My guess given that the that peak is broad and very similar in both channels is that that's a reflection close to the mic, like from a chair. I've always gotten the best results by moving my listening chair out of the room when taking measurements.
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
midrange peak, a problem in room acoustics
- Thread starter Zathrez
- Start date
Keith_W
Master Contributor
All right, let's take a look.
We can rule out "speaker problem" first. This 1m measurement from your speaker shows that the midrange peak at 800Hz isn't there.
We can also rule out "toe in problem". The measurements do not change no matter where your speaker is aimed. But if we look at the manufacturer measurements, we see:
... that your speaker is radiating a lot of energy out the back at 800Hz. Keep this in mind for later. It also appears to be a very wide radiating speaker.
In the first set of measurements, all the measurements differ by a lot, but the 800Hz peak is present in all of them. The descriptions say that the physical location of the speaker was changed.
This is a comparison between Curve 1 and 5 in your first set of measurements. The position of the speaker relative to the right wall is unchanged at 0.92m. The only difference is that the green curve is 1.12m from the front wall, vs. the starting position of 0.92m.
What we see here: SBIR cancellations worsening at 60Hz and 170Hz, but the peak at 800Hz is virtually unchanged. Conclusion: my earlier concern that the speaker is radiating rearwards in the polar plot can be dismissed. It's not some kind of SBIR reinforcement.
So what if the speaker is moved closer to the right wall? This is a comparison of Curves 1 and 2. In the green measurement, the speaker was moved 20cm closer to the right wall. We can see the peak is still there, but it has improved a little.
To summarize what we have found, we now know that:
1. It's not a speaker problem.
2. It's not a toe-in problem.
3. Speaker positioning has minimal effect on that peak.
The remaining possibilities are:
4. Poor choice of listening position
5. Something local in the listening position causing a strong reflection that boosts 800Hz.
And indeed, if we look at the ETC, we can see:
There are 3 clusters of reflections at 1.6ms, 7ms, and 18ms within the Haas fusion zone (where the tonality of reflections is integrated with the direct sound by your brain). These clusters correspond to a path length difference between direct sound and reflection of 0.55m, 2.4m, and 6.2m respectively. That early reflection at 1.6ms is particularly strong, and I suspect this is the culprit.
What you need to do:
1. Make sure the 1.6ms reflection is real and not a measurement artefact. Examine your room for sources of nearby reflections. Please show us a photo of how you set up the microphone for this measurement, clearly showing the back wall and nearby furniture.
2. Starting at your current listening position, take measurements at 30cm increments moving closer and further from your speaker. Overlay the measurements and see what happens to that 800Hz peak.
Your room has three axes - front-to-back, side-to-side, and vertical. We have excluded the side-to-side axis through your measurements. The vertical axis is fixed. The only axis that remains is the front-to-back axis, which you have not measured yet.
We can rule out "speaker problem" first. This 1m measurement from your speaker shows that the midrange peak at 800Hz isn't there.
We can also rule out "toe in problem". The measurements do not change no matter where your speaker is aimed. But if we look at the manufacturer measurements, we see:
... that your speaker is radiating a lot of energy out the back at 800Hz. Keep this in mind for later. It also appears to be a very wide radiating speaker.
In the first set of measurements, all the measurements differ by a lot, but the 800Hz peak is present in all of them. The descriptions say that the physical location of the speaker was changed.
This is a comparison between Curve 1 and 5 in your first set of measurements. The position of the speaker relative to the right wall is unchanged at 0.92m. The only difference is that the green curve is 1.12m from the front wall, vs. the starting position of 0.92m.
What we see here: SBIR cancellations worsening at 60Hz and 170Hz, but the peak at 800Hz is virtually unchanged. Conclusion: my earlier concern that the speaker is radiating rearwards in the polar plot can be dismissed. It's not some kind of SBIR reinforcement.
So what if the speaker is moved closer to the right wall? This is a comparison of Curves 1 and 2. In the green measurement, the speaker was moved 20cm closer to the right wall. We can see the peak is still there, but it has improved a little.
To summarize what we have found, we now know that:
1. It's not a speaker problem.
2. It's not a toe-in problem.
3. Speaker positioning has minimal effect on that peak.
The remaining possibilities are:
4. Poor choice of listening position
5. Something local in the listening position causing a strong reflection that boosts 800Hz.
And indeed, if we look at the ETC, we can see:
There are 3 clusters of reflections at 1.6ms, 7ms, and 18ms within the Haas fusion zone (where the tonality of reflections is integrated with the direct sound by your brain). These clusters correspond to a path length difference between direct sound and reflection of 0.55m, 2.4m, and 6.2m respectively. That early reflection at 1.6ms is particularly strong, and I suspect this is the culprit.
What you need to do:
1. Make sure the 1.6ms reflection is real and not a measurement artefact. Examine your room for sources of nearby reflections. Please show us a photo of how you set up the microphone for this measurement, clearly showing the back wall and nearby furniture.
2. Starting at your current listening position, take measurements at 30cm increments moving closer and further from your speaker. Overlay the measurements and see what happens to that 800Hz peak.
Your room has three axes - front-to-back, side-to-side, and vertical. We have excluded the side-to-side axis through your measurements. The vertical axis is fixed. The only axis that remains is the front-to-back axis, which you have not measured yet.
Last edited:
You've gone to a lot of trouble, and expense, chasing this down - but I don't see any mention of Room Correction (REW).
Looking at your measurements, I think Room Correction can flatten this out, as it has done in my system.
I've attached the method I use. It's easy to do and gives great results.
Worth a try, IMHO.
Keith_W
Master Contributor
I am sorry, I disagree with that advice. He needs to diagnose the issue first before attempting room correction at 800Hz. The wavelength of 800Hz is 43cm, you don't need to move much to change that pattern. The first step is to find a better position in the front-to-back axis, and then refine that by taking an MMM from that position. If the peak is still present, then removing it via EQ is a possible solution. Don't go around chopping high frequency peaks without thinking about it first.
- Thread Starter
- #27
Nice analysis, thanks for that.
I didn't measure the MLP moving forward and backward. True enough. I thought about it, but at the time I wasn't sure about the point of measuring behind that loveseat. That is, in front of the back, or behind the back. I'm sure the reflections in those two places must be quite different. So sorta like comparing apples and oranges? I should have done it just to see what it looked like.
Second way would be to move the seat with the microphone (forward and back) just to keep the loveseat more or less constant in the measurements. Would that help or hurt, or be in any way meaningful? IDK.
Third, I could just take the loveseat out of the room entirely. That might be interesting too.
Which one(s) would you like to see? Are then other alternatives that would tell you something that I haven't suggested yet?
-------------------------
As an aside, what frequencies does your ETC graph cover? I've been trying to get REW to show me something like that (over a small range of frequencies, like this peak) and can't figure out how to get it to this point -- where I can see the valid reflection timing.
Finally, that 1.6ms reflection is... interesting. That's about 0.5m more than the direct sound, yes? I'll string test that one first. I can't think of what it would be, but it's something.
Keith_W
Master Contributor
Show us a picture of the microphone, your listening chair, and your back wall please. A picture of the front showing the loudspeakers would be good, too. Don't worry if it's messy, we've seen it all before. It will help us understand the acoustics of your room a bit better.
Why it matters: if you are measuring using a mini tripod perched on your seat, for sure that will affect the measurement. The microphone needs to be placed where your ears are going to be when you sit down. So you will need a boom tripod to obtain proper height. If you are unwilling to invest in that, an alternative is to duct tape your mic to a broom handle and do an MMM around the area where your head would be.
Why it matters: if you are measuring using a mini tripod perched on your seat, for sure that will affect the measurement. The microphone needs to be placed where your ears are going to be when you sit down. So you will need a boom tripod to obtain proper height. If you are unwilling to invest in that, an alternative is to duct tape your mic to a broom handle and do an MMM around the area where your head would be.
