Same speakers and room measured using ARC3 and REW

[Foxenfurter]

I have been tinkering with DRC using IK Multimedia ARC3 using their MEMMs microphone and REW using a very cheap uncalibrated measuring MIC (Subzero from Gear4Music costs about 20 GBP).

I finally got an opportunity to setup and measure my main speakers chain being laptop -> MOTU M2 -> Purifi EVAL 1 -> Opera SP3 Speakers. The room is in a London Victorian Semi-detached house with suspended flooring bay window dimensions 4m x 4.5m with 3m high ceiling. Furniture in the corners, fireplace between speakers and a rug on wooden flooring.

Speakers are Opera SP3 (Super Pavarotti 3) - floorstanders. Tweeter and 2 x 6.5 inch mid/bass drivers.

First measured using ARC 3 using their semi automated approach and then REW using MMM method.

I have tried to align the charts as best I can.

First ARC3

Then REW

I am assuming that the Arc measurement is slightly more accurate because the process is more idiot proof and the mic is effectively calibrated. But for simplicity I am using the impulse filters generated by REW (this is because ARC only lets you use the correction via their VST plugin, which isn't feasible as my streamer is a raspberry pi.

Any observations or suggestions appreciated.

 

[abdo123]

Is the tweeter blown up? What the hell is happening above 4KHz?

 

[Foxenfurter]

Well that is partly why I posted this.
There is no obvious damage and the speakers don't sound dull when comparing to my iloud monitors or various headphones I own. My hearing is good to about 12 K, but I was surprised to see the dropoff. The Arc measurement is about 6db down at 16k whereas REW is about 15db - so that difference could be down to the cheapo microphone that I did the REW measurements with.

 

[Foxenfurter]

I have been doing some testing using the arc system with the MEMS microphone as a reference and then repeating the measurements using both the MEMS with no calibration and the sub-zero with no calibration

These are my iloud monitors. Green lines being measurments. This is using the inbuilt calibration of the Arc system.

I then repeated the measurement using the MEMS microphone but this time as an external MIC

Notice that the microphone presents a flatter response, but this is not accurate.

Finally I repeated the exercise with the sub-zero

Notice the bass measures a hot and the treble cold vs the MEMS,

I then pasted these images into a spreadsheet and measured the distance vs DB and offset vs the Original MEMS plot. and created calibration files for both the MEMS and Subzero assuming that the ARC system was a good reference.

In theory ARC supports calibration files, but in practice whilst you can upload them, they have no effect. So I then moved over to REW and re-measured the speakers using the Sub-zero MIC and the calibration file I had just created.

I expect that there are some differences between the smoothing mechanism used as well as slight differences in microphone positioning. Also the graph scale is slightly different.

 

[Hipper]

ARC seems to work well whatever mic you use - except over 10kHz which as you say may be a limitation of the cheaper mic.

I use REW and a well thought of mic - the Earthworks M23 - but concentrate my adjustments (using a Behringer DEQ2496) on the 0-300Hz range. Above that I just use wider adjustments more like tone controls with some changes for more old ears hearing deterioration (can't here above 10kHz and weaker around 6-8kHz).

Whilst some of these measurements may look nice, I see ARC, REW, MiniDSP etc., various mics, and measuring methods, as tools to get you a better sound but the final arbiter are your ears.

 

[kyle_neuron]

At £199 for the ARC3 package, I'd be very surprised if there's any real difference in the response between the microphones from 50 Hz to 14 kHz - at least at levels below 100 dB linear SPL.

It's probable that there's a mic correction curve, fixed windowing, or even a high pass filter loaded when the software knows you're using the bundled mic, but even then the deviations should be far less than 3 dB either way. Unless the mic itself is faulty or subjected to relatively extreme changes in temperature or humidity. What microphone preamp is being used?

Have you checked that it can provide a consistent phantom voltage? Some of the bus-powered interfaces don't achieve 48 Volts unless hooked up to mains power, and some microphones don't play nicely with a lower voltage. Likewise for the impedances, which can limit the maximum SPL which can be measured with a certain combination of mic and preamp. It could be that there is clipping occurring at the preamp, even at relatively low output levels, if the gain isn't adjusted accordingly.

A 'fair' test would be to use the different mics in the same software too. There's no reason the ARC3 mic shouldn't work with REW, which would give a nice comparison of the relative responses if you can get the mics at the same physical position. Even a few millimetres out can make quite a big difference in a room, so a nearfield measurement of the drivers would also be a good start.

I'd expect to see much more deviation at the very high-frequency region on microphones up to the real expensive types, so a close mic measurement of the tweeter is a nice and easy test. You also don't have to upset the cats and dogs in the neighbourhood, since you can get away with lower sound levels

 

[Foxenfurter]

Hi - Arc software automatically calibrates when measuring with their own microphone, but if you select the third party option then the load of calibration file does not do anything. The interface I am using is a Motu M2 and it supplies 48 volt phantom power, the mic doesn't work without it.

I read on another forum (AVForum) I think that cheaper Mics often measure poorly above 10K and also found a calibration for the MEMS microphone that someone had made.

Just to close the circle - I have downloaded some open source room measurement software called Cavern. and used their quick EQ option and this has given me results that compare well with ARC but can easily be used in EAPO or Inguz on my LMS.

I will attach the calibration files for MEMS mic and Subzero that I created from this exercise.