Best B/W Setting for 60 & 120 Hz Notch Filters

[dre95060]

How wide should one make the notch filter (lower value of Q) for filtering hum? I've tried the minimum of .01 octave--which is somewhat effective--but I still hear some hum. At what point are we throwing the Stradivarius out with the noise?

Dave

[Craig Maier]

The smaller the bandwidth setting, the less adjacent signal that you are throwing out along with the target. Narrower Bandwidth results from higher filter Q.

I always tune the Bandwidth of the Notch filter to be as small as possible while still attenuating the Hum. Often, I put the filter in "Slot" mode and tune for maximum hum and then switch it back over to notch mode. If the hum contains any "frequency smear" (frequency modulation due to wow, flutter or any other frequency variant effects), you will require a wider (or larger value of Octaves) bandwidth to attenuate it. I am usually pretty happy when I can achive my desired results with a bandwidth of around 0.1 octaves. I would say that 0.01 Octaves is generally too narrow to be effective with most hum signals, unless they come from a digital source (which are quartz time based).

[Dan McDonald]

and sometimes what you are hearing is 120 or 180, which makes it sound like the 60 Hz is not wide enough. At least that's what I find with a lot of material I work with (live recordings especially). You get multiples of the 60 Hz that are often louder than the original sound. So I try to use multiple filters with smaller bandwidths if that's the case. Like Craig says, use the filter as a slot, then back to notch. If it sounds right with the slot, but the noise is still prevalent, then check the multiples.

[Craig Maier]

An alternative to Dan's suggestion of using multiple notch filters in the Multfilter, is to try the Harmonic Reject filter. You can dial in the fundamental (say 60 Hz) and a number of harmonics up to 500. Most of the time, 2 or 3 harmonics run with both odd and even settings (two runs of the filter) gets all of the hum. Or, you can set up two harmonic reject filters in the multifilter, one for odds and the other for evens.

If you are dealing with "Buzz" then many more harmonics may be required - - - I have used up to 50 at times.

[Dan McDonald]

Hi Craig -

Yes, I use the Harmonic Reject sometimes also, but sometimes the presets for it a little harsh for the material, so I developed a modification of one of the presets to get the first group of 3-4 multiples of 60 hz. I was pretty happy with it, but I lost that one when my hard drive was wiped out about a year and a half ago.

Dan

[Craig Maier]

There is a difference in the DSP technology used between the harmonic reject filter and multiple notch filters. The Notch filter uses IIR techniques whereas the Harmonic Reject filter uses a combination of IIR and FIR techniques. So, the results between the multiple notch method in conjunction with the multifilter and the harmonic reject filter can produce different sounding results depending on the situation.

[Dan McDonald]

Hey Craig -

The Harmonic Reject works really well with really bad buzzes you get with very old recordings (like from old open-reel home units from the 50s or from the home wire recorder that I have). The Harmonic reject tends to clear a lot of that up pretty easily.

[dre95060]

Thanks Again...

I get an education every time on this board. Thanks, guys! I will try that slot filter trick. I haven't had great luck with the harmonic filter except for the first two or three harmonics. But I suspect that is highly dependend on the source material.

I have found the spectrum analyzer very useful. My source recording shows very clear peaks at 60 and 120 Hz.

Dave

[Ringmaster]

I've had very good results using the EzClean "Hum" filter. Due to the way the filter works, (I've forgotten the name of the sensing feature); Logically, this should not completely delete the "notch".

Am I on the wrong track?

[Craig Maier]

The EZ Clean Hum Filter(s) are basically multiple Notch filters coupled with some sensing routines.

[Ringmaster]

Craig, there is a name you used somewhere for the sensing routines, but I couldn't remember what it was; but, this is what made me think that this filter would not completely replace the "notch".