Topic: Would someone like to build a little passive mixer for me?

[SparkE!]

Hey faninor,  I think it's you that's correct on this.

Chris is worried about signal inversion between the two mic sources.  A rise in instantaneous pressure at the mic capsule should provide the same polarity of voltage swing at the output of the mic.  If they produce the opposite polarity signal from each other, then they are 180 degrees out of phase with each other and contrary to what Chris says, you can't cure this problem with a simple time shift of one of the signals relative to the other.  You actually have to invert one of the signals and that cannot be done with a passive mixer unless that mixer uses a transformer for a mixing element.  The most commonly used convention is that a rise in instantaneous pressure on the mic capsule should produce a rise in instantaneous voltage at the output of the mic. So, if you send a puff of air at the mic capsule by saying a word that starts with a "P", then the resulting waveform should have a rising edge at the first of the recording of the "P".

So, as long as your mics have the same polarity of output, you should not have to worry about phasing issues if the mics are located at approximately the same distance from the sound source.

As far as placement of the mics go, sound travels about 1137 feet per second.  If you are using a sample rate of 44.1 kHz, your Nyquist frequency is 22.05 kHz and you'll get 180 degrees of shift at that frequency if you are just one sample off.  Of course the phase shift at lower frequencies for that same time shift is smaller. At 220.5 Hz, your phase shift would only be 1.8 degrees.  In one sample period, sound travels just over 3/10ths of an inch, so it does not take much mic separation to give significant relative phase shift between two mic source at the higher frequencies.  Then again, your own ears are about 17 cm apart in terms of the acoustic delay from one side of your head to the other.  Our brains use that phase shift at the higher frequencies to help us locate the sources of the sounds around us.

[faninor]

Ah okay, I hadn't thought about signal inversion -- that makes a lot of sense. I think my mics have the same polarity though so it shouldn't be a problem. I just checked and the microphones have a diameter of just about .25". If I allow just a very small gap between the two microphones then the discrepancy between the two should be between 0 and 1 samples, and by then placing them wisely with respect to the sound source I should be able to keep directions with a 1 sample discrepancy to places where it shouldn't have as important an effect.

[Lil Kim Jong-Il]

I'll add another vote for just getting new mics.  There are some people here who are looking to get in and you could probably sell both those pair and take a nice step up to another mic with the proceeds plus what you would put toward the mixer.   If you are trying to be low profile, the mixer isn't something you want to add.  Even for normal situations adding the mixer and second pair of mics is just adding potential points of failure on your chain.  IMO you will be lucky to mix the CSB and CSC mics and get a result that is not a lesser quality result than either mic alone.  All the guys I know who get good results from a four mic mix are doing the mix in post and appling filters.  I think you will be happier in the long run with a single pair of mics. 

[faninor]

I think I want to hang onto my Core Sound microphones even after upgrading, at least for a while... in the past I've loaned them out to friends, but the more expensive my microphones are, the less willing I'd be to take risks with them. I think I might try to build a mixer myself based off of the schematic I have and just try it out. In the past I've mixed recordings from these mics in post with good results. But sync in post has its own problems with drift and any way I go about fixing that has its own problems quality-wise. And even if there's absolutely no drift, because of the finite samples in digital storage, it is impossible to guarantee the accuracy of synchronization by more than 1/2 a sample -- so recording at 48kHz, the Nyquist frequency is 24kHz, but if the two sources you're mixing were recorded offset by half a sample from each other (or x.5 samples off from each other, your pick of x) then whichever way you line them up you'll still be half a sample out of sync, your 24kHz frequency will be 180 degrees out of phase no matter what. This is a real benefit of recording at 96kHz I think -- if you're just recording with a single microphone set at 96kHz that's great, but if you're recording from multiple mics and then mixing it together 96kHz should give enough precision to guarantee that all your audible frequencies don't get hurt in the mix. Do the guys you know record at a high sample rate? And do they use a multitracker or something that should take care of drift problems?

Given the physics of the situation and intelligent mic placement I think I'll be able to do just as well and most likely considerably better mixing it on the fly as I can in post recording at 44.1 or 48kHz. In the long run I'll be getting a different pair of mics though.

[Lil Kim Jong-Il]

Do the guys you know record at a high sample rate? And do they use a multitracker or something that should take care of drift problems?

Now they use 96K and timecode.  But it hasn't always been that way, it used to be multiple DAT sources.


I still think a single pair of better quality mics will give the best result but it is fun to build things so you might as well try it. 

[Church-Audio]

I think I want to hang onto my Core Sound microphones even after upgrading, at least for a while... in the past I've loaned them out to friends, but the more expensive my microphones are, the less willing I'd be to take risks with them. I think I might try to build a mixer myself based off of the schematic I have and just try it out. In the past I've mixed recordings from these mics in post with good results. But sync in post has its own problems with drift and any way I go about fixing that has its own problems quality-wise. And even if there's absolutely no drift, because of the finite samples in digital storage, it is impossible to guarantee the accuracy of synchronization by more than 1/2 a sample -- so recording at 48kHz, the Nyquist frequency is 24kHz, but if the two sources you're mixing were recorded offset by half a sample from each other (or x.5 samples off from each other, your pick of x) then whichever way you line them up you'll still be half a sample out of sync, your 24kHz frequency will be 180 degrees out of phase no matter what. This is a real benefit of recording at 96kHz I think -- if you're just recording with a single microphone set at 96kHz that's great, but if you're recording from multiple mics and then mixing it together 96kHz should give enough precision to guarantee that all your audible frequencies don't get hurt in the mix. Do the guys you know record at a high sample rate? And do they use a multitracker or something that should take care of drift problems?

Given the physics of the situation and intelligent mic placement I think I'll be able to do just as well and most likely considerably better mixing it on the fly as I can in post recording at 44.1 or 48kHz. In the long run I'll be getting a different pair of mics though.

Theory is nice... Experience tells me your not going to get good sound by placing an omni capsule right beside a cardioid capsule and mixing the signal together with out being able to have separate tracks to address the phasing issues, you have to remember the omni mic in "theory" pics up sound equally from all directions. The cardioid has more pickup on axis. The cardioid also has ports in the back of the capsule that are 180 degrees out from your omni mic.. The two patterns from the back lobe of the cardioid and the back lobe of the omni will mix and cause phase cancellation IN YOUR AUDIO not to mention that the arrival time WILL NEVER BE THE SAME of because your using two different types of microphones "strapped" together. I will be more then willing to bet your not "better off mixing it live" then you are with a multitrack and fixing it in post.

Chris
 

[SparkE!]

Mixing it live, he can align the sources to within one sample of each other.  Mixing in post, he'd be lucky to keep it within about 200 samples.  So, alignment of the signals won't be the issue.  However, getting the proportion of one signal's contribution to the other's could be an issue.  When you are mixing in post, you have the luxury of being able to experiment with different ratios of signal amplitude and choose the ratio that sounds the best.  Doing that mix live, you've got to get it right the first time.

[Church-Audio]

Mixing it live, he can align the sources to within one sample of each other.  Mixing in post, he'd be lucky to keep it within about 200 samples.  So, alignment of the signals won't be the issue.  However, getting the proportion of one signal's contribution to the other's could be an issue.  When you are mixing in post, you have the luxury of being able to experiment with different ratios of signal amplitude and choose the ratio that sounds the best.  Doing that mix live, you've got to get it right the first time.

Brad if he has separate tracks he can fix it. If not he is stuck with things the way they are. I would never do an experiment like this with out having the ability to treat the tracks separately. I dont know what your referring to when you say ( align the sources to with in one sample of each other..) I am not sure I follow you.

Chris

[SparkE!]

Mixing it live, he can align the sources to within one sample of each other.  Mixing in post, he'd be lucky to keep it within about 200 samples.  So, alignment of the signals won't be the issue.  However, getting the proportion of one signal's contribution to the other's could be an issue.  When you are mixing in post, you have the luxury of being able to experiment with different ratios of signal amplitude and choose the ratio that sounds the best.  Doing that mix live, you've got to get it right the first time.

Brad if he has separate tracks he can fix it. If not he is stuck with things the way they are. I would never do an experiment like this with out having the ability to treat the tracks separately. I dont know what your referring to when you say ( align the sources to with in one sample of each other..) I am not sure I follow you.

Chris

200 samples at 44.1 kHz is a little less than 5 ms.   5 ms is 1/2 of a cycle (180 degrees) at 100 Hz and if both sources were perfectly matched in amplitude for the on-axis signals, you'd get comb filtering with nulls at 100, 300, 500, 700... Hz and so on with nulls every odd multiple of 100 Hz.  This is one reason why matrix mixes often are done 70/30 or 60/40 and not at 50/50.  5 ms of skew across a recording is common when you use two separate sources.

If you place the two sets of mics so the both left channel mics are within about 1/3 of an inch of the same distance from their intended sound source, then you'll get less than 1 sample of acoustic delay between those two sources.  Yes, this is only true if both mics are pointed at the intended sound source and you will get phasing issues when you take into account the back pattern of the cardioid mic, but remember, that's precisely what makes a null in the pattern to the rear of a cardioid mic.  By mixing cardiod and omni, you are going to end up with something like a subcardiod pattern if the mics are located very close to one another.  In fact, I think you'll find that this is how some multipattern mics actually make their subcardioid pattern.

[Church-Audio]

Mixing it live, he can align the sources to within one sample of each other.  Mixing in post, he'd be lucky to keep it within about 200 samples.  So, alignment of the signals won't be the issue.  However, getting the proportion of one signal's contribution to the other's could be an issue.  When you are mixing in post, you have the luxury of being able to experiment with different ratios of signal amplitude and choose the ratio that sounds the best.  Doing that mix live, you've got to get it right the first time.

Brad if he has separate tracks he can fix it. If not he is stuck with things the way they are. I would never do an experiment like this with out having the ability to treat the tracks separately. I dont know what your referring to when you say ( align the sources to with in one sample of each other..) I am not sure I follow you.

Chris

200 samples at 44.1 kHz is a little less than 5 ms.   5 ms is 1/2 of a cycle (180 degrees) at 100 Hz and if both sources were perfectly matched in amplitude for the on-axis signals, you'd get comb filtering with nulls at 100, 300, 500, 700... Hz and so on with nulls every odd multiple of 100 Hz.  This is one reason why matrix mixes often are done 70/30 or 60/40 and not at 50/50.  5 ms of skew across a recording is common when you use two separate sources.

If you place the two sets of mics so the both left channel mics are within about 1/3 of an inch of the same distance from their intended sound source, then you'll get less than 1 sample of acoustic delay between those two sources.  Yes, this is only true if both mics are pointed at the intended sound source and you will get phasing issues when you take into account the back pattern of the cardioid mic, but remember, that's precisely what makes a null in the pattern to the rear of a cardioid mic.  By mixing cardiod and omni, you are going to end up with something like a subcardiod pattern if the mics are located very close to one another.  In fact, I think you'll find that this is how some multipattern mics actually make their subcardioid pattern.

Yes your talking about the difference between left and right.. What he wants to do is take a single stereo recorder and plug 4 mics into a mixer and blend them down the sample rate does not factor into the phasing issues.


The phasing issues will be there because although the back of the cardioid mics are 180 degrees out of phase from the front the omni mics are not. So when you combine them together side by side you will have phasing issues.  But dont take my word for it why not build this your self and see if I am right or wrong.. Its pretty simple thing to do. I have done it my self and I have noted the effects. The effects were not pleasant. The best way to make a sub cardioid is simply to block some of the vents on the back of the capsule. There are a few companies that use an omni and a cardioid to make a subcardioid BUT they do not put both mics side by side on the same axis.. If they did they would have a phasing nightmare of a sound.
When they do try and make a cardioid into a sub cardioid using two capsules one method is to place the omni capsule directly behind the cardioid capsule but the vents of the capsule are not facing backwards towards the omni they are facing sideways. Please forgive my very bad drawing Its pretty simple rear lobe of the cardioid 180 degrees out of phase give or take because its dependent on the fundamental + omni rear lobe in phase = phase cancellation and not at a predictable group of frequencies. Because the directionality of a frequency in free space changes as the fundamental gets lower.

Chris

[faninor]

If you place the two sets of mics so the both left channel mics are within about 1/3 of an inch of the same distance from their intended sound source, then you'll get less than 1 sample of acoustic delay between those two sources.  Yes, this is only true if both mics are pointed at the intended sound source and you will get phasing issues when you take into account the back pattern of the cardioid mic, but remember, that's precisely what makes a null in the pattern to the rear of a cardioid mic.  By mixing cardiod and omni, you are going to end up with something like a subcardiod pattern if the mics are located very close to one another.  In fact, I think you'll find that this is how some multipattern mics actually make their subcardioid pattern.

Yes your talking about the difference between left and right.. What he wants to do is take a single stereo recorder and plug 4 mics into a mixer and blend them down the sample rate does not factor into the phasing issues.

The phasing issues will be there because although the back of the cardioid mics are 180 degrees out of phase from the front the omni mics are not. So when you combine them together side by side you will have phasing issues.  But dont take my word for it why not build this your self and see if I am right or wrong.. Its pretty simple thing to do. I have done it my self and I have noted the effects. The effects were not pleasant. The best way to make a sub cardioid is simply to block some of the vents on the back of the capsule. There are a few companies that use an omni and a cardioid to make a subcardioid BUT they do not put both mics side by side on the same axis.. If they did they would have a phasing nightmare of a sound.
When they do try and make a cardioid into a sub cardioid using two capsules one method is to place the omni capsule directly behind the cardioid capsule but the vents of the capsule are not facing backwards towards the omni they are facing sideways. Please forgive my very bad drawing Its pretty simple rear lobe of the cardioid 180 degrees out of phase give or take because its dependent on the fundamental + omni rear lobe in phase = phase cancellation and not at a predictable group of frequencies. Because the directionality of a frequency in free space changes as the fundamental gets lower.

Chris

I'm not certain that the phasing issue of the rear lobe should be that great a concern. Isn't this just what gives the mic the pickup pattern that it has? And whether or not you're mixing live or mixing in post it is impossible to put the rear lobe of the cardioid in phase without taking the front lobe out of phase. I'm not sure what could be gained by mixing in post over mixing live (aside from a greater ability to select the volume of each source) if sound traveling from the source you'd like to record to the front lobe of either microphone is already in phase. In a couple weeks when I have my regular computer with me I will try to do a few tests to see what the actual results of what I'd like to try would be at a variety of frequencies.

[Lil Kim Jong-Il]

Why not filter the mic signals to eliminate the phase issues and then sum them? 

A basic HPF or LPF feeding a buffer is pretty simple.  With complementary pass filters and a summer you could isolate the desired ranges of each mic.   You could run your CSC tracks through a HPF and your CSBs through a LPF. 

[Church-Audio]

Why not filter the mic signals to eliminate the phase issues and then sum them? 

A basic HPF or LPF feeding a buffer is pretty simple.  With complementary pass filters and a summer you could isolate the desired ranges of each mic.   You could run your CSC tracks through a HPF and your CSBs through a LPF. 

Good idea.

[Church-Audio]

If you place the two sets of mics so the both left channel mics are within about 1/3 of an inch of the same distance from their intended sound source, then you'll get less than 1 sample of acoustic delay between those two sources.  Yes, this is only true if both mics are pointed at the intended sound source and you will get phasing issues when you take into account the back pattern of the cardioid mic, but remember, that's precisely what makes a null in the pattern to the rear of a cardioid mic.  By mixing cardiod and omni, you are going to end up with something like a subcardiod pattern if the mics are located very close to one another.  In fact, I think you'll find that this is how some multipattern mics actually make their subcardioid pattern.

Yes your talking about the difference between left and right.. What he wants to do is take a single stereo recorder and plug 4 mics into a mixer and blend them down the sample rate does not factor into the phasing issues.

The phasing issues will be there because although the back of the cardioid mics are 180 degrees out of phase from the front the omni mics are not. So when you combine them together side by side you will have phasing issues.  But dont take my word for it why not build this your self and see if I am right or wrong.. Its pretty simple thing to do. I have done it my self and I have noted the effects. The effects were not pleasant. The best way to make a sub cardioid is simply to block some of the vents on the back of the capsule. There are a few companies that use an omni and a cardioid to make a subcardioid BUT they do not put both mics side by side on the same axis.. If they did they would have a phasing nightmare of a sound.
When they do try and make a cardioid into a sub cardioid using two capsules one method is to place the omni capsule directly behind the cardioid capsule but the vents of the capsule are not facing backwards towards the omni they are facing sideways. Please forgive my very bad drawing Its pretty simple rear lobe of the cardioid 180 degrees out of phase give or take because its dependent on the fundamental + omni rear lobe in phase = phase cancellation and not at a predictable group of frequencies. Because the directionality of a frequency in free space changes as the fundamental gets lower.

Chris

I'm not certain that the phasing issue of the rear lobe should be that great a concern. Isn't this just what gives the mic the pickup pattern that it has? And whether or not you're mixing live or mixing in post it is impossible to put the rear lobe of the cardioid in phase without taking the front lobe out of phase. I'm not sure what could be gained by mixing in post over mixing live (aside from a greater ability to select the volume of each source) if sound traveling from the source you'd like to record to the front lobe of either microphone is already in phase. In a couple weeks when I have my regular computer with me I will try to do a few tests to see what the actual results of what I'd like to try would be at a variety of frequencies.

Again theory is a wonderful thing.. But again when you record with a cardioid mic you are absolutely capturing sound from behind the microphone that back vents are picking up sound that absolutely contributes to the over all sound of the microphone. Want to try an experiment block some of the vents and listen to the difference in sound quality. What is gain by doing things in post is you can rotate the cardioid microphone in time/space by knocking it out of phase say 20degrees or 40 degrees in relation to the omni signal. That's the benefit of separate tracks for all mics. You can manipulate the phase of each mic. With mixing on the fly you can adjust the phase of the stereo signal but that will not fix the phase relationship between two microphones place side by side and mixed to the far right or left, as you would have to with a stereo recording. Control of individual tracks means control of individual phase this means you can change the time/frequency relationship between the capsules.

Another good idea that has been suggested would be the use of hi pass and lo pass filters but filters also introduce phase issues of there own. These could work if you can get the slop past 6 db per octave, if not you will have too much overlap of the frequency bands and still have phasing issues. If you use an active 18db per octave or greater filter you can achieve what you want. You would simply do a fft on each mic find out where the response starts to get back and make that your cross point between the two mics. In effect your creating a two way transducer but your not using speakers your using microphones.

This has been done by AKG with a dynamic microphone and a condenser microphone called the c3000 where they used a cardioid and omni capsule one on top of the other and crossed it over very high around 8-10k and used the omni capsule for sibilance enhancement, because its very hard to get a large diaphragm microphone to have extended frequency response above 10k they used this little trick to make there 3000 microphone a much better mic in the high frequency range with out having to spend a lot of money making a better 1inch diaphragm.

Chris Church

[kuuan]

I had built various small pramps, had plans to build a mixer.
The best preamp in the end is the one of which I had taken out the level control potentiometer and substituted it with a switch for 2 levels. Why? Because all the pots I had bought were too bad and noisy. Panning pots are hardly available, good pots generally cost a fortune, this is why in the end, I never built the mixer.