lag.

In general, each SigmaStudio block involves several code resources:

• The assembly code for the DSP itself, including structures for input, output, processing, and parameter map.
• The PC coding for SigmaStudio to display the block, accept any user input, and generate needed parameters such as filter coefficients.
• The resources for computing the predicted frequency response used by the Stimulus-Probe tool.
• Whatever else I don't know about.

    Thus it's a rather amazing amount of work the SigmaStudio programmers have done for us, which is why it's so wonderfully easy and fun to program DSP this way.

    Best regards,

    Bob