Measuring Room Response - what is a good mic?

[mace]

Hi,

Attached is a graph which shows the response curves for various things:

Aiwa = Aiwa CA-30 boom box (from 1984) in bedroom. 

Enos = My car stereo, measured in my car

KH = stock Klipschorn, with and without sealing the corner pieces with tube insulation

CW = stock Cornwall, two very different speaker positions (W1 and rv)

(RadioShack digital SPL meter, held at listening position in all cases. Warble tones 1/3 octave from Audiophile test CD2. The number after the variable labels is the dB level of the 1 kHz test tone.)

What I notice is the general pattern all of these are similar. Peaks at around 5-10 kHz, dip at around 500-1000 Hz, peak at about 100-200 Hz. I'm interpreting this as a fingerprint of the meter since these are from very different sources and rooms. So, I tend to think my RS meter isn't very accurate. 

My question: What is a good microphone to use to measure SPL? What I envision is simply recording the mic directly to my computer and analyzing the resulting AIF file with something like Audacity (free program) to get peaks or RMS or whatever for each warble tone. I've use a Shure 58 instrument mic and then calibrated the results (per the freq. response graph which came with the mic). It seems a lot more reasonable with the Klipschorn results (see next post). 

I've read through the old posts on this board on room response measurements. What I'm proposing here is something different... it seems the RadioShack meter has a distinct signature response no matter what you measure. 

Thoughts??

Mace

 

response curves all.pdf

[mace]

Here's the graph with the Shure mic vs. Rat Shack...

(maybe I should have posted these in the chart room thread...)

Mace

RS meter vs. Shure vs. seal.pdf

[rcarlton]

Mace,

Try the Home Theater Shack. They have correction curves for the RS SPL meter.

[mace]

Here is the calibration table I used on the Rat Shack curve for the Shure-RS graph. I did not use any calibration of any kind for the "all response curves" graph. 

The calibration table I used only says to add everything, which just makes the peaks more pronounced. It still seems there are some issues with the RS meter, IMHO. 

This calibration table really doesn't change much except at the extremes. I followed your link and didn't see a calibration table directly. Is there another table, different than the one I got off the internet many moons ago? 

Thanks,

Mace

Re- Radio Shack SPL Meter.pdf

[rcarlton]

Mace,

They have the calibration table on the link. Just have to search. You might try Room EQWizard on the link. It will do a nice job of capturing the graphs for you.

I use a Behringer ECM8000 mic.

[mace]

Thanks for the links and mic info! I registered at that site and am waiting on the confirmation email. I didn't realize at first that the correction table was in a special file format (not raw text).  

Mace

[DrWho]

I also use a Behringer ECM 8000 calibrated mic runing through a Behringer UB 802 mixer (to provide the necessary phantom power). To make the measurements, I'm using Room EQ Wizard with a set of calibration files I picked up from the Home Theater Shack forum. Total cost was $100 with a 50 foot mic cable and free shipping [Y]

You can see a bunch of the measurements I took here:
http://picasaweb.google.com/mebentz2/Measurements/

It's a very cool program once you get it calibrated. I highly recommend reading the online help file as it will walk you through everything.

[mace]

Hi,

I think I'll get that mic. I have a little M-Audio box with phantom power which I use for other stuff when I record to computer. 

The Room EQ Wizard software seemed to run OK on my Mac under Java so I just need to figure out how to really use it. 

I'm curious, however, if the ECM 8000 is a "calibrated mic" why do you need "calibration files"? 

Thanks,

Mace

[DrWho]

To correct for a slight roll off above 10kHz and an ever so slight rolloff below 20Hz...just the extremes of the measurement.

[mace]

Right.

If you look at the graphs I posted above you'll see the same pattern, no matter what the source. Maybe it has to do with the C-weighting or something. I noticed that when I imported data into the REW software it asked if I had run the data through a C-weighting compensation or some such. Anyway.. maybe this thread is best served on that HTS board!

Thanks again for the links,

Mace

[DrWho]

The last time I measured my room with an SPL meter, it looked like this:

[DrWho]

If you'd like, you can have the excel spreadsheet I used to automatically implement the Radio Shack meter compensations:
https://netfiles.uiuc.edu/mebentz2/Frequency%20Response.xls

[WMcD]

Regarding calibrated mikes and calibration files.

It is my thought that the microphones are not necessarily perfect at all. However, the seller will compair them to a known near flat microphone's response to a sweep of frequencies, or they will use a known flat sweep.

By this the imperfections of the microphone sold will be known. Then the software file will compenstate when the microphone is used in conjunction with a measuement program.

For example, the LMS system sells a "calibrated" microphone with serial number on it. They also give you a microphone calibration file with that serial number. You load the calibration file into the software which controls the hardware card.

As you can imagine, this allows any run of the mill microphone to act as a near perfect one. There is still a small plus or minus window though.

Gil

[mace]

Thanks much for the spreadsheet. I'll have closer look on the weekend. It seems that the adjustment values in your spreadsheet are quite different than the ones I have been using (which I attached the PDF in a post above).

There are so many variables with this rat shack meter my head spins. First, there are several different calibration tables. Then, there is C-weighting and some sort of C-weighting compensation when analyzing results. Plus, I really question how accurate the thing is based on the fact all my curves have similar trends. Maybe it is user error!

Concerning "calibrated mics"... does the ECM8000 come with it's own unique calibration table? In other words, is each mic tested and the values for each specific mic provided in the packaging?

Thanks much,

Mace

[DrWho]

No, the ECM8000 does not come with a unique calibration table. It does come with a frequency response chart, but it's not a measurement of the specific mic you recieve. The calibration file I used was provided by someone on the HTS (home theater shack) forums. He sent his mic out to some dude who measured its frequency response and created the calibration file. I would assume that all of their mics would be fairly close to being within spec. The absolute extremes of the frequency response aren't going to be very accurate anyway due to room acoustics.

This should hopefully clarify the A and C weightings on your meter:
http://www.sfu.ca/sonic-studio/handbook/Sound_Level_Meter.html

The reason REW (room eq wizard) asks if your readings were accounted for C-Weighting is so that it can undo the intentional change in the measured frequency response. If you note the chart in the link, you should see that the C weighting affects frequencies lower than 125Hz and higher than 2kHz. So if you used an SPL meter with a flat frequency response, but on a C-weighting, then REW will correctly reverse the C-weighting filter and yield a true frequency response. REW will however not compensate for deviations in the frequency response of the meter (for which the Radio Shack meter is notorious).

If you wanted to use your radio shack meter in REW, then you have two options. One, you can take your manually entered data, apply the correction values I provided and dump them in REW - telling it that it wasn't a C-weighted measurement (so don't apply compensation). The other option is to take your manually entered data and dump it straight into REW. Tell it that it was a C-weighted measurement (make the compensation) and then inside the program you can apply the compensation file you found at the HTS forums.

In other words, the correction values I supplied will turn the C-Weighted measurement directly into a flat frequency response. The calibration from the HTS forums only compensates for the mic response, but REW gives you the option of compensating for C-Weighting (making the program versatile with all SPL meters, since a different calibration file can be loaded for every meter). In other words, REW treats the mic response and the weighting response differently. If they wanted, they could add an A-weighted compensation too, but most (if not all) SPL meters have both A and C weightings.

I hope this helps.