High pass filter issue for sigma3.9 version.

Hello Paul,

Your two screenshots appear identical (both from SigmaStudio 3.9), so I'm not sure if you're referring to the coefficient calculations, their location in parameter memory, or a different problem. However, first-order HP filter info in version 3.10's help file does disagree with the Toolbox wiki at: https://wiki.analog.com/resources/tools-software/sigmastudio/toolbox/filters/general1storde

The wiki version is the correct one.

We can make a test by typing coefficients into the General Second-Order Filter, which works as a first-order filter when a2 = b2 = 0 --- its diagram reduces to this:

I was able to match responses with the stock First-Order HP filter by entering coefficients as shown below. Note that the formula for b0 should use e (2.718....), although the approximation 2.7 may work. Also the sign of a1 is always a bit confusing since the DSPs need to have their "a" coefficients negated. If the filter doesn't work, try flipping the sign(s) of a(n).

Best regards,

Bob

Hi Bob

I am sorry. I am fill another screenshots appear. Thank you for your help.

Thank you.

--Paul

--

Hello Paul,

We can see how SigmaStudio has changed since version 2.7 (the earliest I go back is ver. 3.2).

It's interesting that the old version set all three first-order HP coefficients the same except for sign. This goes contrary to what I see in books on the subject. However, plugging the "old" coefficient set into the General IIR filter the same way as above, also provides a proper high-pass response (see below):

Apparently, a change was made in a SigmaStudio revision after 2.7 that affects the coefficient calculations. I searched through the SigmaStudio Release Notes (included with the standard installation) and found this entry, which may explain the change:

So my advice is simply -- if the filter performs to your satisfaction, don't worry about it.

Best regards,

Bob

Hi Bob

Thank you for your help. Thank you.

Hello Bob

Can you explain elaborately why " Also the sign of a1 is always a bit confusing since the DSPs need to have their "a" coefficients negated" I see this in my parameter window ..I want to understand why this is a necessity