• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. 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!

UMIK-2 First time Room EQ measurements.

  • Thread starter Deleted member 35357
  • Start date
OP
D

Deleted member 35357

Guest
I noticed I could add EQ filters on top of each other to make additive corrections if you don't mind having a crazy long list of adjustments.

Are there any drawbacks on just doing correction on-top of corrections and summing the results for a "clean" smooth end result not minding what happens in-between?
Seems there were no real performance disadvantages on the PC but to many corrections might not work in other devices if you want to export them elsewhere I guess.
 

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,840
Likes
5,775
Yeah, it's the left channel without the sub.
Then the corrections I see seems WAY too heavy for both amp and speaker,you're in the realm of destroying something trying to make it operate way above it's capabilities.
 
OP
D

Deleted member 35357

Guest
Then the corrections I see seems WAY too heavy for both amp and speaker,you're in the realm of destroying something trying to make it operate way above it's capabilities.
How could it be destroying them when I'm subtracting power for most part?

I have more than -12dB across the spectrum right now before they go to the speaker inputs with the corrections and then I'm not even maxing the DAC output. SUB gain is not even 1/5th of it's power. I Might even be only using 1/6th of the gain at the moment.

These speakers can be loud if you put a full signal into them, uncomfortably so. I had the pleaser to do so when I had them connected to PC outputs years ago and didn't check the volume level wasn't at minimum output.
 

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,840
Likes
5,775
How could it be destroying them when I'm subtracting power for most part?

I have more than -12dB across the spectrum right now before they go to the speaker inputs with the corrections and then I'm not even maxing the DAC output. SUB gain is not even 1/5th of it's power. I Might even be only using 1/6th of the gain at the moment.

These speakers can be loud if you put a full signal into them, uncomfortably so. I had the pleaser to do so when I had them connected to PC outputs years ago and didn't check the volume level wasn't at minimum output.
Only the +10db at 57Hz or so needs X10 the normal power.Now add the rest positive values and you get close or beyond your subtraction and that without knowing the specifics of the phase and impedance of these freq and how hard they can be for the amp.
It's great to test and learn but keep in mind that there are limits.
 
OP
D

Deleted member 35357

Guest
Does the minimum noise floor of the Room have a large effect on the measurements?

Did a measurement for the Room to see what the constant noises were.
3 MMM measure average and the lowest measured.

Noise floor.png



The PC is causing some issues and there is this constant 25Khz peak in everything but also a high 25Hz & 50hz constants that wreck some havoc on the AUTO EQ if I don't limit it trying to adjust the frequency at that level. (I guess I can remove some PEQ filter there as it isn't from the speakers output)

60dB 25Hz noise is constant. Those peaks in 100-600hz range are "water trickles" from my PC I guess as I have it Water-cooled and there have been this constant issue with air bubbles I can't seem to get out from it. I think they are loud enough to effect some measurements. (They are in the range where I get issues with peaks & valleys.)

I was surprised by the high 25Hz & 50Hz measurements of the background. Are they internal noise from the electronics or are they actual background noise in the room?

There is this Air-duct system just above this room that goes on and off at times and it has a distinct low humm that can be heard when everything is somewhat silent.

EDIT: Discard the thing about the water droplet squelch, turned down the pump and it didn't have really much if any effect even without the water trickle sounds though I moved the average downward.

Just showed that the 25, 50, 100hz are always there, so is the 200hz peak.
300hz might have been some of that water but it's hardly different. It was mostly above 1K Hz that went lower without the water squelching happening.
So power supply & internal electronics noise I reckon. (they reach 65 SPL peaks???)
 
Last edited by a moderator:
OP
D

Deleted member 35357

Guest
I level matched the 2 channels using the 7-Band Analogue Equalizer and then did a quick REW EQ "Room correction" of it.

Channel matched results.png

Did a MMM measurement of the results.
Kinda flat even though the PEQ was supposed to have a 0.5dB/Octave slope from 200hz.

Code:
Filter Settings file

Room EQ V5.20.13
Dated: 2022-dec-19 18:51:50

Notes:7-band EQ channel matched before REW EQ L+R single point measurement calculation, 50hz cutoff.

Equaliser: Generic
dec 19 L+R 3
Preamp: -8.5 dB
Filter  1: ON  PK       Fc   60.80 Hz  Gain  -5.70 dB  Q  5.489
Filter  2: ON  PK       Fc   98.30 Hz  Gain  -6.00 dB  Q  2.000
Filter  3: ON  PK       Fc   178.5 Hz  Gain  -8.10 dB  Q 12.940
Filter  4: ON  PK       Fc   192.5 Hz  Gain  12.00 dB  Q  3.162
Filter  5: ON  PK       Fc   226.0 Hz  Gain -19.90 dB  Q  4.999
Filter  6: ON  PK       Fc   250.0 Hz  Gain   8.40 dB  Q  7.500
Filter  7: ON  PK       Fc   296.0 Hz  Gain  -7.60 dB  Q  4.997
Filter  8: ON  PK       Fc   319.0 Hz  Gain  12.00 dB  Q  1.384
Filter  9: ON  PK       Fc   372.0 Hz  Gain  -8.30 dB  Q  4.999
Filter 10: ON  PK       Fc   422.0 Hz  Gain  -5.60 dB  Q  4.999
Filter 11: ON  PK       Fc   467.0 Hz  Gain  -3.90 dB  Q  4.995
Filter 12: ON  PK       Fc   799.0 Hz  Gain -14.10 dB  Q  3.840
Filter 13: ON  PK       Fc   813.0 Hz  Gain  11.20 dB  Q  7.489
Filter 14: ON  PK       Fc   899.0 Hz  Gain   7.70 dB  Q  4.927
Filter 15: ON  PK       Fc    1408 Hz  Gain  -1.60 dB  Q  1.003
Filter 16: ON  PK       Fc    2773 Hz  Gain   2.80 dB  Q  2.712
Filter 17: ON  PK       Fc    3310 Hz  Gain   7.30 dB  Q  4.994
Filter 18: ON  PK       Fc    5857 Hz  Gain   2.40 dB  Q  7.500
Filter 19: ON  PK       Fc    7214 Hz  Gain   6.40 dB  Q  5.822
Filter 20: ON  PK       Fc   11313 Hz  Gain  -2.50 dB  Q  5.000
Filter 21: ON  None   
Filter 25: ON  None

Will take my time to check the results of this on how it sounds.
 

Attachments

  • Channel matched results PEQ prediction.png
    Channel matched results PEQ prediction.png
    319.9 KB · Views: 60

ThatSoundsGood

Active Member
Forum Donor
Joined
Aug 12, 2022
Messages
147
Likes
114
that's not how PEQ works. Filters influence each other. For example, here is a 20dB boost right next to a 20dB cut:
View attachment 250947
as you can see, the two filters almost cancel each other out.

And here's the actual filter response of those crazy settings:
View attachment 250948
That's interesting. That means it's using a more advanced algorithm to get to the target curve rather than just implementing those filters that it lists? That would make more sense after looking at those filters.
 
OP
D

Deleted member 35357

Guest
So you correct single point and remeasure MMM @Nighthog ?
Was a little lazy when I did the level matching as it was quicker to do single points to see the effects.
Used one of the results of those to do the EQ in REW, Should have probably done MMM first not after the EQ.

The MMM measurements always lift the high end I've seen when compared to the single point measurements.
 

staticV3

Master Contributor
Joined
Aug 29, 2019
Messages
7,535
Likes
12,002
The MMM measurements always lift the high end I've seen when compared to the single point measurements.
Could be 90deg cal file loaded, but mic pointed at the speaker while doing MMM

Edit: try putting the mic on a stand, then do a regular measurement sweep and an MMM right after, with the mic on the stand.
Are the results still different?
 
Last edited:
OP
D

Deleted member 35357

Guest
Could be 90deg cal file loaded, but mic pointed at the speaker while doing MMM

Edit: try putting the mic on a stand, then do a regular measurement sweep and an MMM right after, with the mic on the stand.
Are the results still different?
I think it might be more that when you move the Mic around you might get closer to the speakers than when having it stand still at one point.

EDIT:
Yeah, not moving the microphone around gives pretty much the same measurement to a single point one.
 
Last edited by a moderator:

staticV3

Master Contributor
Joined
Aug 29, 2019
Messages
7,535
Likes
12,002
Yeah, not moving the microphone around gives pretty much the same measurement to a single point one.
Possible that you're moving the mic too fast, causing some handling noise.
IIRC the PDF recommends <30cm/s.
 
OP
D

Deleted member 35357

Guest
Possible that you're moving the mic too fast, causing some handling noise.
IIRC the PDF recommends <30cm/s.
I'm aware of this and made sure not to rush with it.

I'm in near-field listening position so you don't need to move around much to notice different sound levels from either side of the channels.

I was aware there was a difference on levels for a long time but before I got the UMIK-2 I didn't know by how much I should have corrected.
I had to correct the LF on the one side and he HF on the other side, they were 2-4dB mismatched across the frequency range crossing around 1Khz were they had the same output.
 
OP
D

Deleted member 35357

Guest
Would it be wise to buy a miniDSP 2x4 Flex to go with this setup? I could get a better SUB integration with it even if I use other speakers later as the functionality would then be transferable to whatever you get in the future, even if getting Genelec or Neumann or other stuff.

Would then not need to invest in GLM if I want that functionality as you already have it in the miniDSP 2x4 as external functionality that works with anything.
I would reckon the functionality is superior in miniDSP than the GLM propitiatory solution.
 

brk

Member
Joined
Mar 15, 2020
Messages
28
Likes
20
Some comments based on spending too many hours on this for my speakers/room, which had a somewhat similar frequency response to yours.
1. I could never get consistent measurements using the MMM method. Averaging fixed measurements around the listening position was the only way I could get repeatable results. The trick is to determine - as best as possible - the exact position of your ears in your "normal listening position" and then move the mic forward/back, up/down, side-to-side from there. Having your body in the measurement - the MMM method - changes the response. You can test this by taking a fixed mic measurement with and without yourself sitting next to the mic. Although it is unclear which is "correct", the house curve for a "perfect speaker in a normal listening room" was obtained by fixed mic.
2. If you can get anechoic measurements of your speakers - say from a review - correct these first and then measure and listen in your usual spot. Use this response to determine where to "draw the line" for any further EQ- place the curve right through the middle of these. Judge the sound quality of any further EQ in comparison to the anechoic corrections - a tasteful approach is needed for further improvement. Many changes look better but clearly sound worse.
3. The "old" advice of boosting wide and cutting narrow turned out to be true for me. Any boost with a Q value >3.0 over 200 Hz sounded bad.
4. Not sure you can boost the 100-300 region as much as you are doing without making it sound weird. May have to leave it lower as this
Good luck,
 
OP
D

Deleted member 35357

Guest
Some comments based on spending too many hours on this for my speakers/room, which had a somewhat similar frequency response to yours.
1. I could never get consistent measurements using the MMM method. Averaging fixed measurements around the listening position was the only way I could get repeatable results. The trick is to determine - as best as possible - the exact position of your ears in your "normal listening position" and then move the mic forward/back, up/down, side-to-side from there. Having your body in the measurement - the MMM method - changes the response. You can test this by taking a fixed mic measurement with and without yourself sitting next to the mic. Although it is unclear which is "correct", the house curve for a "perfect speaker in a normal listening room" was obtained by fixed mic.
2. If you can get anechoic measurements of your speakers - say from a review - correct these first and then measure and listen in your usual spot. Use this response to determine where to "draw the line" for any further EQ- place the curve right through the middle of these. Judge the sound quality of any further EQ in comparison to the anechoic corrections - a tasteful approach is needed for further improvement. Many changes look better but clearly sound worse.
3. The "old" advice of boosting wide and cutting narrow turned out to be true for me. Any boost with a Q value >3.0 over 200 Hz sounded bad.
4. Not sure you can boost the 100-300 region as much as you are doing without making it sound weird. May have to leave it lower as this
Good luck,
I noticed that the EQ in the 300-1000Hz range usually messed up the sound in the wrong manner. While above it was more a taste matter if you liked it or not.
Doing EQ 60-300Hz had some benefits after doing some listening tests and redoing measurements and trying EQ another pass on the EQ. Looking at the differences and enabling/disabling the individual PEQ filters to see if I can hear the difference which most of the time I could.

I noted anything messing with the 300-1500Hz range usually wasn't good, I mostly felt the sound got really narrow and smaller versus the wide-field stereo it was without any EQ settings there. Felt artificial with the PEQ used to correct the frequency response there.

The peaks and valleys in the 150-300hz range were "better" to have these large corrections used there rather than leaving it undone. Though they required 8-12dB boosts & swings down to correct with the AUTO settings, I felt they were a little heavy so I lowered the individual swings a little as not to go too high. Around a 8dB swing was enough I felt versus the 12dB swings REW produced with the AUTO tune.

I noticed I didn't like the bass being messed with @ 60hz and below. Felt so anaemic there if there wasn't enough oomph to the bass with the flatness the house curves I tried resulted in. Though it was "tighter" it was not loud enough to be felt like with it having nothing done. Doing correction above 60hz usually sounded better.
So I just unticked the 60Hz filter and left the others in place. Though it might be a bit boomy without any adjustment so some smaller correction might be tried at next listening test.

MMM, is a more of a hassle to measure, but I've found the EQ settings I get from those measurements are generally better sounding than from a individual measurement.
You get a better average EQ for you listening position.
Though I think it might be better advised to get a smaller area with it than a wider area unless you have a home theatre setup for multiple people.
For a Desktop listening position moving the mic around just in your own seating area was good enough.
 
Top Bottom