I use the off-axis calibration for my mic that was calibrated by Cross Spectrum. It makes more sense, as reflections act as a sort of secondary point source anyway, and there's reflections everywhere (as you have noticed, lol). Honestly though, the on-axis calibration works just as well, as there is very little deviation between the two, and only well up into the higher frequencies where absolute precision isn't required.

There's differing opinions on where to measure, with most people (I think) doing 4-8 measurements all over the head position, left, center and right for both channels. I prefer to point my mic primarily at the source of the sound because I've gotten far more consistent results, and better tunes that way. If I'm measuring, for example, the right channel, they would be like this: pointed directly R, 90 degrees away from the length of the car (2 measurements, one at top of ear level and one below), next I'll move to about 45 degrees away, and do another set of "top" and "bottom of the ear measurements. Then I point the mic almost, but not quite, straight ahead and take two more.

One thing I should point out: if you are measuring with your head in place, the head related transfer function (HRTF) will interfere with higher frequencies. That's not necessarily a bad thing per se, but it has to be accounted for and shouldn't be lumped in with non HRTF measurements for averages.

 

Honestly though, the on-axis calibration works just as well, as there is very little deviation between the two, and only well up into the higher frequencies where absolute precision isn't required.

What mics are everyone using? I just recently started taking actual measurements of my car using REW and a MiniDSP UMIK. I was curious to see if I would get different measurements by holding the mic at different angles but keeping the end in the same location. The variance was so tiny it's completely negligible. I would assume any good omni mic would have the same results.

So, like I said, I am new to measuring but it would seem to me that taking a measurement at the left ear, then the right and averaging their responses and adjusting as best you can would be the way to go. Maybe my thinking is off, but I would also think having both left and right channels playing while measuring both sides of your head would be ideal as well, considering all the reflections, both ears will be hearing both channels at the same time. So if you're measuring your left ear position, for example, you will definitely be hearing things there that are coming from the right channel, delayed of course, because of the reflections, but it will still be there.

 

So are you saying you are using the 90 degree off axis file and holding the mic next to your ear with the tip of the mic pointed up toward the roof. And taking a set of measurements? And then rotating your head for another two sets. Or are you using the 90 and pointing it directly at the speaker and then the other two? I find it easier to get the mic next to my ear when its straight up and down. I have seen this suggested by several people including helix.

 

I don't tune with my head in place because it's not necessary and only gets in the way. I tune form outside the car or the passenger seat. My speakers are all in the dash so my body doesn't get in the way (probably helps that I'm 130lbs soaking wet, lol). You can point the mic up, down or straight ahead. Try them all. There will probably be a slight roll-off above 10khz with a straight up or down position, but not enough to change anything drastically.

What mics are everyone using? I just recently started taking actual measurements of my car using REW and a MiniDSP UMIK. I was curious to see if I would get different measurements by holding the mic at different angles but keeping the end in the same location. The variance was so tiny it's completely negligible. I would assume any good omni mic would have the same results.

So, like I said, I am new to measuring but it would seem to me that taking a measurement at the left ear, then the right and averaging their responses and adjusting as best you can would be the way to go. Maybe my thinking is off, but I would also think having both left and right channels playing while measuring both sides of your head would be ideal as well, considering all the reflections, both ears will be hearing both channels at the same time. So if you're measuring your left ear position, for example, you will definitely be hearing things there that are coming from the right channel, delayed of course, because of the reflections, but it will still be there.

I agree with the bolded statement. As for mics, I've owned a ton of them. I measure with this mic, calibrated by Cross Spectrum Labs:

iSEMcon GmbH

I also have several DIY mics that I measure with, including one "shorty" mic/preamp unit I built specifically because it's hard to position a long mic wand in a car with a wall right behind the seat, lol. I had contemplated doing a small run of these mics but I just don't have the time right now.

As for your other question, tune whatever way makes sense to you, but if that's the way you do it you won't get results that are consistent with what everyone else is doing. Typically you tune one side at a time, with the other side completely off. By doing this, you get independent frequency and phase information, which you can then match to the other side. Ideally you want both sides to look exactly the same in both the time and frequency domains to establish the best soundstage (with perhaps some variance in the high frequency amplitude levels to prevent side bias). One you tune both sides, you can measure the combined result to see if the phasic interaction between the two is creating nulls or peaks in the FR. In a perfect world, the L and R measurements would overlay on top of each other perfectly, and the combined measurement would look exactly the same but slightly higher in amplitude.

Naturally you will hear L information from your R ear and vice-versa, this is known as crosstalk and will happen in every audio system except headphones. This is probably why some people use the "all over the head area" mic positioning technique. This gets into the discussion of dispersion (see: Patrick Bateman) and sound power (see: ErinH), and other areas too complicated to go over in this thread. I find it best to generally ignore the idea of crosstalk when tuning. Know that it's there, know that it's affecting your results, accept it, and go with it.

 

There are essentially three kinds of measurement mics. Pressure mics, free field mics and random incidence. Free field mics are the ones (or the measurement condition) for which you'd use the 90 degree orientation. Pressure and random incidence are essentially interchangeable. The inside of a car is not a free field, it's random incidence at high frequency and pressure at low frequency. In this case and with an omnidirectional mic, the regular pointing forward orientation is correct. Most of the microphones available at reasonable prices are pressure mics and the 90 degree calibration file is there in case you need to make a free field measurement. Such a measurement might be a ground plane measurement of a speaker in a box or in an infinite baffle outside.

So, the answer for this question is that it doesn't matter.

Measuring Sound with Microphones

 

So how different will measuring with the UMIK-1 pointed upward using the 90 degree file differ from straight forward using 0 degree calibration for the high end. I feel my tweeters are much brighter than shown pointing the mic upward. Probably due to reflections but I’m not sure how to see that info?

 

The difference between the two calibration files. Very small difference and only at VERY high frequencies.

 

Thank you for that clarification.

Ge0

 

Below is a quick graph of my UMK-1 USB mic for giggles and grins. The spec sheet says with a calibration file it is +/-1 dB but who knows for sure. Either way here is some data. Other mic's may measure better or not. between 0 and 90.

0 = on axis file
90 = 90 degree file
Diff = a difference between the two cal files.

Just some data for the discussion

 

Are both of these measurements with the microphone pointed up, just switching the calibration file?


Sent from my iPhone using Tapatalk

 

The files are the calibration files (offset values) for the microphone for doing measurements to a level of accuracy of +/- 1 dB supplied by Mini DSP.

 

Ah, nevermind. These are just calibration files graphed. Sorry!


Sent from my iPhone using Tapatalk

 

So does this mean that the mic at 90 degrees is MORE sensitive to the higher frequencies (hence the correction to decrease it more)? That seems counter to a lot of people’s experience where people feel that their top end is a bit bright when measured with mic pointing up and then get a reading with greater loudness with the microphone pointed at the tweeters...


Sent from my iPhone using Tapatalk

 

Just replying to myself here, but if the mic will actually not measure THAT different between the two calibration files (as Andy stated) then maybe this difference in calibration files from miniDSP is what accounts for what people (and I have also) experienced (with higher frequencies measuring lower than they actually sound)?

Or, in other words, the calibration files seem significantly different, but if they actually shouldn’t make much of a difference with this type of microphone in a car, then maybe this is what causes the phenomenon that people experience at the higher frequencies?


Sent from my iPhone using Tapatalk

 

Like, is miniDSP assuming people are using the 90 degree file in a free field application (NOT a car). I would think so, right?

Regardless, it would seem that the choice of calibration file makes a significant difference and we should all be using the zero degree file in our cars? (A 2dB difference starting around 7k is certainly something we could notice...)


Sent from my iPhone using Tapatalk

 

The calibration files show that there is a difference between 0 and 90 degree (pointing straight up) and it is not insignificant for the Mini DSP UMK-1 mic To use the 0 file while holding the mic straight up you would most likely get a substantial error above 2Khz according to the calibration file. All things being equal or according to the manufacture supplied data. If you use the wrong file with this mic the tune will become very bright in the top end as the wrong mic correction is used and extra frequency energy will be added by the person tuning to compensate because the calibration offset is not being applied to the microphone measurement before it is displayed on the RTA. Hopefully that makes sense.

 

The calibration files show that there is a difference between 0 and 90 degree (pointing straight up) and it is not insignificant for the Mini DSP UMK-1 mic To use the 0 file while holding the mic straight up you would most likely get a substantial error above 2Khz according to the calibration file. All things being equal or according to the manufacture supplied data. If you use the wrong file with this mic the tune will become very bright in the top end as the wrong mic correction is used and extra frequency energy will be added by the person tuning to compensate because the calibration offset is not being applied to the microphone measurement before it is displayed on the RTA. Hopefully that makes sense.

It does. Thanks! Just trying to combine this knowledge with what Andy is saying:

1) Andy is saying that a car is NOT a free field environment, which is the only place where a 90 degree calibration file should be used

2) Many of us, myself included, have been measuring with the mic pointed up and with a 90 degree file

3) this is incorrect and has been leading to potentially bright top ends.

Now should we all be pointing the mic at the speakers when measuring? I would think yes...

We should all be using zero degree files, right?

I admit different mics might be different, but it still doesn’t change the fact they the car is not free field...


Sent from my iPhone using Tapatalk

 

As an aside, or maybe related point?, I may disagree with your statement that it has nothing to do with the microphone orientation. I think it MAY have something to do with microphone orientation if that is causing you to use a 90 degree file.

Andy is saying that 90 degree calibration files are for free field measurement environments, which the car isn’t. Just because the mic is held upward or orthogonal doesn’t make the car a free field environment. In this case the 90 degree file still is not the correct choice.

Again, just trying to make sense of what Andy is saying and the use of these files in a car.

Totally agree that the use of the incorrect file can lead to unintended results...


Sent from my iPhone using Tapatalk

 

Some old photos from posts I could find of the measurement rigs of Skizer and another member (John Kiser?).

I believe these setups have the mics pointed forward.

Of course these photos do NOT mean that pointing the mic at the speakers is the only or best way to measure. I’m just trying to collect some expert opinions.

Are they aimed this way for a particular purpose / reason? I believe Nick’s mics are the ECM Behringer (an omnidirectional condenser mic) - so in a car, an environment that is NOT free field, the mic should be pointed at the speakers (and a zero degree calibration file should be used, no?)

At any rate just interesting. Again, people can measure anyway they like.

Sent from my iPhone using Tapatalk

 

I spoke with Nick about his mic array because I was interested in trying something similar. He mentions it not working the way he wanted and that he does not use it anymore. Rather, he sits in the car and waves the mic from ear to ear while taking long pink noise averages using REW. He mentions getting better results that way.

Interesting enough, I get the best results I've achieved so far by using this rig:

It waves the mic back and forth automatically while I take measurements. I use the sine sweep measurement function in REW. Take 5 to 10 measurements. Then average them together. I also use the mics 90 degree cal file.

Two important take aways here. The first is that this method is repeatable. Results do not vary wildly from one session to another. The second is it produces a natural sounding frequency response to me using a "Harman like" house curve. I may make slight tweaks from song to song using the tone controls on my head unit. But, that is more recording dependent that system playback dependent.

Ge0

 

The "difference" between the two curves above at 20k is not 0dB, its about 12dB. It appears you've ADDED the two curves rather than subtracting one from the other. For example, 6+(-6) = 0 but 6-(-6) = 12.

 

It looks like at 20k it should be 0 - (-6) = 6, but maybe I’m missing something?


Sent from my iPhone using Tapatalk

 

How many of you are Bluetoofing the stimulus signal to the head unit when you're tuning cars? How many of you have a pink noise track ripped to your iTunes or some other app that you use when you're tuning cars?

 

I use a pink noise track that I play through the Apple Music app. My phone is plugged in via USB to the car’s factory head unit. I play most of my music in this configuration.


Sent from my iPhone using Tapatalk

 

Maybe this is why I have a bright top end and sibilance issues. I was using the 90 degree file with mic pointed up. Strange how this seems to be touted by a lot of people as the correct method.

 

Just to throw it out there, I tune with pointing the mic straight up and waving it around my ears with infinite averaging in REW. I think the results are pretty great but I’m always interested in improving...especially if there is good reason to do it another way.


Sent from my iPhone using Tapatalk

 

If you're averaging while sitting in the car and waving the microphone around each side of your head, you'll probably require an adjustment to the target. You're going to measure less high frequency with the mic on the far side of your head and more when it's on the near side of your head.

I have a little array I've used for years that works really well with my target curve (which is the "harman" target curve because I used to work at "Harman". unless they've published something since I left).

 

If you're averaging while sitting in the car and waving the microphone around each side of your head, you'll probably require an adjustment to the target. You're going to measure less high frequency with the mic on the far side of your head and more when it's on the near side of your head.

I have a little array I've used for years that works really well with my target curve (which is the "harman" target curve because I used to work at "Harman". unless they've published something since I left).

I'd like to see some details on your array. I want to build something like Earl Geddes suggested in his white paper. But, don't have a software that can process multiple microphone inputs and average them. REW is powerful and free. But, does not support multiple mic inputs.

It's been a few years since I've visited Harman. To my knowledge nothing has changed with their system applications. Do you know a guy named Brad Hamm? He now leads the group and is a member of my local audio club. I can ask him at our next meeting.

Ge0

 

Just remember if you decide to set up some kind of an array, that the average of the micropohne measurements is not the sum of the measurements. You can't just get a 6 input USB interface and combine the output of the mics. You need software or hardware that provides an average. The spatial average is the average of the MAGNITUDE responses WITHOUT the phase response.

 

So use Periodic Pink Noise as the stimulus and RTA measurement. The microphones measure magnitude only. Each microphone will produce its own response curve. If I have software average these response curves it should result in the spatial average correct?

Ge0

 

Just as another data point, Helix also recommends the "mic pointing straight up and moving from ear to ear" method of measuring in their Sound Tuning magazines.

I also take measurements with the MIC pointing up, moving ear to ear and using the 90-degree calibration file - and I use a 20-minutes-long "Lab Grade Pink Noise" track that is in .wav format and played directly from my headunit (stored on a 256gb USB thumb drive that is always installed).

Ultimately, most people are only going to use the measurements for comparison to their other measurements, so I'm not sure it really matters. Personally, I don't follow any pre-determined curve. I started out tuning towards a curve, but quickly realized that I didn't like the sound of those curves and came up with my own curve - and honestly, I don't use a curve at all anymore. I basically just adjust the response to whatever sounds good to me.

I get extremely consistent measurements using the method described above (ear to ear, mic point up). Very quick and easy. I measure each speaker individually, each "pair" of speakers, each "side" and then the entire front (dash, doors and sub). Only takes about 10 minutes to get a complete set of measurements.

 

Totally agree. There is a point where your measurement equipment and process are “good enough” where the minutiae doesn’t matter too much. Also, you bring up a very important point that over time you have to become less restricted by a target curve and learn your tastes and your car. Still, my OCD likes to know the most correct way if there is one, especially if it could have saved me some trial and error and frustration at the beginning wondering why my tune sounded bright when others report that the target curve is pretty flat and not very lively. I’m sure this is a can of worms but hopefully you get what I’m saying.


Sent from my iPhone using Tapatalk

 

No, that still doesn't work because different frequency response has different phase response and that cannot be included in the sum.

 

Well respectively what explained is a basic delay and sum mic array used for decades.
The microphones relationship ( distance ) to each other needs to be determined and known. With that a time domain alignment is perfectly doable with in a range of sound source distances. The frequency response may be directly impacted by the microphone spacing and the polar plot of the microphones. All microphones should be matched to some acceptable tolerance otherwise it will be much harder to create a coherent signal

 

Well respectively what explained is a basic delay and sum mic array used for decades.
The microphones relationship ( distance ) to each other needs to be determined and known. With that a time domain alignment is perfectly doable with in a range of sound source distances. The frequency response may be directly impacted by the microphone spacing and the polar plot of the microphones. All microphones should be matched to some acceptable tolerance otherwise it will be much harder to create a coherent signal

So... Mmmm. Ahhhh... No. Just no.

I was going to go there. But its Friday night. I'd rather sit back and have some beer and pizza while watching a Scifi movie with my boys . I'll start thinking again tomorrow.

Ge0

 

Stupid question probably but are we supposed to use the foam "cover" on the tip of the microphone. I've got a Umik 1 and in the package it came with a foam piece that goes over the tip. Just wondering, does it help eliminate wind noise while moving the mic around for example?

Sent from my HD1905 using Tapatalk

 

D@mn. How fast are you moving your mic . I'm going to go out on a limb here and say no. The foam piece is supplied to cut down on the affect of wind turbulence. Hopefully you are not creating wind turbulence while waving your mic around. I can picture wild and violent jerky actions .

Ge0

 

LOL, no I'm not swinging the mic around like a baton, just wondering if that was it's purpose or not. You've got a vivid imagination! [emoji28]

Sent from my HD1905 using Tapatalk

 

Thank you 😀. I'm glad I could make you crack up . It kind of lightens the mood around here...

Ge0