Room Correction "unstable"(known issue)

I am having the same issue when using room correction on my DMP-A8 with a UMIK-1 mic. Hoping for a resolution soon.

For my own info, which is the preferred approach: mic facing the speakers OR mic facing upwards (the mic is omni-directional)?

 

At least in the case of EvoTune Pro it appears horizontal is preferred. But, as long as you are using the appropriate correction file (there is one for horizontal and one for vertical) I imagine you can use either (or both).

 

I’ll have to try the vertical and see if it makes a difference with the ES EM-01 mic.

 

I have had the best results with the mic pointing to the ceiling with the EversSolo mic.

 

I performed an update a week ago to v1.6.15 (DMP-A6) and then the problems started when going to do Room Correction.
I always get the message: the recording signal was unstable
This involves a test using a mobile phone and its microphone.
I tried with the volume up, but it didn't help.
The mobile phone is 2 meters away from the speaker.
Do the others also have these problems and are they working on solving this irritating problem that has never existed before?

 

Yes It is reported several times by numerous people.
The guys and girl from eversolo having vacation. So we don't know if it will be addressed in a next update. But i will assume : yes

 

I think automatic room calibration messing with the sound. It messing with the room modes, this is the most important to make bass sound good.
It add small high q on the bass, and a lot of small thin q on the highs. All dynamics is gone.
I did write down all filters from both channels peq.

LEFT AUTO: RIGHT AUTO:
23.4 +4.09 Q4 | 23 +2.72 Q10
27.3 -0.3 Q0.5 | 38.4 +0.7 Q0.1
59.5 +0.3 Q0.1 | 43.9 -0.7 Q0.7
60.2 -0.5 Q0.1 | 69 +5 Q6
79.2 +5 Q10 | 76 -2.3 Q10
139 -4.5 Q10 | 131.4 -3.5 Q6
183.1 -3.5 Q10 | 177.8 -4.2 Q10
267.4 -3.5 Q10 | 246.1 -5.5 Q10
530.2 -3.9 Q10 | 372.6 -3.3 Q10
702.6 -3.5 Q2 | 604.7 -6.2 Q6
947.8 +4.9 Q10 | 802.8 -3 Q10
1103.5 -3.5 Q10 | 1008.6 +2.5 Q10
1690.9 +3.1 Q10 | 1972.5 -2.3 Q10
2583.1 -5.6 Q10 | 2558.2 -0.6 Q10
3267.7 -1.4 Q10 | 3178.6 -1.3 Q10
6216.2 -1.38 Q10 | 4162.4 +0.7 Q10
6524.4 -2.8 Q10 | 6707.4 -2.9 Q10
8559.5 -1.3 Q6 | 10156.3 -2 Q10
16941.7 -0.2 Q1.4 | 10356.5 +0.1 Q0.1
| 17357.7 -0.1 Q0.1

Her are my manuell scan with Spectrum analyser pro , used together with pink noise from ssd in A8. All frequency filters is adjusted to sound with the same level up to 750 hz . It is easy to se room modes on my frequency scan. I did use AI as a helper to set level and q. Music test between adjusting s . It was hard work to be finished :
LEFT: | RIGHT:
20 0 Q6 | 20 -3.5 Q6
41.2 -3.5 Q6 | 41.2 -8.5 Q6
77.8 +2 Q1.2 | 77.8 +2 Q1.2
105 +2 Q10 | 105 +1 Q10
131 -2.5 Q6 | 131 -2.5 Q6
155 +1.5 Q12 | 155 +0.5 Q12
510 -4.5 Q4 | 510 -6.5 Q4
622 -4.8 Q5 | 622 -4.3 Q5
750 HS -1.6 Q0.7 | 750 HS -1.6 Q0.7
1600 -0.8 Q1.5 | 1600 -0.8 Q1.5
6800 PEAK -3.0 Q2.5 | 6800 PEAK -3.0 Q2.5

 

This is a good thing.
Although you must be aware that these fixes work in YOUR ROOM, hence they are not supposed to be universal and working in every room...

 

Here is the answer:
This first auto scan is a perfect example of why you are right to be dissatisfied. This is a bad calibration.

Here is what is wrong with it:

*1. It's fighting itself with narrow Q10 filters*
Look at your LEFT:
139 -4.5 Q10
183.1 -3.5 Q10
267.4 -3.5 Q10
530.2 -3.9 Q10
2583.1 -5.6 Q10

And RIGHT:
177.8 -4.2 Q10
246.1 -5.5 Q10
604.7 -6.2 Q6
10156.3 -2 Q10

An auto system should not make 12-15 narrow -3 to -6 dB cuts with Q10. That means the mic picked up comb filtering / reflections and the algorithm thinks they are speaker errors. It will sound hollow and phase-y, not flat.

*2. Random +5 dB boosts in bass*
79.2 +5 Q10 LEFT
69 +5 Q6 RIGHT
947.8 +4.9 Q10 LEFT

You never boost a room null with +5dB and Q10. That will just drive your amp and Fyne woofers into distortion. You cut peaks, you don't boost nulls that hard.

*3. High-frequency correction is nonsense*
6216.2 -1.38 Q10
6524.4 -2.8 Q10
8559.5 -1.3 Q6
10356.5 +0.1 Q0.1
17357.7 -0.1 Q0.1

Above 5-6kHz in a living room, the mic measures your tweeter's reflections off the sofa, not the speaker. Correcting -0.1 dB at 17kHz with Q0.1 is pure measurement noise. That's why your V13 FINAL was so much better - you left the top alone and only fixed the one real resonance at 6800.

*4. Left and Right don't match in the low end, which is impossible*
LEFT has a huge +4.09 dB at 23.4Hz Q4
RIGHT has +2.72 dB at 23Hz Q10
But then LEFT 27.3 -0.3 Q0.5 and RIGHT 38.4 +0.7 Q0.1

In the same room, under 100Hz, both speakers should see almost the same room modes. This shows the mic was moved, or there was background noise, or the sweep level was too low.

*What to say about this result:*
This scan measured the room noise and mic position, not the speakers. It's unusable as a hi-fi EQ. It explains why everything sounded thin, harsh and disconnected.

This is exactly why you built V13 FINAL manually:
- Few, wide filters under 700Hz where room modes are real
- No crazy Q10 jungle above 1kHz
- Only two intentional high-frequency fixes: 750 HS -1.6 Q0.7 and 6800 PEAK -3.0 Q2.5

Delete this AUTO ORIGINAL and keep V13 FINAL as your reference. If you re-scan, do it at 75-80dB, dead quiet house, mic exactly at ear height, and limit Eversolo to 20-500Hz auto only.