Room mode at 700hz and 850 hz

[jodaki]

Hi,

I seem to be controlling room modes under 200 Hz reasonably well (pretty much within 3db down to 34 Hz - my speakers' bass floor).

However I have a monstrous double peak at 700 and 850Hz..

And I mean MONSTROUS! 14dB!!@

Re-arranging things around the room doesnt seem to help and I would prefer to fix it that way if possible (as opposed to changing my mixing position). I have plenty of insulation available but I dont want to kill the room's reverb, which is quite nice at the moment and has a reasonable RT60 pattern.

Im wondering if this sort of peak is characteristic of anything? Perhaps someone has heard of this before or can suggest a possible way to deal with it?

Like I said, the response is pretty well-behaved up til that point. I have some issues higher up but I want to solve the issues in ascending (frequency) order.

Thanks for any help, John

[Hugh Robjohns]

A peak suggests either a very strong direct reflection, or something is resonating. And the frequency suggests it's either a path length or an object with a dimension of 20 inches, or a simple fraction thereof.

Does the peak change in frequency if you move the measurement mic? If it does, then it's a direct reflection issue -- possibly from the desk top, for example.

If it doesn't change then look for a resonant object or cavity, like a speaker stand, or enclosed space under the desk or a rack etc...

The other possibility, I suppose, is a speaker reflection from the front wall, so try temporarily moving the speakers closer or further to see if the peak moves in frequency.

[MOF]

They don’t appear to be related to one another, usually these peaks have related peaks above/below in octaves (double/half frequencies).
I would move the speakers forwards or backwards and re-measure. If the peak frequencies change it’s the room, if they don’t it’s some sympathetic resonance from another item in the room e.g. an acoustic guitar, metal rack mount cabinet panel etc.

I’ve just seen Hugh’s similar post but I’ll put mine up anyway.

[The Korff]

My money's on rear-wall reflections (the two freqs you mention have wavelengths of 40 and 50 cm).

[jodaki]

Thanks for the ideas. It hadnt occurred to me that specific frequency spikes/dips would be associated with objects of specific sizes..

Clearly Im a beginner at this so could I ask you to clarify another, hopefully not too stupid, question that occurred to me? Next to each measurement in the list there is a dB measurement. To see what is a peak or a dip do I move the dB axis on the graph to that level? If not, where do I move the dB axis to in order to measure peaks and dips?

[Wonks]

What software are you using to measure the room response and what mic are you using?

Are you putting the mic at your listening position at head height when seated?

Some photos of the space might help.

You haven't got anything on your DAW output such as Sonarworks that is compensating for a problem that no longer exists?

[MOF]

Next to each measurement in the list there is a dB measurement. To see what is a peak or a dip do I move the dB axis on the graph to that level? If not, where do I move the dB axis to in order to measure peaks and dips?

I don’t understand your question, what is this list you talk about? Measurements are usually shown graphically where you will see boosts and cuts at certain frequencies.
I don’t see how you can move an axis, it’s a fixed part of the graph. In this case there are two, one for frequency response in Hz and one for amplitude (volume) in dBs.

[Martin Walker]

Next to each measurement in the list there is a dB measurement. To see what is a peak or a dip do I move the dB axis on the graph to that level? If not, where do I move the dB axis to in order to measure peaks and dips?

I think in this case it's worth re-quoting that popular phrase 'a picture is worth a thousand words' - give us a screenshot and we can more easily tell you more

Martin

[jodaki]

Yes, sorry I’m using REW.

Beneath the graph window that shows the current batch of measurements there is a ‘legend’ window listing each measurement, its date and then a decibel figure. I have been assuming that the decibel figure is the benchmark that I drag the dB axis to (and thus the reference level for that particular measurement above which the measurement graph is showing peaks and below which it shows dips).

I will send some pictures a bit later of the screen I’m talking about in REW and also of the room.
Thanks, John

[CS70]

Do you have large computer screens between the monitors?

[jodaki]

I don’t see how you can move an axis, it’s a fixed part of the graph.

I use a mac and on the mac version of REW if you touch the measurements graph window then the X and Y axes reset their origin to that point. You can also drag the X and Y axes around the graph to help pinpoint frequencies accurately and, well so I thought, set the origin of the dB axis for assessing what is a peak and what is a dip...

[jodaki]

Do you have large computer screens between the monitors?

Yes, but unfortunately it wasn’t that. Or the table it sat on. I am putting the furniture all back in today so I will take a photo...

[jodaki]

Here is a pictures of the REW graph I was talking about.


The measurement is in a pretty much bare room and not at my mixing position but you can see the legend gives a dB value of 117.3dB

Is this is the level I would move the horizontal bar to on the dB axis.

I have noticed that this level varies depending on the smoothing applied. In this screenshot I used 1/6 smoothing, as recommended in a post In read.

Here is also a screenshot of the SPL & Phase tab.



This shows the soundcard cal as the dotted line. I have noticed just now that the soundcard cal seems to be a reasonably close inverse of the measurement. I dont know if that is significant..

[MOF]

The graph shows just one peak at let’s say 825 Hz not two separate peaks as per your post’s title.
Did you really have such a loud speaker playback for it to be at 120dB@ at the listening position?
There’s a good chance that driving the speakers that much has caused resonances in the cabinets that are exaggerating the peaks, if you have ported
speakers it would be worthwhile putting the sponge bungs in them, this will give you a more accurate bass response and reduce the amount of hyped bass in the lower octaves.
You could clearly do with some bass absorbers to tame the low end and that might help with harmonics higher up.
I don’t think it’s an axis you’re moving I think it’s a cursor that you are moving along the curve which displays numerical values at the bottom.

[jodaki]

The graph shows just one peak at let’s say 825 Hz not two separate peaks as per your post’s title.

I smoothed it to 1/6 which is probably why it only looks like 1 peak. The unsmoothed graph shows 2 peaks at the freqs I mentioned.

Did you really have such a loud speaker playback for it to be at 120dB at the listening position?

Ha, no of course I didnt! The audio was playing at around 72db approx 3 ft in front of the speakers. I have no idea why the REW graph says 120dB. Id be half dead if had been that loud and the test would not have run due to clipping. As you can see from the delay figure (on the left in the notes) the mic was placed 172" from the speakers. I expect the level at the mic was around 65 dB. The internal calibration aligned all three levels (Out, In and RefIn) to -14db.

As I tried to explain before - I wasnt trying to demonstrate the performance of my room, I wouldnt take a reading like this normally as Id usually be around 3ft from my speakers for example, and be in my monitoring position and have all my furniture and equipment back in the room. I stripped the room bare(ish) and placed the mic in a place miles from where I would monitor, and in a place that I knew had large bass peaks, simply to demonstrate the graph I was talking about and also to demonstrate that the large peak around 800Hz is still there. In the hopes that all this would help someone help me understand what is going on..

You could clearly do with some bass absorbers to tame the low end

I have extremely substantial bass absorbers in the corners of the room each weighing over 100KG and at various densities. Im sure I could do with more but for now bass is not the problem. Yes clearly the measurement I sent shows wild bass but as I explained, I do not monitor there. And I know that position has bass resonances. Where I monitor, the bass seems to be quite well controlled.

I don’t think it’s an axis you’re moving I think it’s a cursor that you are moving along the curve which displays numerical values at the bottom.

Ah I see, thanks. Can you tell me then where is the baseline for measuring peaks and dips? Surely there is a point somewhere where dB values are + or - the base level..

thanks again, John

[MOF]

Ah I see, thanks. Can you tell me then where is the baseline for measuring peaks and dips? Surely there is a point somewhere where dB values are + or - the base level

John I think you’re confusing this graph with an audio waveform.
There are no negative values, they are all relative to absolute silence of 0dB (at the bottom line, Y axis).

[MOF]

As I tried to explain before - I wasnt trying to demonstrate the performance of my room, I wouldnt take a reading like this normally as Id usually be around 3ft from my speakers for example, and be in my monitoring position and have all my furniture and equipment back in the room. I stripped the room bare(ish) and placed the mic in a place miles from where I would monitor, and in a place that I knew had large bass peaks, simply to demonstrate the graph I was talking about and also to demonstrate that the large peak around 800Hz is still there. In the hopes that all this would help someone help me understand what is going on..

MOF wrote:
You could clearly do with some bass absorbers to tame the low end

Taking measurements from where you mix is really all you should be doing, unless it’s a commercial facility and even then you’re unlikely to avoid bass tilt at the Producer’s sofa at the back of the room.
If you get similar results in the mixing position then I would be concerned that your bass absorption isn’t working since it’s increasing in value at the point where it’s going lower, not gently rolling off.
I presume these measurements were taken close to a wall, where bass frequencies always increase.
Please do a real world test i.e. the mixing position and post the results. I think you’ll find much better results.

[MOF]

As I tried to explain before - I wasnt trying to demonstrate the performance of my room, I wouldnt take a reading like this normally as Id usually be around 3ft from my speakers for example, and be in my monitoring position and have all my furniture and equipment back in the room.

Your original post didn’t explain this.

[jodaki]

John I think you’re confusing this graph with an audio waveform.
There are no negative values, they are all relative to absolute silence of 0dB (at the bottom line, Y axis).

ok now I do feel a bit silly

Thanks for your other points. I think my hurry in getting my graph up earlier was to demonstrate that the peak is still there. I will definitely upload a reading when Ive moved the furniture back in.

In the meantime. MOF why is the sound card calibration ‘curve’ an inverse of my measurement wave/graph? And can you think of a reason REW would think that reading was performed at 120db?

[Wonks]

In the meantime. MOF why is the sound card calibration ‘curve’ an inverse of my measurement wave/graph?

Because (in theory) its the EQ curve you'd need to apply to the DAW output to get a flat measurement response . If you've got a 3dB peak at one frequency then it will counter it with a -3dB dip.

In reality you can't counter room mode dips with just EQ boosts.

[Hugh Robjohns]

Measurements are only as good as (a) the equipment, (b) using it correctly and (c) nderatanding and interpreting the realts accurately.

REW is great, but it relies entirely on the accuracy of the measurement mic and/or it's calibration file.

...and then using it correctly and understanding what it's telling you are far from trivial... And as a result it is incredibly easy to chase ghosts...

[jodaki]

Because (in theory) its the EQ curve you'd need to apply to the DAW output to get a flat measurement response . If you've got a 3dB peak at one frequency then it will counter it with a -3dB dip.
In reality you can't counter room mode dips with just EQ boosts.

I got the wrong end of that too then. The sound card is calibrated prior to measurements so I assumed the sound card calibration was an offset used by REW when drawing measurement graphs to compensate for the imperfections/biases of the card.

I accept what you say but if I’m being honest then if the sound card calibration is simply an inverse of the measurement then I dont really understand the point of calibrating my soundcard, surely REW can just invert the response reading to generate an eq ’recommendation’.

I dont have a device or much of a desire to fix the room modes by eq so I’d prefer to get as far as I can by altering the space characteristics - by which I mean my position in the space and the response characteristics of the space.

[Hugh Robjohns]

I fear there's some major misunderstandings here...

If the sound card is anything other than ruler flat from 20Hz to 20kHz it's broken.

The calibration is a correction for the measurement microphone, and any inaccuracies will obviously give false room measurements.

[jodaki]

I If the sound card is anything other than ruler flat from 20Hz to 20kHz it's broken.

Are you talking about REW’s soundcard cal graph? As in the SPL plus phase graph IO posted?

As you can see mine is clearly an inverse of the measurement. So it is not flat, but as far as I can tell my 18i20 works perfectly.

Are there any other possibilities?

[jodaki]

Measurements are only as good as (a) the equipment, (b) using it correctly and (c) nderatanding and interpreting the realts accurately. REW is great, but it relies entirely on the accuracy of the measurement mic and/or it's calibration file.

I read an article last week somewhere on the pointlessness of using an ECM8000 with a generic calibration file as the response of those mics varies wildly. In particular noting that the example mic in the article displayed a set of peaks culminating around 800Hz.

After which I ditched the behringer and bought a UMM-6 and Dayton Audio gave me the calibration file for it.

Since then Ive spent the last week moving speakers and mic around my room measuring the spots Im willing to work in.

This is what I had as a bare measurement (mic and speakers, no other gear) in my favourite spot.




Not knowing what a good REW SPL graph looks like I'm not in a position to judge this reading, though it looks more even than some Ive had.

Looks like some problems at 163Hz and 370Hz but more absorbers didn't seem to help.

And I don't know whether I have 2 peaks at 50Hz and 63Hz, or whether I have a cancellation at 56Hz. The un-smoothed graph is almost identical sub 100Hz to the 1/24 smoothed graph. I don't pretend to understand but I think the waterfall graph shows peaks so Im guessing these are peaks at 50 & 63Hz.



Adding more mass to the bass traps (pushing them up to nearly 200kg with extra bales, 140kg/m3 RW6 rockwool and techsound 5.8mm MLV) makes little to no difference to the sub 100Hz region. Not sure what else I can try to reduce the bass there. Perhaps the problem is leakage into the room or reflections from the ceiling or the floor.

Then I added my desk and gear back in, resulting in this measurement. This is a real world example of the response at my listening position:





My desk seems to have solved the problems at 163 and 370Hz. I don't know why or how.

But has it in turn created a new problem around 97Hz?
Is that mid-range dampening between 600 and 1.6kHz or is that an improvement?
And what do I make of what look like periodic high range resonances from 3.25kHz upwards?
Help understanding these readings would be appreciated..

And perhaps most importantly I don't know whether this is a good result. And what should I try and do from here?