Room Measurements

I now have all the software, cabling and mic equipment to take my own room measurements and thought I would present the results as I went along. Here is the initial frequency plot. As you can see relative to the midband the treble is raised and the bass is depressed - which does a lot to explain the observed brightness of the system. Next step is to make the adjustments to the crossover and remeasure. Crossover points are 380 hz and 3800 hz which matches almost exactly with the observed frequency plot.

 

Could you tell what equipment and software you have used since I would be interested in doing the same in my room

Thanks

 

Well done Nick, but are you sure thats right? It doesn't look the way I remember your system being...

Especially since it looks like the bass extends to 0Hz?

Where was the mic?

I'll do some screenshots to help guide you through it later on if you want.

 

Yes it doesnt look right and whatever I fiddle with on the crossover doesnt seem to change anything

Its the behringer ecm 8000 mic using cool edit pro on my pc

 

It's more likely to be the lack of energy in the midbass area that's causing it.

It is funny though. I was speaking with a friend today, a sound engineer of some experience. I was bemoaning the performance of the transmission line loudspeakers I heard at the show and he told me something I didn't know. That in theory, a transmission line is just about as good as it gets when it comes to bass loading, but that in practice it leads to an in room response that goes up down up down up down all the way through the bass region.

Looking at your plot, he is dead on. Most rooms will exhibit two distinct abherrations in that area. The PMC's are producing far more than that. It seems digital EQ might be a prerequisite with these kinds of speaker, some thing my friend did suggest actually. Good luck,

 

Transmission lines produce ripple from the harmonics that get excited with the fundamental quarter wave resonance you are actually trying to get. This can be lessened by placing the driver a distance from the beginning of the line and by playing with stuffing, but the ripple rarely goes below about 80Hz and the magnitude is far less than a room will usually exhibit.

What is perhaps interesting is that TL's actually tend to roll-off earlier than a standard ported design. The difference is that they roll-off much less sharply.

Anyway, it doesn't look quite right there Nick, but I'm sure it will work out. Seems like you may be recording or analyzing something other than the microphones output.

Are you free any day this week I could come and help out.

 

Alwayd wondered why I'd not heard a good one Tenson. Thanks for that.

 

I wouldnt read too much into this plot SM - as Tenson pointed out it goes to 0 hZ!

 

I just assumed the software got confused below 20hz. What do you think you measured then? It also drops off at about 17khz, and the plus/minus 12db variences below 100hz suggest room measurements of some description. Sorry I have only worked with Tact and MLSSA software in the past so this behringer stuff looks strange anyway.

 

I dont know I will try again today with the volume switched off and see what happens - maybe the software is measuring accross the sound card or something.

 

Yup even with the mic disconnected I still get a high level noise signal recorded on the track - so the track isnt pointing at the mic ... hence the problem! So the features observed by SM sadly cant be used to argue TLs sound the way they do

 

What software are you using for measurement?

When you get the inputs sorted and some more meaningful measurements posted it would be useful to know what the time window was and whether any smoothing has been applied. Incidentally when I first measured my room I was shocked at the appearance of the response, it doesnt look quite as pretty as you might expect from listening. Lots of comb filtering can produce some quite wild variations in level.

 

Cool edit pro. Time window was about a minute - I used max fft. That may be too much. I didnt apply any smoothing.

 

I would stick with max FFT, it just gives you greatest detail in the result. Smoothing can be useful to illustrate trends, or make the response look a lot more pleasing when friends look aghast at how awful the room measures

A couple of things that I found useful:

http://www.etfacoustic.com/demoroom.all.html
An article on etfacoustics website going through the process of how they optimised speaker placement in the Bryston demo room. They split the room treatment into 3 sections, below 80 Hz; 50-500Hz and 500Hz +. They outline a principle to objectively arrive at an optimimum placement for main speakers using response measurement. I found this to be an excellenty starting point before then using room correction to deal with 20 - 200 Hz. Finally you can use some absorbtion to deal with any limitations in the 500Hz + region.

http://homepage.ntlworld.com/john.mulcahy/roomeq/
A great free application specifically intended for assisting in setting up room correction. This came out of an app the guy developed when he was at Tag Mclaren. He has broadened the application out now though so it is more generic. I found it very handy for experimenting with different filter settings in real time, not having to wait for the sweep to test the effect of each filter change you make.

 

Your not going to like this...

stop screwing around with the crossover, because without robust, accurate measurements you are not going to be able to make good judgements about how to improve your loudspeakers. In fact, you run the risk of making them a hell of a lot worse, and you'll end up fooling yourself into thinking they sound better based on some erroneous graphs you've produced (see your remarks in your initial post for example).

Let me put it like this: PMC have built a few speakers, have proper test instruments, access to an anechoic space, and the time to take hundreds of measurements; you simply can't compete with them, give it up.

 

They will be robust when I am actually measuring it At the moment I dont seem to able to get the software to feed from the mic.

I bought the actives (this isnt a passive system) 2nd hand so I have to do it myself. PMC have never setup my system.

I havent made any comments about them sounding better yet as I havent actually changed them, you misread my comments.

 

From your original post I assumed you were talking about room correction, but I see that you were just talking about adjusting the crossover. Is the intention just to get the levels correct for the crossover?

 

No they won't be. At home you have to deal with environmental noise and the interference of room acoustics, so you won't be able to measure just the speaker. You simply won't be able to take high resolution measurements (in the frequency domain) because you'll have to use short gating times - which will likely preclude doing the bass/mid transition properly. Statisical averaging for noise removal requires a substantial number of measurements.

On axis measurements aren't enough, you'll need to averge from a large number of different off axis measurements.

This kind of cluelessness makes me wonder how you have the nerve to use the snide smiley face ->

So just do what PMC would do, with whatever electronics they would do it with. You clearly have more money than [engineering] sense [in speaker design..]

 

Yes thats correct - I should have mentioned that this thread follows on from another one where this was discussed. I will also do some eq in the bass to correct for room effects.

 

Measurement at the listening position is sufficent for me to set the three levels on the active crossover. I am not trying to tweak a passive with eq.

The wink was directed at myself for not getting the software to record from the mic correctly yet. I have set it to the correct feed but it isnt working. This is not a trivial problem - according to tenson cool edit needs some fiddling to get it to work so that is my next task.

PMC would be doing exactly what I am doing if I had bought the active system new.

I am familair with statistics, measurements and engineering ... with the right technique all these things are tractable. Once the software recognises my mic.