Imaging Issue – Center Image Pulled to the Left

[gianluca_v]

Hi folks,


Lately, I’ve been noticing a significant shift in the soundstage of my stereo system—specifically, voices and some instruments in the upper-mids and low treble range are clearly pulling to the left of center.


This happens regardless of whether I’m using Dirac correction or not. For context, I’m using an Arcam SA30 driving a pair of Dynaudio Special Forty speakers. I haven’t made any major changes to the room or speaker positioning. The only recent addition is a turntable placed in the cabinet about 30–40 cm behind the speaker baffles. I considered whether the plastic dust cover might be introducing some reflections, but it seems unlikely to cause such a noticeable shift. In fact removing such dust cover does not solve the issue.


I initially suspected an issue with the right speaker and swapped the L/R channels, but the imbalance remained unchanged. I’ve also ruled out hearing issues, as the soundstage is perfectly centered when listening through headphones.


At this point, the only remaining suspect would be the amplifier, but it seems unlikely that a subtle fault would only affect imaging without other symptoms.


To investigate further, I ran a set of measurements using a UMIK-1 mic and REW. I captured each channel separately and then both together, with and without Dirac correction. While I’m no REW expert, I didn’t see dramatic differences in SPL between the channels. Phase alignment isn’t perfect but not wildly off either. The only significant difference I noted is in RT60 around 100 Hz.


Here’s a link to the REW measurements: https://drive.google.com/file/d/13g-9AqNKMySv3sKYinfRd_X18M-BSces/view?usp=sharing


If anyone is willing to take a look and share their thoughts, I’d greatly appreciate it!



Thanks!

 

[kemmler3D]

While I’m no REW expert, I didn’t see dramatic differences in SPL between the channels.

I'm on my phone so can't look at REW but if you have even a 1-2dB excess in the left channel in mids or treble it might be enough to cause what you're describing.

 

[gianluca_v]

I'm on my phone so can't look at REW but if you have even a 1-2dB excess in the left channel in mids or treble it might be enough to cause what you're describing.

The only region where one channel is systematically above the other is over 1k (but it is the right channel in green, not the left). In the mid region sometimes they are even, sometimes either of two is slight higher. Graphs are shown with 1/6 smoothing

 

[kemmler3D]

View attachment 458927

The only region where one channel is systematically above the other is over 1k (but it is the right channel in green, not the left). In the mid region sometimes they are even, sometimes either of two is slight higher. Graphs are shown with 1/6 smoothing

Interesting, definitely no smoking gun that I can see here. For an in room measurement I think this looks pretty much even.

I guess if the only other change is the turntable then maybe try removing it entirely to see if the problem goes away?

Edit: it occurs to me that you might want to measure the speakers close-mic'd since room reflections would tend to overwhelm a small defect in the tweeter or something. Try measuring from ~0.5cm away from the drivers and see if anything is different that way?

Or maybe do a gated measurement to see just direct sound at MLP, I think that's possible?

 

[SIY]

This might be a long shot, but... could you have some wax buildup starting in your right ear?

 

[Count Arthur]

Have you tried some test sounds:

Left / Right Stereo Sound Test (Online)

Check your loudspeakers left/right position

www.audiocheck.net

Maybe try it with your eyes shut and point to the apparent source of the sound, so that when you open your eyes, you can see if you are actually pointing off-centre.

Sometimes, objects in the room can draw the eye and your brain can have a tendency to localise sounds to them, if you see what I mean.

 

[Sokel]

This 3-5dB difference up to the most sensitive for the ear freq range might be the guilty party.
If of-axis response is nice and smooth try to tow out the louder one a little and see if it fixes it.

 

[gianluca_v]

This might be a long shot, but... could you have some wax buildup starting in your right ear?

I thougt about it too, but i have no problem when watching tv or listening music through headphones

 

[gianluca_v]

Have you tried some test sounds:

Left / Right Stereo Sound Test (Online)

Check your loudspeakers left/right position

www.audiocheck.net

Maybe try it with your eyes shut and point to the apparent source of the sound, so that when you open your eyes, you can see if you are actually pointing off-centre.

Sometimes, objects in the room can draw the eye and your brain can have a tendency to localise sounds to them, if you see what I mean.

That is exactly the problem: with open eyes the shift to left seems minor, almost unnoticeable. But upon closing the eyes the off-center becomes clearly apparent

 

[respice finem]

Could some early reflexion be the cause? Any new or moved furniture etc.?

 

[gianluca_v]

Could some early reflexion be the cause? Any new or moved furniture etc.?

Nope apart from the turntable addition

 

[Keith_W]

ETC for the left speaker, post Dirac.

ETC for the right speaker, post Dirac. What we see are two early and loud reflections at 1.25ms (about 0.43m) and 2.95ms (about 1m). These are missing from the left ETC. ASSUMING that both measurements were taken correctly, this shows that your right speaker is very close to a wall, about 20cm away. The presence of loud and early reflections on the right would normally cause the reflections to be perceptually fuse to the main sound, and should cause the image to pull to the right. So i'm not sure why you say the image is pulled to the left. Could be that the measurements are mislabelled, or my understanding of psychoacoustics is wrong.

We can see the same pattern when we look at the IACC (interaural cross-correlation). This compares the similarity of left and right across a few frequency bands (63Hz, 125Hz, 250Hz, etc) and across time periods (early, late). A perfect IACC, where the two channels are perfectly identical, is 1. "Good" is above 0.8, and "poor" is below 0.5. Focus on the "early" IACC and you can see that the similarity between left and right is very poor.

When the base is so poor, Dirac can't help very much.

Poor IACC is caused by the left and right measurements being dissimilar in the frequency and time domain. It is very rarely due to your left and right speakers being mismatched, by far the most common cause is room asymmetry.

Your measurements make it pretty obvious that your speaker setup is asymmetrical. Time to think about positioning your speakers better in your room.

 

[gianluca_v]

View attachment 458932
ETC for the left speaker, post Dirac.

View attachment 458933
ETC for the right speaker, post Dirac. What we see are two early and loud reflections at 1.25ms (about 0.43m) and 2.95ms (about 1m). These are missing from the left ETC. ASSUMING that both measurements were taken correctly, this shows that your right speaker is very close to a wall, about 20cm away. The presence of loud and early reflections on the right would normally cause the reflections to be perceptually fuse to the main sound, and should cause the image to pull to the right. So i'm not sure why you say the image is pulled to the left. Could be that the measurements are mislabelled, or my understanding of psychoacoustics is wrong.

Thanks for looking into that. I confirm right speaker is positioned closed to an angle (45cm from rear wall, 25-30 cm from right wall. That is exactly what is puzzling: i would expect the image to be slightly to the right given the setting, but the opposite is true. I guarantee labeling of the measure are correct (in fact the right channel has some boost in the low frequency range due to the angle)

 

[ahofer]

Have you implemented a convolution filter? It would be interesting to see if it solves the problem.

 

[gianluca_v]

Have you implemented a convolution filter? It would be interesting to see if it solves the problem.

Dont have a dsp apart from the dirac live implementation of my arcam sa30

 

[Keith_W]

And BTW I also checked that you took your measurement properly - i.e. mic properly centred between the two speakers. See that double impulse in the first peak of the ETC? One is your left speaker, and the other is the right. The time discrepancy between the two peaks is 0.11ms, or about 3.5cm. Meaning your mic is 3.5cm closer to one speaker than the other. I can't tell if it's closer to the left or right speaker from the ETC, but regardless the mic does seem "acceptably" centred. Mics which are wildly off-centre can affect the measurement and give misleading results, not the case here.

Anyway I think you should try repositioning your speakers and see if the problem goes away.

 

[gianluca_v]

View attachment 458937

And BTW I also checked that you took your measurement properly - i.e. mic properly centred between the two speakers. See that double impulse in the first peak of the ETC? One is your left speaker, and the other is the right. The time discrepancy between the two peaks is 0.11ms, or about 3.5cm. Meaning your mic is 3.5cm closer to one speaker than the other. I can't tell if it's closer to the left or right speaker from the ETC, but regardless the mic does seem "acceptably" centred. Mics which are wildly off-centre can affect the measurement and give misleading results, not the case here.

Anyway I think you should try repositioning your speakers and see if the problem goes away.

Thanks again for this check. Yes I will try to move speaker. Only I hoped that REW measurement would allow me to have a better understanding of the problems and hence the solutions.

 

[gianluca_v]

View attachment 458932
ETC for the left speaker, post Dirac.

View attachment 458933
ETC for the right speaker, post Dirac. What we see are two early and loud reflections at 1.25ms (about 0.43m) and 2.95ms (about 1m). These are missing from the left ETC. ASSUMING that both measurements were taken correctly, this shows that your right speaker is very close to a wall, about 20cm away. The presence of loud and early reflections on the right would normally cause the reflections to be perceptually fuse to the main sound, and should cause the image to pull to the right. So i'm not sure why you say the image is pulled to the left. Could be that the measurements are mislabelled, or my understanding of psychoacoustics is wrong.

View attachment 458935

We can see the same pattern when we look at the IACC (interaural cross-correlation). This compares the similarity of left and right across a few frequency bands (63Hz, 125Hz, 250Hz, etc) and across time periods (early, late). A perfect IACC, where the two channels are perfectly identical, is 1. "Good" is above 0.8, and "poor" is below 0.5. Focus on the "early" IACC and you can see that the similarity between left and right is very poor.

View attachment 458934

When the base is so poor, Dirac can't help very much.

Poor IACC is caused by the left and right measurements being dissimilar in the frequency and time domain. It is very rarely due to your left and right speakers being mismatched, by far the most common cause is room asymmetry.

Your measurements make it pretty obvious that your speaker setup is asymmetrical. Time to think about positioning your speakers better in your room.

Sorry I missed the bottom part of your reply about IACC. I do confirm the room is asymmetric. In fact it is an open space kitchen plus living room. So left speaker has open space to the left while right speaker has a wall to the right and then a glass windows. I know it is pretty poor but there is not much I can do apart from installing some sound absorbing curtains. But I swear that such imbalance was not so relevant in the past despite the setting be identical. Also I dont understand why soundstage is to the left while logic but also measurements would suggest it should be to the right. Anyway thanks again for the detailed check intorno my measurements!!

 

[Keith_W]

That's OK, I was editing my post to include the remarks on the IACC when you posted your reply.

I had a look at the other measurements as well. All of them show the same thing, your right speaker is dominated by early and loud reflections. I don't really see anything else that might account for why your image is pulled to the left, nor can I explain why it happened recently.

 

[Titurel]

View attachment 458937

And BTW I also checked that you took your measurement properly - i.e. mic properly centred between the two speakers. See that double impulse in the first peak of the ETC? One is your left speaker, and the other is the right. The time discrepancy between the two peaks is 0.11ms, or about 3.5cm. Meaning your mic is 3.5cm closer to one speaker than the other. I can't tell if it's closer to the left or right speaker from the ETC, but regardless the mic does seem "acceptably" centred. Mics which are wildly off-centre can affect the measurement and give misleading results, not the case here.

Anyway I think you should try repositioning your speakers and see if the problem goes away.

I agree with your assessment that this is likely a positioning issue, but I would try adjusting toe-in before making any more radical adjustments. From the OP's description, I'd start with right speaker toe-in, increasing or decreasing the toe-in so that the listening position is SLIGHTLY more on axis with the tweeter/midrange, increasing that adjustment until the images locks into the center. If a 15 degree change doesn't help, reset the right speaker and adjust toe in for the left, moving its tweeter/midrange axis gradually away from the listening position. N.b., the wider the distance between the R & L speaker, relative to the listening position, the more critical channel imbalances tend to be. If adjusting toe in doesn't help, try moving each speaker a couple of inches (5 cm) closer to the center line. This could easily be enough to help and it is a small enough change not to disturb your basic room tuning.

Over the years, I chased this problem in multiple systems. Toe in and the L to R distance has usually been the culprit.

My question to the more tech savvy: could differences in 1st reflection spectra also cause the center image to shift to one side? In my present system, only the right speaker has a nearby (1.5m) side wall. The center image tended to collapse left until I put a gobo and a poly-cylindrical diffusor between the right speaker and its side wall, substantially weakening its reflections. Curious?