What if you don't have calibrated microphones?

[Blumlein 88]

Okay a bit messy. This is a Umik 1 calibrated mic plus some musical condenser microphones. 3 cardioids and 2 omni. Source was a speaker 12 feet away in my video setup which is why the uneven response. I tried to keep the capsules as close to the same exact spot as possible. You are looking at 1/6th smoothing to make it legible. However even without smoothing zooming in all the mics pretty much agree. The cardioids depart somewhat below 200 hz and more so below 100 hz. Proximity effect.

I'll explain later how I calibrated those without using the calibrated microphone I have. Though I used it to check the results. The Umik is red btw. The uniformity below 100 hz even leads me to wonder about the Umik below 50-60 hz.

In any case this works out better than I expected. So following my rather simple home calibration I believe most condenser mics could be utilized to good effect if you don't have a calibrated measurement mic to work with.

The other mics were a Shure KSM32, Shure KSM27, Audio Technica AT4033 which are cardioids. Then an Avantone CK1 omni and CAD M179 omni. Though not shown the CK1 and M179 also have cardioid settings and fit right in with what is shown.

More to follow.

 

[Blumlein 88]

Here is the Umik 1 (red) compared to a Shure KSM32 (green) with 1/24th smoothing. Above 100 hz a pretty good match.

 

[amirm]

That's great work. But I don't understand how the model response changed so much in low frequencies with Umik1 being the outlier. Here is similar work that Ethan did on measurement mics: http://realtraps.com/art_microphones.htm

What is your take on your measurements?

 

[Blumlein 88]

That's great work. But I don't understand how the model response changed so much in low frequencies with Umik1 being the outlier. Here is similar work that Ethan did on measurement mics: http://realtraps.com/art_microphones.htm

What is your take on your measurements?

I don't have an explanation. I intend to repeat the Umik later and make sure there wasn't some mistake with it. I have seen Ethan's article.

It would make no sense to purchase any of the mics I tested for this purpose as they all cost more than a Umik or several other measuring microphones. I was mostly wondering how much they differed from a calibrated microphone. I was surprised they ended up as close as they did. Which made me think if someone already had these for recording purposes they could be pressed into service for some measurements if needed. When I get time later I'll go over how I calibrated them. I also have some others I didn't include because the graph was getting too crowded, but they don't differ from those I put up much either. Now uncalibrated these are not very close to the measurement microphone with one exception.

 

[Blumlein 88]

Okay, repeated the measurements and mea culpa. Don't know what happened. Maybe a passing distant train I didn't notice. So here is yesterday's Umik and todays Umik as well as the Umik done at 90 degrees off axis. Yesterday's measurement in red. A good reason for replication.

Here is a repeat of my initial posting with the new Umik measure in place amongst the others.

Here are the three omnis I have on hand. Avantone CK1 omni, CAD M179 omni and Umik1. The M179 and Umik are the two greens and the gold color is the CK1.

Finally here are the three cardioid micrphones. They all show a bit of a roll off below 200 hz no doubt due to proximity effect which omnis don't suffer from.

 

[Blumlein 88]

I was surprised the microphones ended up so close to the Umik1 calibrated mike. Especially that all three omnis matched so closely. Even more so when you read how I calibrated these other microphones.

I simply grabbed graphs of the response from the makers and created calibration files for REW from those. Such graphs are maligned for being smoothed and being written by the marketing guys. Especially one like this for the Shure KSM 32. They were a little optimistic below 100 hz, but apparently these kinds of response graphs are for real.

Some are more bumpy like for the Audio Technica At4033a. Yet when corrected based upon this the results line up well when an uncorrected graph would be considerably different.

I have pairs of most of these other microphones. Another surprising result is each pair though some were made years apart drop on top of each other almost identically. The difference being about equal to what you might get running the test twice on one microphone a few minutes apart.

If you ever have reason to make your own cal files for REW, WebPlotDigitizer is free software that makes it a piece of cake.

http://arohatgi.info/WebPlotDigitizer/

It runs as a Chrome add on, or in a browser window, or an app in Chrome OS or a downloaded offline version for Windows.

I can grab an image of the graph and spit out a .csv cal file in two minutes or less. REW takes .csv as well as .txt files for calibration.

Take a few minutes to learn the software and it is pretty handy. You open an image file. Pick two points on the x axis and on the y axis. Select linear or log scaling. Set the size of the scales. Then you have two choices. You can use a mouse to place points on the curve as many as required to follow the curve. If you can find graphs like those above where the curve of interest is a different color than the grid it is even easier. You take a paint brush and paint a wide line over the curve, then use a color picker. The software then places plenty of points on the curve of that color and spits out a file with the values in the .csv format ready to drop straight into REW as a cal file. It is easier than it sounds. There are a couple of 9 minute youtube tutorials by the author of the software. The software also can do much more like polar graphs or mapping.

Recording hacks is a good place to find graphs for nearly any microphone all in one location. They have a fairly extensive listing of microphones with descriptions, basic specs and usually graphs of response.

http://recordinghacks.com/microphones/

 

[amirm]

Ah, all is well with the universe again. Thanks for all the work. It is so easy to make mistakes and posting them online is a good way to find such faults.

 

[Wayne A. Pflughaupt]

Did you have the UMIK’s calibration file loaded in REW before you took its measurement? Did you remove it before you measured the other mics?

So here is yesterday's Umik and todays Umik as well as the Umik done at 90 degrees off axis.

Was the 90 degree calibration file loaded? If not, response should start to droop below ~ 2 kHz.

While it appears that the results from the non-calibrated mics can get you a decent idea of your speakers’ response, it should be noted that manual equalizing based on a non-calibrated graphs is not a good idea. There is as much as 3-4 dB deviation in some places between the various mics.

Regards,
Wayne A. Pflughaupt

 

[Blumlein 88]

Did you have the UMIK’s calibration file loaded in REW before you took its measurement? Did you remove it before you measured the other mics?


Was the 90 degree calibration file loaded? If not, response should start to droop below ~ 2 kHz.

While it appears that the results from the non-calibrated mics can get you a decent idea of your speakers’ response, it should be noted that manual equalizing based on a non-calibrated graphs is not a good idea. There is as much as 3-4 dB deviation in some places between the various mics.

Regards,
Wayne A. Pflughaupt

The direct cal file was loaded for the Umik 1. It was pointed at the speaker. If you notice above, first graph in post #5, the green one is the new measure aimed at the speaker with the direct cal file. The blue one is pointed at the ceiling with the 90 degree cal file loaded for that one. The red one had something causing noise in the low end it appears when it was done.

Certainly I think for the cardioids one is best keeping to response above 200 hz. While there are some differences in the mics they are smaller than the peaks and dips in the speaker itself. So not perfect, probably in my opinion close enough to be helpful if one had nothing else. Now the omnis did much better being effectively identical below 200 hz. So if one made their own cal files like I described and stuck with omnis they are unlikely to go wrong by much below the Schroeder frequency fixing the low end.

Also I didn't intend that one doesn't need a calibrated mike. Just tested these out of curiosity and found they might be useful with care should one wish to do so.

 

[amirm]

When we (JJ) put in room EQ in Windows Vista, you could use any microphone you wanted. You can get a lot done even if the overall response is not accurate. Then there was a checkbox to indicate you had a measurement mike and it would do a broader correction.

 

[DonH56]

Many vocal mics have a "presence boost" that emphasizes the upper midrange. Good for voices cutting through the band, bad for EQ work. Large-diaphragm condensers tend to be generally flatter though that is not always true, and many mics have LF rolloff (intentionally). A friend had always used his Shure SM81 for reference and we were surprised by both how flat it was (compared to your average recording mic) and how far off it was compared to my Earthworks M30, a true measurement microphone.

Note that it is hard to get decent 90 degree response from anything but a small-diaphragm omni. More important in the HT world, perhaps, but I prefer using an omni pointed at the ceiling for quick room measurements. Last time I helped with a pro room measurement we used fancy SW that stitched everything together very nicely from (many) individual measurements but I don't have $10k to spend on that... Also note a 90 degree response is not a complete panacea; any reflections/emphasis/changes added by the ceiling signal will become part of the measurement and can be difficult to de-embed.

 

[Wayne A. Pflughaupt]

Also note a 90 degree response is not a complete panacea; any reflections/emphasis/changes added by the ceiling signal will become part of the measurement and can be difficult to de-embed.

Indeed. There was a case at the Shack some years back where (I assume) the REW user had a reflective ceiling, and a 90-degree measurement got him a graph with unnaturally hot high end response.

Regards,
Wayne A. Pflughaupt

 

[Blumlein 88]

Many vocal mics have a "presence boost" that emphasizes the upper midrange. Good for voices cutting through the band, bad for EQ work. Large-diaphragm condensers tend to be generally flatter though that is not always true, and many mics have LF rolloff (intentionally). A friend had always used his Shure SM81 for reference and we were surprised by both how flat it was (compared to your average recording mic) and how far off it was compared to my Earthworks M30, a true measurement microphone.

Note that it is hard to get decent 90 degree response from anything but a small-diaphragm omni. More important in the HT world, perhaps, but I prefer using an omni pointed at the ceiling for quick room measurements. Last time I helped with a pro room measurement we used fancy SW that stitched everything together very nicely from (many) individual measurements but I don't have $10k to spend on that... Also note a 90 degree response is not a complete panacea; any reflections/emphasis/changes added by the ceiling signal will become part of the measurement and can be difficult to de-embed.

Yes, all good points. That is why I pointed these music condensers at the speaker not at the ceiling. And the response was adjusted according to the published response graph of the microphone which it appears is generally reasonably accurate. That mostly took out the presence boost in these. If you ask me it would be better practice to have flat mikes and add the boost yourself in post production. That isn't how it is done however. A common complaint against the KSM32 which is pretty flat with only a hint of boost is that is "merely" gives you on the recording what is in front of it without cutting thru the mix or having "good character". It certainly is easy to adjust such a boost if you desire it tailor made for your vocalist if you used the KSM32. The Umik is an SDC omni of course as was the CK1. The others were either MDC (around 3/4 inch) like the AT4033 or KSM32 while the M179 and KSM27 were LDC mikes.

BTW the cal curve on the Umik shows a presence boost of around 3 db and an off axis dip of about 2 db.

 

[Fitzcaraldo215]

I guess possiby workable solutions abound, if you are trying out some EQ software and have no suitable omni mike that can successfully interface with it. But, if you can get a UMIK-1 for $75 or for $100 with the custom individual calibrations, why screw around and waste your own time with someting else? The Umik-1 isn't perfect either, but it is a step in the direction of doing it right with a widely used and recommended tool with an omni mike that has been calibrated within reasonable tolerances.

 

[tomeh]

View attachment 5937

Okay a bit messy. This is a Umik 1 calibrated mic plus some musical condenser microphones. 3 cardioids and 2 omni. Source was a speaker 12 feet away in my video setup which is why the uneven response. I tried to keep the capsules as close to the same exact spot as possible. You are looking at 1/6th smoothing to make it legible. However even without smoothing zooming in all the mics pretty much agree. The cardioids depart somewhat below 200 hz and more so below 100 hz. Proximity effect.

I'll explain later how I calibrated those without using the calibrated microphone I have. Though I used it to check the results. The Umik is red btw. The uniformity below 100 hz even leads me to wonder about the Umik below 50-60 hz.

In any case this works out better than I expected. So following my rather simple home calibration I believe most condenser mics could be utilized to good effect if you don't have a calibrated measurement mic to work with.

The other mics were a Shure KSM32, Shure KSM27, Audio Technica AT4033 which are cardioids. Then an Avantone CK1 omni and CAD M179 omni. Though not shown the CK1 and M179 also have cardioid settings and fit right in with what is shown.

More to follow.

Hi,

I just came across your great post while looking for some CAD M179 info.

I approached this in the slightly different way a few years ago as well. The only change was that when Logic Pro came out with thier Match EQ plug-in, I used it instead of REW.

I ran pink noise through one of my monitors, lined up 16 channels of mic's, closely packed which included a Behringer BM8000. I used the Behringer as the "standard" and used the Match EQ plugin to sample the others and it creates a corrective EQ profile for each.

It allows you to apply a percentage of "correction" and also a minus %. So applying a minus 100% correction gave me a decent profile for each mics response.

This corresponded very well with what I heard for each mic and applying about 67% correction made them match pretty well.

What it couldn't do was make each mic's polar response match of course. It was a one trick pony with it only match well when the source hit each one at the same angle, position, etc.

What it did lead to was a better understanding of why it is best to use the Behinger measurement mic at 60 degrees to the speaker.

I look forward to comparing your responses to what I got with the mic's that overlap. The CAD M179 and one of the Shures.

Cheers and thanks.

Tom eh

 

[dominikz]

Perhaps I can add another datapoint to this excellent thread:

Some explanations in this post.

 

[Blumlein 88]

Perhaps I can add another datapoint to this excellent thread:

Some explanations in this post.

You could try creating a Cal file using this FR chart and get it even closer.

RØDE NT2-A | RecordingHacks.com

NT2-A, Multi-Pattern Condenser Microphone , aka: NT2A, NT2 - detailed microphone profile, specifications, manuals, reviews, frequency-response graphs, self-noise data

recordinghacks.com