Measurement Mic Shootout (EMM-6, WM-61A, RS 33-2055, Audyssey) - Page 3

How much does that cost?

I have Galaxy cm140's, a couple of ecm8000's including a calibrated one, about 3 audyssey mics and a matched pair of calibrated Earthworks m30's which don't require a calibration file until below 9hz and past 20khz. For normal measurement I dont use one with them. When I get time I will look at the comparison. I do not expect all of the mics to match up very closely at the extremes.

Notnyt Im not saying you did anything wrong. What myself and Bosso are saying is that if your calibration is right on that mic and everything matches so closely then a cheap audyssey mic with no cal is basically dead flat till 5hz. That just does not make any sense. Btw I don't know this for sure but the same Panny capsule or similar might be in most of the units. That's just a hunch though.

My desktop soundcard has a rolloff built in to the microphone input. Different input jack for the microphone input on my desktop PC's internal soundcard.


Pink noise sound source for the microphone input and for the line input. Red is the FR for line in, and green is the FR for microphone in.

Here's the EMM-6 mic against the ACO and extrapolated Aud mic from my exercise:



I'm tellin' ya, my extrapolated curve is darned close to what the actual uncorrected Aud mic response is.

Also, notice the reference mic Dayton used.

Bosso

You can start telling regular users to plug in their audyssey mics to their notebooks, then use REW to get some idea of their room response...interesting stuff indeed.

It begs the question, why in 2011 are we just finding this out? What other historical data exists on this topic.

My microphone input is filtered, so that is a no go for me. My PC's microphone input does not produce "flat / accurate" results.

I don't have an Audyssey microphone either!

No clue.

I bought mine in 2003-4. The UA1000 was around $600 and the PS9200 ACOPacific kit was around $1650.00.

The Edirol UA1000 is long discontinued and I have no idea what the ACO kit goes for today.

Here's the catalog and kit with options, including the SPL calibrator. You could call and inquire. The guy knows his stuff and is cool to talk to about measuring audio.

http://www.acopacific.com/acopaccat.pdf

Main site products page:

http://www.acopacific.com/products.html

Bosso

You mean the reference microphone that Cross Spectrum Labs uses to generate "caliibration" files for a Dayton microphone.

Considering that the calibrated Dayton measured less than 3 dB lower than the reference microphone at 10 Hz, the difference in "accuracy" is hardly worth thinking about.

I PM'd Herb (Cross Spectrum Labs) over at HTS earlier today and got the following response:

"I took a quick look at the thread and I'll say from the start that Audyssey mics are all over the place in terms of their frequency response so you absolutely cannot generalize based on the performance of one mic (they have individualized correction curves in the hardware that flatten the mic response).

Out of the box, the accuracy of the ECM8000 (or EMM-6) doesn't hold a candle to a pro mic from ACO Pacific, B&K, GRAS, etc. Those pro mics are flat down to 1 Hz or so. The cheaper ECM8000/EMM-6 mics have high-pass filters that cause a roll-off starting around 100 Hz which severely degrades the low-frequency response. The correction files I generate compensate for this so in theory, which a correction file, you can get a ruler-flat response down to about 5Hz. That said, there is still a bit of error since the measurement procedure I use has some error, plus the amount of correction required for those mics below 10 Hz (which can be as much as 15 dB) may introduce some noise or distortion that can affect the measurement."

So what conclusions can be drawn from the data we have here?

Well, first - no one asked me....

Second, old RS meters are evidently not nearly as accurate as new RS meters. Lots of respected folks measured them and showed us the results years ago. We've all relied on that since.

Third, and IMO, the most important.....

Anyone with a Windows PC or laptop, a 30 dollar USB sound card, a simple measurement mic, and REW can measure things extremely accurately with just a little bit of effort. This capability did not exist at all 10 years ago, and to be honest, has only become widespread in the last 3 years or so. This is the new and important part.

None of the mic technology is really new. There is a lot of information about the Panasonic electret capsules out there, and the Linkwitz Mod to these electret capsules is not exactly new news either, being dated 1988. Wallin mentioned it in his meter mods back in the 90s. Another mention of the panasonic capsule is the Mitey Mike from Audio Amateur, that's also back to 1990.

The low-cost availability of fast laptops and fast low-latency USB audio cards are probably the fundamentals here. The laptop I use for measurements is a refurbished Dell I picked up for $300, all in, shipped to me. It has plenty of CPU (2.2 GHz, dual core), hard drive (120 GB), and RAM (2GB), and came with Windows XP Pro at that price. Install some free software (REW & Holm), order in a Behringer UCA-202 for $30, and build yourself a lilmike mic for $10. You're pretty well set to handle any acoustic measurement a DIY sub or speaker builder needs to make. That sort of capability was not available for $340 5 years ago. Cobble up a proper set of test leads and you can do impedance and Thiele Small parameters for another $10 too.....do it wrong and you'll be ordering in another UCA-202....

I only brought up the idea of an accurate measurement mic because watching people apply close to 30 dB of "correction" factors that they downloaded from the Internet to their measurement sweeps, then claim frequency response extending into to the single digits (when their measurement hardware can not possibly resolve that) sorta runs against the grain of the "Science" in this site's name.

I figured that if people couldn't find another $10 and an hour of effort to measure the results of weeks of enclosure building and thousands of dollars worth of subs and amps and a bit more accurately, they were at the wrong site, reading the wrong forum. Sorry if that hurts feelings, it is not intended to.

Simply put, data that has 30 dB of correction applied to it is not exactly accurate in my eyes. 30 dB is HUGE, that exceeds the range of a typical analog SPL meter by a factor of 2. There is no need for this.

Not calling anyone out, not naming names, or anything like that. I really just wanted to clear the air and hopefully elevate everyone's game for the cost of a pitcher of beer. A little bit of direction and education does a long way.

Honestly - this has turned out a LOT better than I had hoped. Above 10 Hz, nearly any of mics presented is accurate enough to be trusted, and clearly, modest measurement gear will get us there. Certainly - extending the low end below 5 Hz separates the toys from the tools, but I'm not claiming I can accurately measure data that low, nor do I really care to. I'll leave that corner of the DIY sandbox for the bigger kids to play in.

Big thanks to all that have participated, and huge thanks to notnyt for helping me kick this off, as well as taking things as far down the road as he has. I'll have more to add once things calm down for me a touch. I have stock and modded RS 33-2050s, as well as my EMM-6/Xenyx to add to the mix.

you know what, i am going to give this a try.

i have an htpc, an audyssey mic, and an old RS analog meter. but i don't have a sub.

can i measure my mains instead? if so how? is there a tutorial?

ok, downloaded REW and tried to run it but keep getting "level to low" warning.

any suggestions?

got my audeyssey mic hooked up to the "line in" input on my htpc.

volume on my integra 9.9 is at -20 (which is quite loud for me).

edit: also switched the mic to the "mic in" input on the htpc, still same message.

i have absolutely no idea what i am doing. need to read up.

"Big thanks to all that have participated, and huge thanks to notnyt for helping me kick this off"

very interesting thread guys...+1

True.

Before REW and just after purchasing my measurement gear (around the end of '03, beginning of '04) I bought Pure Bits Sample Champion, who had just extended their softwares capability to 1 Hz per my ragging.

Just about that time they posted their comparo of a B&K, Sennheiser and the Panny WM-61B electret, which is still partially on their site:

http://www.purebits.com/miccomp.html

He measured to 10 Hz:



So I was aware of the possibility, but preferred to go as low as I could. I also corresponded with the SC author on the subject so I was equally aware of the importance of quality fantom power/interface, etc, to successful results.


Double rolleyes for this one, J.

How would you know it measured -3dB at 10 Hz without the reference mic? That's the whole point of using one, isn't it?


That's just it. Yes, the Aud mic does need correction to match your corrected EMM-6. Both my measurement and Audyssey's own specs and graphs say so.

I'd really like to get to the bottom of that.

Maybe they took the FR pages down because they use a different mic with a different roll off? Doubtful.

Maybe you got a freak Aud mic? Doubtful as well.

I just don't know at this point. Suggestions and comments, please?

Bosso

thanks, that's good info. but sadly, didn't fix my low level thingy.

giving up for now. will look at it another day.

thanks, i'll pm you when i get back to this again.

this is getting a bit complicated, and i thought i am good at following instructions.

i did turn the vol up to max in recording device, got 5 green bars (or was it 6?).

That is really good. My Cross Spectrum calibrated ECM8000 needs 17 dB at 5 Hz, 13 dB at 10 Hz, and 4.84 dB at 20 Hz.

Rolling eyes means sarcasm. I must have missed the sarcasm!


A few general points:

1. An uncalibrated microphone is uncalibrated. Accuracy for any individual unit is an unknown. Individual units of the same model vary. FR of individual units of the same model vary. Uncalibrated means uncalibrated (AKA accuracy unknown).

2. An expensive measurement microphone is expensive for a reason. Better quality element, known, calibrated and documented FR, higher SPL capability, and probably a better phase response (etc.).

3. The vast majority of weekend HT warriors need not be concerned with measured FR below 20 Hz. You get what you get below 20 Hz.

4. I can close mic measure my sealed main speakers 20 times in a row, and the FR below 20 Hz will vary measurement to measurement. Since each one is different, then which one is the accurate measurement below 20 Hz? I do not for one nanosecond think that the use of an expensive measurement microphone will change that fact.

5. I use a RS SPL meter for a measurement microphone because I just look for general relative trends when I measure in room. Kind of hard to get an accurate speaker measurement at the listening position in a room anyhow. Move over a couple of feet and the measurement is different. Room reverberation is always measured along with everything else in the bass department. You can't gate it out!


Below here is an old steady state FR measurement of my system (R, L, SW together) measured at the LP. Calibration factors have not been applied. The SPL C scale compensation factors have not been applied. There is no easy way to apply any compensation factors even if I had them. Accuracy at the low end is an unknow.

At the high end, anything above 8 kHz has unknown accuracy. In reality it just does not matter how that area measures. My speakers tweeters are crossed at 3kHz, so I get what I get. I am not going to EQ the tweeters anyhow!

That being said, I do have a good idea on what is going on in my room. Note that at 10 Hz, the RS SPL meter filters C scale by around 15 dB. How much more drop off is caused by my specific RS SPL meter's microphone is an unknown.

Well, that is not what I see when I measure. All my sweeps measure different in the low end. RS Digital SPL meter is used as the microphone, with the SPL C scale adjustment factors activated (MIC CAL button is ON).

These sweep measurements were made with TrueRTA. Under normal conditions, I would delete the obvious "bad looking" sweeps, and then average them. REW sweeps are a bit different, and REW goes off track at 8Hz when I measure with REW.


Top set of curves are 15 consecutive TrueRTA QuickSweeps overlaid no smoothing. 110 dB SPL sweeps measured at the close mic SPL meter position.

Second curve from top is the average of all 15 sweeps no smoothing applied.

Third set of curves is the soundcard calibration on and off. Minimal soundcard compensation is required.

Fourth curve down from top is the noise floor of room measured with microphone on and no noise except for the PC fan. No smoothing is applied.

Bottom is noise floor of electronics and cables with SPL meter turned off but still connected. No smoothing is applied.


One note, if you clip the RS SPL meter you can smooth out the curves (AKA SPL meter reads over range). Same thing applies to all electronics in the chain!