Renderer Settings

To change renderer settings, click on the "Render" menu item and select the "Settings" submenu item.

The renderer settings are stored in the filter document file.

The user can control the data type, double precision, float, 32-bit integer, or 16 bit integer, values for the filter coefficients stored in the file.

For integer coefficients, overflow and underflow are much more of a concern. In general, the signed integer coefficients should be considered to be normalized from -1 to very close to 1, with the minimum negative value being -1, and the maximu positive value close to 1, by placing the sign bit after the first binary digit.

The user can control whether all coefficients are stored, or only half the coefficients, or for an odd-length filter, half the coefficients plus one more, i.e. the center coefficient. This can be done because all filters generate by this program are either symmetric or antisymmetric, i.e. the last half of the coefficients are negated. If only half the coefficients are stored, the filtering code will have to handle the antisymmetry. Currenlty only the Differentiator filter and the Hilbert Transformer filter produced by the Parks-McClellan design method will be antisymmetric.

Settings changes will only take effect if the "OK" button is clicked.

The "Render Type" radio button controls allow four selections.

Double - The filter coefficients are rendered as "double" precision values.
Float - The filter coefficients are rendered as "float" values.
32-Bit Integer - The filter coefficients are rendered 32-bit signed integer values.
16-Bit Integer - The filter coefficients are rendered 16-bit signed integer values.

The "Render Format" radio button controls allow three selections.

All coefficients - All filter coefficients are rendered to header file.
First half (+1) - The first half of filter coefficients are rendered to the filter file. If an odd number, the center coefficient is rendered last.
Second half (+1) - The second half of filter coefficients are rendered to the filter file. If an odd number, the center coefficient is rendered first.

If the "Discard End-Zeros" checkbox is checked, then any contiguous zeros at either the start or the end of the filter coefficient array will be discarded and the filter length shortened in the header file. Because the filters are always either symmetrical or antisymmetrical, this will always be the same number of values at the start and end of the filter. Because this can change the length of the filter, this can also change the group-delay of the filter.