Comb Filtering - Part 2
Be sure to read Part 1 - before jumping into Part 2
Important! For all of the following, do not use any effects that alter pitch and tone, meaning pitch shifters, wah-wah, etc. If you will use such effects, plug them in after completing the comb filtering tests.
You should open the previous article for important details as both articles directly relate.
Comb filtering got its name because the waveform looks like a hair comb. This is caused by one sound source and two (or more) mics picking up the sound when the mics are not the exact same distance in relation to the sound source. In this set up we have two mics positioned in front of a guitar amp speaker. Let's say both mics are fairly close to each other and one mic is positioned one inch back from the speaker grill cover while the other mic is positioned three inches back from the grill cover.
The mic signal that is positioned three inches from the speaker will arrive later in time at the mixer compared to the mic that is only one inch from the speaker. The difference is two inches. Since sound travels around a millisecond per foot, there will be a delay of around .2 milliseconds between the two mics.
With both mics set to approximately the same level, if looking at the combined mic signals on a dual trace scope or on a hard disk recorder format with waveform track readouts, the combined mic signals waveform will look like a hair comb. This is because there are major buildups in some frequencies and major suck outs in others caused by a very short time delay between signals.
Try This Cool Audio Test and Trick
If you want to hear what this sounds like, using one mic on a guitar amp, record a simple E cord and let it sustain for 10 seconds. After recording, mult off the guitar track to a delay line that allows for at least single millisecond steps of delay. Assign the delay line to an unused mixer module and match the levels of both the guitar track and the delay line. Set the delay line to one millisecond. You will hear that the frequencies sound very strange. Now set the delay line to two milliseconds. Again, the sound will be strange but with a different set of frequency build ups and suck outs. You get the drift. If you use an oscillator to shift the delay time, you will hear the flanging effect! The comb filtering effect stops when you delay the signal to the point of hearing two distinct attacks. This starts around 40 milliseconds or so of delay.
Back to the Article
OK, in our set up, if using one speaker, we have two mics that may be positioned fairly close to each other. This set up will make things easy when fixing the comb filtering.
We may be using two different speakers and possibly two different speaker cabinets. With that in mind, the frequencies will not be exactly the same since speakers surely have different tonal characteristics. Hopefully not a problem when dealing with the comb filtering fix.
Yes, there are always exceptions, such as using two separate amps/two separate speaker bottoms. Let's say on one amp, the tone is set with full bass and no mids or treble. Let's say the other amp the tone is set to no bass, some mids and major treble. In this case, the mics would be seeing two signals that would produce waveforms that do not cross each other (frequency wise) so even if the mics were not positioned at the same distance from the sound source, the comb filtering effect would not really be in play. Yes, this example is strange but there are no rules when recording! The Comb Filtering fix
If you're using a digital recorder format, and if you're using two mics to two recorder tracks, the fix can most likely be done after recording. Even so, you may want to do the following test to avoid the time shifting later.
Two mics recorded to one recorder track
After setting the basic mic placements and you are happy with the sound, the concept is to move one of the mikes to equal the same distance from the sound source as the other mic. The mic to be moved could be either but I tend to move the mic that is farther back in comparison and move closer to the speaker to match the other.
The routing has been set up for either two mics to one recorder track or two recorder tracks. For two mics to one recorder track, mixer module #9 and #10 are the mic sources and the recorder track is track #7. Recorder track #7 is monitored through mixer module #7. For two mics to two recorder tracks, mixer module #9 and #10 are the mic sources and the recorder tracks are #7 and #8. Recorder track #7 and #8 are monitored through mixer modules #7 and #8. Pan mixer modules #7 and #8 to center and set their faders to around 10 dB on their fader throws. We are setting up for a phase test.
If you have set up the path in full and have gotten a good guitar sound, and if a compressor is in the chain, bypass it for now. Also switch out the EQ for the time being. We simply want the mic path unaltered.
You also need to mark mixer fader settings since we will change levels. Put a piece of artist or masking tape next to the mixer faders used in the mic path and recorder return path. Mark the tape (using a sharpie pen or any marker that can easily be seen) next to the little pointer on the fader so as to be able return to those settings later.
If you have set guitar amp levels, tone, etc., we are going to move the settings for the test. Put a piece of artist or masking tape next to the controls and mark with a sharpie pen or any marker that can easily be seen. If the controls do not have a little pointer also put a piece of tape on top of the control and mark both pieces of tape with a little arrow pointing towards each other.
The guitar could be used as the sound source but since the signal dies off in sustain land fairly quickly, its better to use a non-periodical waveform sound source such as your favorite artist CD or cassette tape.
Plug the CD or cassette player output into the guitar amp input. Since the CD or cassette player is a line level source, its most important to make sure the guitar amp input volume is off to start. If the CD or cassette player has outputs that allow the volume control to be active, use as to not overload the guitar amp input. We do not care about the stereo output meaning use either the left or right output only. Note that if your amp has an effects send and return, the odds are huge there will be level controls. In this case, plug the CD or cassette player output into the effects return jack and set the effects return level very low as to start.
Play the CD or cassette and slowly bring up the guitar amp volume. In any case, all we need is a volume level that is around an average listening level. You may want to adjust the guitar amp tone control settings if the sound is too dark. The idea is a to make sure to hear a wide range of frequencies. After you have set the controls to taste, hit stop on the CD or cassette player for now.
Back to the mixer. The idea is to move one of the mics to the point of hearing signal cancellation. This is achieved by flipping the phase of one of the mic mixer paths and setting both mic levels (and recorder return levels if using two records tracks) to the exact same levels.
If your mixer modules have phase switches, you could switch the phase on either mixer module #9 or #10 since all we want to do is swap either. Lets reverse the phase on mixer module #9. Doing so will lead to signal cancellation when the levels of both mics are the same and the mics are positioned at the same distance from the sound source. Note that full cancellation may not happen for many reasons but the idea is to find the point of the most cancellation.
(Important: if your mixer modules do not allow for phase reversal, the fix is to wire up or alter a mic cable inverting the phase. If a two wire mic cable, on one end of the cable at the connector, reverse the hot and ground wire. If three wire mic cable, on one end of the cable at the connector, swap pins 2 and 3. Make sure to label the cable inverse phase!!!)
Play the CD or cassette player. If using two mics that will be recorded to one recorder track, set both mixer modules #9 and #10 to around 10 dB on their respective fader throws. Very slowly move mixer module #9 up and down so as to listen for the cancellation spot. After finding the level setting with the most signal cancellation, time to move one of the mics regarding the distance to the guitar amp speaker. Again, I typically move the mic that is positioned farther away.
If you are working by yourself and can get to the guitar amp wearing headphones monitoring the mixer signal, simply move one of the mics until you hear the most signal cancellation. Now go back to the mixer and slowly move mixer module 9s level up and down as to hear further signal cancellation. Repeat this over and over until finding the point of the most signal cancellation as possible.
If you are working with a 2nd engineer, or someone to help you, all the better. The 2nd engineer or helper needs to wear headphones hearing the mixer signal you are hearing. The 2nd engineer or helper will hear what to do meaning hearing the point of cancellation or more build up when moving the mic (if the guitar amp is not too loud in the studio). While the 2nd engineer or helper is moving the mic slowly, keep moving the level on mixer module #9 very slightly up and down looking for the spot of the most signal cancellation. Use the mixer talk back to communicate with the 2nd engineer or helper. When hearing the most signal cancellation, ask the 2nd engineer or helper to stop moving the mic.
The mic needs to be secured as usual so refer to the Part 3 Miking The Guitar Amp article. Make sure the mic is not moved when doing so!
Now flip the phase on mixer module #9 back to normal. Reset all mixer module settings, amp settings, etc. Since the mic was slightly moved, you may want to revisit EQ settings and a slight level change with that mic.
Most engineers do not take the time to perform the test. I have blown it off when in a hurry and have ended up with a sound that could have been better.
Fixing after recording two mics to two recorder tracks
Using a hard disk recorder with waveform track readouts: The easiest way to fix comb filtering requires a hard disk recorder format that allows sample accurate timing offsets. After recording the guitar mics on individual recorder tracks, the fix is to look at the first attack of both guitar tracks and time shift the late track waveform to match the track with the first attack. Make sure to shift the total length of the track meaning from beginning to end.
Using a stand-alone digital recorder that allows sample accurate track timing offsets: In that case, you will need to experiment using track delay timing offsets as to start. More on this as things unfold.
Reverse the phase on either recorder return mixer module #7 or #8. (If your mixer modules do not have phase switches, see the above section and look for a paragraph that starts with, (Important: if your mixer modules do not allow for phase reversal). In this case, you will need to substitute a cable from the recorder track output that is wired reverse phase for one of the two recorder outputs.) Mute all other instrument/vocal tracks meaning only listen to the two guitar tracks. Set both mixer modules #7 and #8 to around 10 dB on their respective fader throws. Also pan both mixer modules to center. While listening to the playback, very slowly move mixer module #7 up and down as to listen for some or total signal cancellation. If hearing some signal cancellation, or none, in either case, time to experiment with track delay offsets. Start by delaying track #7 in five-sample steps. Again, very slowly move mixer module #7 up and down as to listen for some or total signal cancellation. When the signal cancellation is becoming more noticeable, go to single-sample steps. Experiment with single-sample delay settings until you hear the most signal cancellation between the two tracks. Go past the point of hearing the signal cancellation and then work in reverse (backwards) meaning use single sample delay steps in reverse (less track delay). When finding the general area of cancellation, go back and fourth with delay settings until finding the track delay setting with the most cancellation. Write down the amount of delay samples since we will need these numbers later. If you are not hearing any signal cancellation and the sound becomes two distinct attacks after using delay numbers that are in the thousand of samples, track #8 is the track that needs to be delayed. In this case, repeat the above steps track delaying track #8 instead. By now, you have discovered the track that needs to be delayed and have found the best cancellation point. You have also noted the track delay setting.
So what happened? If track #7 needed to be delayed, we know that track #7 is sounding first. If track #8 needed to be delayed, we know that track #8 is sounding first. No matter what, you get the logic. Let's assume the following fictitious layout as to set up the time shifting fix.
Track #7 was delayed 18 samples in comparison to track #8. We want to get track #8 in the future by 18 samples as to make up for the delay.
Time to put the late guitar track in the future. This is cake if using an ADAT recorder format (or a similar format) using two or more recorders that have a sample accurate time lock.
You have written down the track delay setting. (Our fictitious settings are noted above.) Take the first ADAT recorder out of track delay mode. Digitally record (bounce) tracks #8 to an open track on another ADAT recorder. Best to use a blank tape on the 2nd ADAT recorder to be safe. We need to digitally bounce this track back to its original track (ADAT recorder #1) offsetting the bounced ADAT recorder in the future. This is performed by putting the 2nd ADAT recorder in machine offset mode. In our example, we know that we need to get track #8 in the future by 18 samples. Offset ADAT recorder #2 to minus 18 samples this allows the 2nd ADAT recorder to send the bounce 18 samples early correcting the delay problem!!! The 2nd ADAT machine offset setting has been set so now digitally bounce it over to its original track on the first ADAT recorder. Check playback to make sure the bounce was recorded correctly. If you want to play things very safe, meaning not recording over the original track; record the offset track to an open track. Mark this track as the comb filtering fixed track and mark its original as do not use and also note the offset time used on the track sheet this allows you to go back and change offset times if needed. Now take the ADAT recorder used for the offset out of machine offset mode. Now switch the phase back to normal on the mixer module that was phase reversed.
Important: If using any digital recording format, its always smart to back up the tape or hard disk data at least once per session. Yes, this a time burner but trust me on this if you do not back up your stuff often, the odds are very good you will loose recorded information at some point in time. This happened to me a few times when working with drum samples and I was backing up every few hours! The sinking sick feeling in my stomach will never be forgotten. This caused only a few hours of extra work but I was pissed off! Yes, I always say no rules but take this rap as a rule!!!