Effect of Mode on Filter Measurements

On February 5, 2009 in an email to the Elecraft List, Joe Subich, W4TV wrote: "The DSP begins to roll off at 200 Hz even if LO = 0.00. Any measurement where FC (Pitch) is less than [BW60/2 + 200] will be inaccurate on the skirts and any measurement where FC is less than BW6/2 + 200 will have essentially identical lower frequency responses."

This suggests that Pitch has a significant effect on filter skirt response and low frequency response. For example, if we are measuring a 500 Hz filter in CW mode with a -60dB bandwidth of 1,640 Hz, Joe's formula BW60/2 + 200 would indicate a Pitch less than 1,020 Hz (1640/2 + 200) gives inaccurate results on the filter skirts. As we demonstrated in Effect of Pitch on Filter Measurements, Joe is correct. Unfortunately, in CW mode the maximum available Pitch is 800 Hz.

To work around this dilemma we ran filter sweeps in Data A mode and compared them to CW mode. In Data A mode there is no Pitch setting, and the DSP bandwidth can be set to 4,000 Hz to avoid effects of DSP on filter measurements. We ran a test using Data A mode with DSP BW set to 4.0kHz and compared it to the results for CW mode at 800Hz pitch with DSP BW set to 2.8kHz. The first plot below shows the results on one graph for both Data A and CW modes. In the second plot I normalized the data so the curves would lie on top of each other. You can see the low frequency effect in the CW curve. The -60dB bandwidth will be significantly narrower than it should be if the filter sweep is run in CW mode. Thus, we recommend users of K3 Filter Tools run their filter sweeps in Data A mode to get the best representation of their CW filters.

Return to Software page.