Sigma studio coeff calculation filter broken

Hello Gestalt,

I suspect your Q factor is set for higher value, so it started clipping. when you drag and drop the coeff calculation filter block (any 2nd order filter), the Q value is set for 1.41 as default. I drag and drop the filter you said and set to LPF of 1KHz and let Q value be 1.41 'as is'.

Below is what I got for Q=1.41

Since it's clipped, it seems like a square wave.

Set the Q value to be something like Q = 0.7, below is the result

It seems fine for me now. I have attached my program too above.

You can see this in the frequency response using stimulus and probe. I used SS just for this functionality currently; this will give you an idea about the frequency graph. (I used SS+ for simulating your query)

for Q = 1.41                                                                                          

 for Q = 0.7                                                                                

Could you please try this and get back? 

I will reply soon about the param.h file related question.

Regards,

Harish

Thanks! Yep, your definitely right about the Q. I had selected butterworth HPF and there's no Q on that one, but I can see in the parameters that it doesn't change it to 0.71 and leaves it as 1.41. You can fix it by selecting another filter and changing the Q there, then changing it back. The output was fixed after that. Hopefully there's an easy fix for the parameter file too.

I figured out the shelf thing. It was because I was using a version of the equations that only had one gain. The version used in sigma studio uses a linear gain too. The gain in the other equations is the same as what's being called boost in sigma studio. The parameters didn't change between highshelf and lowshelf because I was putting the gain into the the linear one instead of boost. With 0 boost the coefficients are the same. How I got some of the coefficients to line up doing that, I'm not sure.

The equations used in sigma studio are here except I think there is an error: https://wiki.analog.com/resources/tools-software/sigmastudio/toolbox/filters/general2ndorder?rev=1592943758

The high-shelf a0 middle term should be negative I think. Like this: "a0=(A+1)-(A-1)*cos(w0)+2*sqrt(A)*alpha"

Anyways, I still don't understand how to get the addresses to interact with the on DSP coefficient calculation filters from the header files. If you know how to do that, I would prefer to offload the calculation to the DSP. I see the line:

#define MOD_GSOCOEFFCALCULATION_0_ADDRESS_DSP_CALC0___FIXPT          0x00000076

But, it seems like this just points to the actual coefficients b2, b1, etc. All the other address lines are 0 like this:

#define MOD_GSOCOEFFCALCULATION_0_FREQ_DSP_CALC0___ADDR              0

Hello Gestalt,

I'm glad that you figured that shelf coefficients thing out yourself.

Regarding that param,h file, 

Yeah, I can simulate the bug. it indeed shows 0 as address for all the parameters in the file.

But you can get these addresses from the parameter window. Below is how to do this.

Go to view>> parameter window

and you can see the parameters and its addresses there.

Sorry for the trouble caused.

Hope this helps!

Regards,

Harish

Thanks. I was thinking there was some checkbox I missed or something. It seems it has to be hardcoded for now.