Help needed understanding graph

[MikeOates]

Hi,

I am a real newbie at this so please bear with me.

What I need is some help understanding the graph I am posting here. Just what does this show to you experts. Is this a poor result?

I should explain what I am doing...

I want to see how the monitors I have on my PC are performing in a rather small room 2x3m. I am not going to change any EQ or use filters in anyway. All I want to do is alter the room acoustics by adding absorbers / reflectors etc to make any changes. The position of the Monitors are not symmetrical in the room (not easy do do).

I am using the PC for mixing original recordings, not what most of you use the software for to tune your systems and room to improve listening of existing CD’s movies etc.



On the graph Left is Red, and green is right. I turned off the right speaker /monitor when doing the measurements for the left etc. Is this correct? The right monitor is nearer to a wall than the left monitor. When I first did this I assumed the software would only use one speaker at a time, but it does both, have I set it up wrong?

I used the Radio Shack SPL Meter with the cal file I downloaded. The monitors are Alesis M1 Active Mk2 with that have a flat output between 50Hz and 20KHz with cutoffs at 40Hz and 23.5KHz and is driven by an M-Audio Fast Track Pro USB box. (I don’t use a PC sound card quality is not good enough)

Any help appreciated.

Mike

[brucek]

I would increase the level of the test from the ~60dB you're at to something like 80dB. (edit)

I would change the vertical limits of the graph from the (10dB-96dB) to something like (45dB-105dB).

I would turn on smoothing of about 1/3 octave to remove the spikes and make it a bit easier to read.

All in all the response is about as expected. The lower end is fairly flat down to about 45Hz and begins to drop off. The upper end cannot be verified above 10KHz, because the RS meter basically sucks above that range. But your response is quite flat.

brucek

[MikeOates]

Brucek thanks for your advise.

Here is another graph with the changes you suggested, except that the level is at 70dB any higher and it clipped.



So basically you say this is ok, but it does have a couple of dips in the ranges 240~550Hz and 1KHz~3.5KHz. Is the best way to flatten this out by removing the peaks in other areas using absorbers tuned to certain frequencies thereby reducing the peak levels to fall in line with the lower parts of the curve. Or is it the other way round and try to increase the dips with reflectors? The latter sounds more difficult, but as a novice I would be grateful of any advise.

Or would you be happy with the graph as it is?

Would I be correct in saying that the right speaker (green) is higher at the bottom end because it is closer to a wall, in fact it is almost in a corner of the room. And to improve the bottom end I would either have to box in the left speaker or add a bass trap behind the right speaker? (By improve, I mean even out the left and right so they are the same levels.)

Thank you,

Mike

[brucek]

Yeah, the small lower end boost you're getting on the green speaker is likely due to corner placement. As is the case with a sub, corner placement accoustically couples the speaker better to the room and excites all resonances to give a boost in gain.

I'm afraid my experience is limited to low frequency equalization and not to room treamant. Hopefully others can chime in and help out with that.

Overall though I would say you need to lower somewhat the peak between ~120-150Hz and try and lower the upper end which must be a bit bright? A slow declining 'house curve' I've shown usually sounds best.





brucek

[Sonnie]

We have some very knowledgeable room treatment guys hangin' around and I would think those areas that brucek point out would definitely benefit from room treatments. I would probably suggest this be your first avenue to tame those areas too, if you can afford it or you could use DIY acoustical panels.

You might list your equipment in your profile too... so we know what you have to work with.

[JohnM]

The response shape above 2k is primarily due to the meter, if you look at the graphs in the leaflet that comes with with the meter you'll see the same shape. To do any EQ above a few hundred Hz you really need a calibrated microphone, else you are likely to do more harm than good.

[MikeOates]

Great help so far, many thanks, getting there slowly

I have got some Acoustic tiles 40mm thick see ebay item: 7381039576. Works well for high frequency absorption, can also be fitted multiple thickness to lower the frequency. I can also do some DIY. But can't find a supplier of the rigid Rockwell wool (Manchester, UK) only the softer type used for loft insulation.

I will compile an equipment list and post in the next few days.

Re John's message about a calibrated microphone, any suggestions? I am prepared to go down this route to get it done right. Would prefer a not too expensive one though. But can a good microphone with a published profile do instead (by this I mean a small response graph that comes with the mics and on the website, no figures just a graph). I have a number mainly by Rode:
NT1a
NT2000
NT5
NTG-2
An example of a Frequency Response graph is http://www.rodemic.com/downloads/NT5_Brochure.pdf

Thanks,

Mike

[Sonnie]

How much can you get the NT5 for? The Behringer ECM8000 will run you about 50 bucks and its farily flat from 20hz - 10khz. You can have any of them professionally calibrated from 10hz - 46khz.

[JohnM]

If you have a good mic may as well use it, but ideally you'd want an omni otherwise the mic's polar plot affects the measured levels of off-axis sound and you end up a response that is not quite as your ears perceive it. This calibrated mic/preamp set is good value: http://www.ibf-acoustic.com/2005/shop-us/pd1041155380.htm?categoryId=8, but chances are you have a preamp you could use and would only need the mic: http://www.ibf-acoustic.com/2005/shop-us/pd1954442790.htm?categoryId=2

[MikeOates]

Fantastic, I am going to get the Behringer ECM8000 it's only £38.78 inc VAT from
Boomerangsounds

Thanks for the help, I will get the mic, do some more measurements and perhaps post another graph and take it from there. May take a couple of days.

Incidentally the Rode NT5 is not an omni and costs about 3 times the price of the ECM8000. The Rode NT2000 has variable polar pattern including omni, which I will also try alongside the ECM8000. It would be interesting to see how they compare.

Your a great bunch of guy's

Mike