[timandmonica]

I have been studying REW video tutorials and the manual for days. I initially had problems getting things working but have finally made it through these steps:

Calibrated soundcard via feedback loop
Calibrated SPL, balanced mic input to match pink noise
Did the sine sweep to measure the room
Generated EQ correction curves to compensate for my room

But when I click the "Predicted" graph, it looks NOTHING like I would expect (and nothing like the example I saw on Youtube, which made the new curve relatively flat.)

I had a few issues when doing the calibration.

One was that b/c you need to put a mic in front of a speaker, and that mic'd sound goes back into the system and back out the speakers, I was having a problem controlling feedback. I turned my SPL down to the min. recommended amount but I still heard a small feedback resonance (at an exact frequency) sometimes. I did the sine sweep over and over until I thought it recorded no feedback but I can't be sure. Why would REW play the mic'd pink noise back through the speakers instead of just listening and measuring?

The other was that I saw conflicting opinions of how to angle the mic. I have a Behringer ECM8000. I saw opinions that you should 1) Point it up to the ceiling, 2) Point it directly between the speakers, and 3) Point it between the speakers at a 30 degree angle upwards. I chose #3. But wouldn't doing it any other way give a different reading? If so, how would one know which one is correct? It's an omni mic but placing off-axis like that should change the frequency response, shouldn't it?

So keep those two points in mind when you see my graph. Maybe I did something wrong that made this strange predicted frequency fix happen. I've attached both a JPG and the entire data set. The purple is the "Predicted" results.

Thanks for any help offered!

 

[Wayne A. Pflughaupt]

Welcome to the Forum, Tim!

One was that b/c you need to put a mic in front of a speaker, and that mic'd sound goes back into the system and back out the speakers, I was having a problem controlling feedback.
Why would REW play the mic'd pink noise back through the speakers instead of just listening and measuring?

That’s not an REW issue, it’s a sound card issue. You need to turn off the monitor or set it for “Computer.” If your sound card doesn’t have such a provision, you’ll need one that does.

The other was that I saw conflicting opinions of how to angle the mic. I have a Behringer ECM8000. I saw opinions that you should 1) Point it up to the ceiling, 2) Point it directly between the speakers, and 3) Point it between the speakers at a 30 degree angle upwards. I chose #3. But wouldn't doing it any other way give a different reading? If so, how would one know which one is correct? It's an omni mic but placing off-axis like that should change the frequency response, shouldn't it?

I feel your pain, I’ve even seen one fellow advocate pointing the mic at the floor. :dizzy:

I’m not sure where the idea of upright or upside-down mic orientation came from, but pointing the mic at the signal source for frequency response measurements has been the professional standard since the dawn of room measurement technology, and examples to that effect can be seen in vintage literature from measurement-mic manufacturer B&K, as well as the manuals for professional hardware RTA equipment. In fact, the venerable RTA hardware manufacturer Goldline still makes the same recommendation in their current-production devices.

Your question didn’t mention anything about calibration files. A mic pointed at the ceiling requires a 90° calibration file. Without it high frequency response will sag.

But even with a 90° file, frequency response results with upright measurements can be unpredictable, as seen with this thread. There’s no reason not to try both ways – pointed at the speaker or at the ceiling (again assuming the correct calibration file for each), but if they get markedly different results, then the 0° measurement is the one to trust.

For the record, Herb Singleton of Cross Spectrum Labs and REW author John Mulcahy have generally recommended 0° orientation for in-room frequency response measurements of main-channel speakers, and 90° for acoustics measurements.

Regards,
Wayne

 

[JohnM]

That’s not an REW issue, it’s a sound card issue. You need to turn off the monitor or set it for “Computer.” If your sound card doesn’t have such a provision, you’ll need one that does.

I'm not so sure about that - soundcard cal looks fine (so no monitoring when that was done).

Main issue is the response has no smoothing applied, so comb filtering is obscuring things. If REW's variable smoothing is applied things start to look more sensible, it is best to apply the smoothing before asking REW to match the response to the target. The target level is also set a bit low, should probably be 60 dB, and generally better off not applying EQ above a couple of kHz at most or the result may be very sensitive to position, try measuring at slightly different positions to see how consistent the response is.

 

[rvsixer]

...For the record, Herb Singleton of Cross Spectrum Labs and REW author John Mulcahy have generally recommended 0° orientation for in-room frequency response measurements of main-channel speakers, and 90° for acoustics measurements.

Five years ago when I started on this DIY/REW journey and purchased a Cross Spectrum calibrated EMM-6, it was recommended to me to use 90 degrees for in-room measurements (which is all I have used since). And now moving on to a USB UMIK-1 and HDMI PC setup, and just now read in the latest "Getting Started with REW: A Step-by-Step Guide", still pointing out 90 degrees is the angle used for most REW measurements. I had always thought 90 degrees made sense, as one would want to capture reflections etc.

And now I come across this thread, and the recommendation to use 0 degrees. Also I am confused that 90 degrees is for acoustics measurement, are LP measures of in-room main-channel speakers not acoustic measurements?

I can certainly see using only 0 degrees for near-field speaker measurement, but what about LP only measurement? 0 degrees or 90 degrees for all LP testing, or 0 degrees for L/R and 90 deg for C/SR/SL/LFE (which does not seem correct at all, but what I am getting from the above quote)? Very confused at this point :help: ? Thanks for any clarification as I set up my new measurement system.

 

[thrillcat]

I can certainly see using only 0 degrees for near-field speaker measurement, but what about LP only measurement? 0 degrees or 90 degrees for all LP testing, or 0 degrees for L/R and 90 deg for C/SR/SL/LFE (which does not seem correct at all, but what I am getting from the above quote)? Very confused at this point :help: ? Thanks for any clarification as I set up my new measurement system.

I would imagine that one big advantage to using 90º mic placement is that you can leave the mic in the same position to measure all speakers in a system. Would this not make comparisons between pre- and post-eq much more accurate, since the mic can remain in position? If you use a 0º placement, the chances you will be able to reproduce that placement perfectly are very slim.

 

[rvsixer]

I would imagine that one big advantage to using 90º mic placement is that you can leave the mic in the same position to measure all speakers in a system. Would this not make comparisons between pre- and post-eq much more accurate, since the mic can remain in position? If you use a 0º placement, the chances you will be able to reproduce that placement perfectly are very slim.

Exactly my thoughts, and confusion about using/switching to 0 degrees. The lab recommended 90, and the REW guide recommends 90. BUT, I can also see why to use 0 degrees in certain situations.

Well my UMIK-1 came with both 0 deg and 90 deg files, this weekend will see if there is any trending difference between the two orientations.

 

[Wayne A. Pflughaupt]

Acoustics measurement are RT60, Energy-Time, etc.

Regarding the “Getting Started” tutorial, please note that AustinJerry participated in the thread linked in my previous post. So he’s seen what you can get with 90° frequency response measurements yet doesn’t seem inclined to change the Guide.

I imagine 90° orientation was started by one or a few “early adapters” to software measurement programs who had no professional experience in the field. Possibly they were simply measuring subwoofers, where orientation wouldn’t matter anyway. They put their experiences up on the internet and it just spread from there.

Another example of the type, back when the BFD was getting popular for EQing subwoofers, one or two well-respected and otherwise knowledgeable parties participating on a few home theater forums were telling everyone that it wouldn’t perform at its best unless the input levels where pushed as high as possible. Pretty soon everyone was repeating that stuff and you were seeing it over the place.

All that aside, it is indeed convenient to measure multiple speakers with a single mic placement. If you use the proper calibration file and see no difference in your room between 0° and 90° orientation, then there’s no reason not to use the latter.

Regards,
Wayne

 

[AudiocRaver]

An additional thought or two:

The smaller the mic tip diameter and body size, the better the response off-axis. The Beyerdynamic MM1, for instance.

But - as Wayne P. says - if you have the right calibration file you are going to be very close. At that point, in practical terms you are picking at details that don't matter much and the convenience of the 90-degree mic is hard to beat. My opinion.

 

[bazsound]

mic definatly shouldnt be going back through the speakers, this is going to really mess thi ngs up.

Look at the mixer and in playback mute or turn down the mics ouput.

 

[rvsixer]

Thanks Wayne and AR for your insight and comments.

I did install linux REW on my HTPC with the new UMIK-1 this weekend, and a bit disappointed with the results (sharp roll off past 12kHz, at both 0 deg and 90 deg). Unsure if its a setup and or UMIK issue at this point, I will pull out my old windows laptop/EMM-6 I previously used (which showed flat response well out past 20KHz), and conduct a measurement experiment with the only difference being the mic and associated cal file.

Will post any further disappointing results in a more suitable forum as to not continue thread semi-hijack :blink: .