Numeric format for State Variable (Q/F Input)?
Dear ADI Engineers:
I'm trying to use the State Variable (Q/F Input) in my ADAU1701 design but this modue is not documented at all in Sigma Studio 3.5. I'm having trouble figuring out what numeric format I should use for the F input. I'm pretty sure it cannot be 5.23 since that only allows values upto 15 or so; so is it 28.0 which would be pure decimal or is it something else?
I believe the Q-input is 5.23.
Thanks! - Ram
Audio Signal Processors
numeric
format
variable
state
SigmaDSP Audio Processors
adau1701
Show More
The format for F is 5.23, and the calculation is:
Input to F pin = 2 * sin(pi * cutoff / fs)
I believe the format for Q 5.23, and it is inverted. In other words...
Input to Q pin = 1/Q
Thanks Brett. I just had a look at the "Tracking" filter spec too. The F-input formula to the State variable (Q/F) is not the same as what is used for the Tracking Filter which is
Input to F pin = Cutoff/(Fs/2).
In particular one of the consequences seems to be that in the formua you provided the effective F is limited to Fs/2, corresponding to a F-input of 2*sin(pi/2) = 2.
- Does this mean that the allowable range for the Tracking filter is greater than Fs/2?
- Also, is the Q-input to the State-Variable (Q input) also meant to be 1/Q?
Thanks very much!
-Ram
Hi Ram,
Limiting the filter cutoff to Fs/2 makes sense because Fs/2 is the Nyquist frequency, and any signal components above that frequency would be aliased back into the audio band. However, the analog components external to the ADCs in the system should guarantee that the input signal is bandlimited to Fs/2. So, this is not really a limitation. The tracking filter could mathematically be set to work above Fs/2, but in that case, I think it would simply alias back below Fs/2.
The tracking filter's Q value is stored as 1 / (2Q).
Oops - I was stupid to have missed that point on the Nyquist frequency. Thanks.
I do wish however that there was greater consistency between modules. Knowing that all modules that have an F-input use the same formula would help greatly and reduce the need for explicit documentation or - as in this case - posts to the forum to get the magic formulas! Hope you can plow this feedback to the DSP design teams.
One of my questions remains: Is the Q-input to the "State-Variable (Q input)" also meant to be 1/Q?
Best - Ram
Please log in to reply.