Levels, Levels, or Levels, Which One Is Correct?

[ttocs]

I've been trying to understand which Levels test is correct. We can use the Levels Tone Generator in the processor, REW, OPPO Tone Generator, MKV files, Dolby ATMOS Test Disc, Apple TV Studio Six Digital ATMOS app, but which one is correct?

I get a variety of results that largely depend on which Sound Field is used, 5.1, 7.1, ATMOS, etc. If running the test audio through the processor it makes a difference which decoding is used. Also, there is a overall system output level difference between 5.1/7.1 and ATMOS. This is not a G3P only thing, it's replicated with the OPPO and using several Sound Field sources. Don't know why, but it just "is".

I'm getting some confusing results and am setting up for some official testing that I want to document. The hope is that some good ideas get offered up as to why some devices are playing differently. I'm going to post results as simply as possible, and to that end it will only involve two channels: Front Left, and LFE. LFE in this case, most of the time, means Subwoofer, which is erroneous from the outset but this is what OPPO is calling the subwoofer channel and the same is true for the Studio Six app but at least they put (Subwoofer) in parentheses so I'll give them partial credit.

If anyone can test and put together the results in a easy to read format, please do. Please include all devices used, the Sound Field used and what the system is setup as, and if using a G3P try using Auto, Direct, Surround, and DD Surround, because in some cases Auto will produce a different result.

On to the testing!

[405x5]

Feb 20, 2021 13:05:56 GMT -5 ttocs said:

I've been trying to understand which Levels test is correct. We can use the Levels Tone Generator in the processor, REW, OPPO Tone Generator, MKV files, Dolby ATMOS Test Disc, Apple TV Studio Six Digital ATMOS app, but which one is correct?

I get a variety of results that largely depend on which Sound Field is used, 5.1, 7.1, ATMOS, etc. If running the test audio through the processor it makes a difference which decoding is used. Also, there is a overall system output level difference between 5.1/7.1 and ATMOS. This is not a G3P only thing, it's replicated with the OPPO and using several Sound Field sources. Don't know why, but it just "is".

I'm getting some confusing results and am setting up for some official testing that I want to document. The hope is that some good ideas get offered up as to why some devices are playing differently. I'm going to post results as simply as possible, and to that end it will only involve two channels: Front Left, and LFE. LFE in this case, most of the time, means Subwoofer, which is erroneous from the outset but this is what OPPO is calling the subwoofer channel and the same is true for the Studio Six app but at least they put (Subwoofer) in parentheses so I'll give them partial credit.

If anyone can test and put together the results in a easy to read format, please do. Please include all devices used, the Sound Field used and what the system is setup as, and if using a G3P try using Auto, Direct, Surround, and DD Surround, because in some cases Auto will produce a different result.

On to the testing!

Seems clear to me you understand better than many and have answered your own questions.
What’s important is once you’ve established a “baseline” by whatever method that works best for you, that your not afraid to tweak your non persistent speaker settings on the fly, depending on the current source material you’re listening to. Results will always vary a bit.

[DYohn]

The only level that matters is what you measure in your listening area. Since you are not dealing with a calibrated system, none of the so-called reference levels will have much meaning outside the device generating it. Measure the output and adjust to achieve 85dbSPL and that is as close to a reference level as you can get - realizing of course that your measurement tool is likely not a 100% accurate indication either. It's all relative.

[ttocs]

Thanks for the responses.
As an FYI, I'm not attempting to level to a predetermined SPL, I'm just wondering why subwoofer output varies with the sound field employed.

I don't mind at all the varying volume differences from one source to another, or one commercial vs a tv show. And I never level the system to any SPL. Everything in the system is relative, so I just level down to the lowest level speaker and adjust the volume based on content.

[ttocs]

Here's the first batch. Just using an OPPO 105 connected directly to the Front Left amp and the subwoofers. The app on the ATV4K is the Studio Six Digital ATMOS tone generator. The disc is from Dolby.

NOTE: There is no processor used for this. It is only an OPPO, ATV, 1 amp, 1 subwoofer. The OPPO has its own outputs that go directly to the amp and subwoofer.



    

[DYohn]

Feb 20, 2021 14:36:10 GMT -5 ttocs said:

Thanks for the responses.
As an FYI, I'm not attempting to level to a predetermined SPL, I'm just wondering why subwoofer output varies with the sound field employed.
.

Every decode scheme uses a different algorithm for speaker levels, especially for the subwoofer. Your system seems to be working the way it is supposed to work.

[ttocs]

Feb 20, 2021 14:41:41 GMT -5 DYohn said:

Feb 20, 2021 14:36:10 GMT -5 ttocs said:

Thanks for the responses.
As an FYI, I'm not attempting to level to a predetermined SPL, I'm just wondering why subwoofer output varies with the sound field employed.
.

Every decode scheme uses a different algorithm for speaker levels, especially for the subwoofer. Your system seems to be working the way it is supposed to work.

While I agree that the OPPO is outputting in a similar fashion to the XMC-2, and that it may be the "correct" way for it to operate. I question why the subwoofer level varies within a range of 20dB based solely on the sound field and what the downmix might be.

When I first tested the XMC-2 a few weeks ago, I was stunned by the variance. I had no idea that the subwoofer could be -10dB with one setup and +10dB with another. With the tests I just posted the only difference within a test was the audio track channel count. When I changed the speaker config of the OPPO from 7.1 to 5.1 the results changed, but this part I understand there will be a difference. And as the results show, at least for the Studio Six app, while the numbers look way different - they are simply offset from the 7.1 numbers by 4. Slide the numbers down by 4dB and they are identical to the 7.1 results. The Dolby disc has more variation.

Questions remain, but maybe they will be answered once a better pattern emerges. But the OPPO is taking a disc made by Dolby with Dolby tracks and changing how loud the subwoofer is vs the Large Front Left. Next test is the Small Center w/XO=120Hz, so it will be the same subs doing BM and LFE (presumably LFE).

[ttocs]

This time the Speaker Config in the OPPO is 7.1, and the measurements are from the Center Speaker Small w/XO=120Hz, and LFE. So it's bass vs bass. Should always be a 10dB difference, right?

NOTE: There is no processor used for this. It is only an OPPO, ATV, 1 amp, 1 subwoofer. The OPPO has its own outputs that go directly to the amp and subwoofer.

[marcl]

Here's my version of this.  (Yes, ttocs and I do talk  )



I have used the XMC-2's Levels - after Dirac calibration - to verify that every channel is within <1db of the Left channel reference, including the Center Subwoofer which plays only LFE.  I didn't touch the XMC-2 throughout the test, had Volume set to -20db, and it's fine that different sources happen to play at different nominal levels.  So with the Dolby Demo BD the left channel is always 66db; with the Surround Generator on ATV4K the left channel is always 57db; and with the noise generator in REW the left channel is always 72db.

1. I can see no justification for the LFE channel to have different values for the Dolby Demo BD.  They should all be the same, and I think they should be 76db ... 66 + 10
2. I understand why the LFE value for the Surround Generator is different for Full bandwidth vs Mid.  Even though I don't know where Mid rolls off, it would have less bass.  But once again. I would expect the value for Full to be 57 + 10 = 67 .... certainly NOT +25db higher than the Left channel
3. With REW I would expect the LFE to be 82db ... 72 + 10.  And this should be the same for both the Full bandwidth test and the limited bandwidth where the speaker was measured 500-2KHz and the LFE 30-80Hz.

+/- 1db I could understand ... but especially the Dolby BD ... I see no way the LFE should be different for any of the configurations.

[DYohn]

You guys are way overthinking this, and like I implied before, assuming that your relative levels have anything to do with each other. Your systems are not calibrated, so anything they do is unpredictable. Even within the Dolby decode schemes LFE level can vary by +/- 10db and still be "correct" from one piece of gear to another. Does it sound OK? Yes? Good. No? Adjust it.

[marcl]

Feb 20, 2021 16:42:46 GMT -5 DYohn said:

You guys are way overthinking this, and like I implied before, assuming that your relative levels have anything to do with each other. Your systems are not calibrated, so anything they do is unpredictable. Even within the Dolby decode schemes LFE level can vary by +/- 10db and still be "correct" from one piece of gear to another. Does it sound OK? Yes? Good. No? Adjust it.

I understand that for the same source and Volume setting, ttocs system will play at one volume and mine at another (they do, and by a LOT).  But not different per channel.

We have to be able to set a baseline where all the channels are balanced to each other.  From there, the Dolby spec says the processor should play all the channels at the nominal level except the LFE.  The LFE should be +10db.   Isn't that correct?

So for any source or configuration, on a given system all the channels should play at X db, and the LFE should play at X+10db, right?

[ttocs]

This is one that anyone can try.

XMC-2
Set Fronts as Large.
Set Center Sub as Mono, and Left/Right Sub:Nono.
Go to Levels and verify the level of the Fronts, mine are 65.4dB.
Now change the Center Sub and make it LFE.
Go to Levels and verify the level of the Fronts is the same as above.

Now play something on tv that has a constant volume, like a musical commercial.

When I play that commercial while the Center Sub is LFE the volume of the Fronts is 64.7 SPL peak.
When I play that commercial while the Center Sub is Mono the volume of the Fronts is 76.8. That's an increase of 12dB just from the Fronts!
The Center speaker and all other speakers never changed volume !!!!!!!   Let me repeat that, no other speakers changed volume, but the Fronts did, and processor Levels still shows the Fronts as being the same volume as the other speakers.

The Center Sub changed volume. It got 4dB quieter when setup as Mono. You would think it would get louder with more bass content running through it, but the opposite happened.

Changing anything in speaker size setup does not change any Levels for any speaker or subwoofer. The Levels for all speakers and subwoofers all remain the same before and after changing the Center Sub between LFE and Mono.

The exercise in this thread is to try to discover the root of the differences. I only want to know if setting Fronts up as Large has a flaw in how it messes up the volume level of the Fronts and Subwoofer.

[DYohn]

If your channels are at different levels when playing a test tone then either one is attenuated in comparison to the other in your speaker setup or there is a problem in your processor. If the signal level is the same but the in room measured level is different then either you have an amplifier issue, or a loudspeaker issue, or a room interaction issue. If you think you have a problem you need to systematically go through those one at a time. If on the other hand you have tested each thing independently and find no issue then it is possible your measurement technique is the problem.

[ttocs]

Sorry if I'm not explaining well enough David.

If I change one setting in the processor, the Front channels are 12dB louder the instant the setting is changed. That setting is Center Sub:: LFE or Mono.
The other key setting is that the Fronts are setup as Large. Can't happen if all speakers are Small.
When doing this and checking Setup::Speakers::Preset1::Levels, the Levels pink noise never changes level in any speaker no matter how the subs are setup.
Source can be DD5.1, mode Direct.

This happens with any speaker setup as Large, but the actual volume difference is dependent on audio content in that channel.

That is only one small part of things I've noticed and not any part of the tests posted above.


This is the flow used for the tests above. OPPO>Amp/Speaker, OPPO>Subwoofer, ATV>OPPO.

[DYohn]

Yes and sorry if I’m being obtuse. . Sounds like a processor quirk. Since I no longer own an Emotiva processor I guess I can’t help you.

[megash0n]

Feb 20, 2021 19:31:35 GMT -5 ttocs said:

Sorry if I'm not explaining well enough David.

If I change one setting in the processor, the Front channels are 12dB louder the instant the setting is changed. That setting is Center Sub:: LFE or Mono.
The other key setting is that the Fronts are setup as Large. Can't happen if all speakers are Small.
When doing this and checking Setup::Speakers::Preset1::Levels, the Levels pink noise never changes level in any speaker no matter how the subs are setup.
Source can be DD5.1, mode Direct.

This happens with any speaker setup as Large, but the actual volume difference is dependent on audio content in that channel.

That is only one small part of things I've noticed and not any part of the tests posted above.


This is the flow used for the tests above. OPPO>Amp/Speaker, OPPO>Subwoofer, ATV>OPPO.

To be sure I understand this thread, in general... Are we basically stating an opinion that, 3ish years later, we still have basic channel routing and volume issues? At a high level. That is what I am gathering, but want to be sure that, upon months of FW development to arrive at 2.2, we still have fundamental issues?
I'm asking because I really haven't spent more than 2 hours with my G3P since upgrading, and all that time was simple documentary watching on Amazon Prime Video.

[ttocs]

Feb 21, 2021 8:20:52 GMT -5 megash0n said:

Feb 20, 2021 19:31:35 GMT -5 ttocs said:

This is the flow used for the tests above. OPPO>Amp/Speaker, OPPO>Subwoofer, ATV>OPPO.

To be sure I understand this thread, in general... Are we basically stating an opinion that, 3ish years later, we still have basic channel routing and volume issues? At a high level. That is what I am gathering, but want to be sure that, upon months of FW development to arrive at 2.2, we still have fundamental issues?
I'm asking because I really haven't spent more than 2 hours with my G3P since upgrading, and all that time was simple documentary watching on Amazon Prime Video.

The main purpose of this thread is to figure out why Subwoofer vs Fronts levels seem so wacky, and it's not an Emotiva thing as the results show. This diagram has no Emotiva processor, it' ONLY an OPPO, ATV, Amp, 1 speaker, 1 subwoofer. THIS is what is providing all of the results in the posted spreadsheets.

My question is why is there no rhyme or reason to the results I'm getting while using a Dolby Disc? It seems to be baked into the Dolby spec that some speaker setups have one result from one audio sound field and others have a different result from another sound field. Forget about that Mono/LFE thing in the processor. I never even got around to delving into that one because LFE works so well for me with Large Fronts. And the LFE/Mono thing is not at the root of what I'm trying to find.

[DYohn]

What you are seeing is the result of the bass management built into WHATEVER is controlling the various relative signal levels (your OPPO DVD player?) IN GENERAL things tend to work like this:
1. LFE level is set by the decoding shame (Dolby Digital, Atmos, DTS, whatever) and assumes a 120Hz low-pass (the default for LFE channel). Your Dolby test disc will contain 20-120Hz in teh LFE channel, and if your speakers are set to large your processor MAY be doubling this information in both the LFE and the main channel.
2. LFE channel will always contain the LFE encoded program material PLUS whatever is being directed to it from the bass-management processor.
3. Usually, when a main speaker (L-C-R-Surround) has a crossover set, program material below that Xover point is sent to the LFE (or sub) channel and not to the speaker.
4. In some cases, when a L-C-R-Sur is set to large AND has a crossover set, the Xover is ignored (set to large) and the low frequency program material below the Xover setting is DOUBLED by being sent to both the LFE and the large speaker. In other cases, if the speaker is set to large then noting is doubled to the LFE channel. You need to understand your bass management processor to know which you have.
5. In some cases, the LFE channel is fixed and will only play whatever is encoded in the LFE program content, and if the L-C-R-Sur is set to small you lose contecnt below the Xover setting (this is rqre and commonly only in older processors)
6. In some cases, your speaker Xover settings do not matter as the bass management is always pre-set based on the decoding shceme called for by the program content

[ttocs]

DYohn
You bring up something that's important to consider, and that is to setup the OPPO with all the speakers Small for one test, then as Large for the next. This way, all bass goes to the sub channel for the Small test, and only LFE goes to the sub channel for the Large test.

I'm using only the OPPO for several reasons. First is that it's easy because it can be connected to an amp directly, but the biggest reason is that it's an autonomous device that doesn't need my XMC-2. Plus it's been around for a long time and can be considered as a product that is well setup.

The end result of all this is for a comparison with the XMC-2. I have been flummoxed by the seemingly inconsistent nature of how specific channels are altered in volume level vs the other channels when using audio test sources. It's looking like the OPPO is doing things very similar to the XMC-2, so it may simply confirm same behavior.

So while writing this I did a quick check of the AllSmall vs AllLarge and the results are more consistent to each other vs SomeLargeRestSmall. This is steering into a good direction. Thanks David.

[DYohn]

The key to any analysis like this is understanding what the bass management processor is doing - or what it is SUPPOSED to do anyway given the limited information often available in owner's manuals. And bercause the information is limited it often becomes more frustrating than it is worth. IMO, anyway.

I hope you are only using the OPPO for this test and not setting your system up this way permanently? Use a proper processor/preamp. The analog outs from the Oppo are designed to be connected to the analog ins on a preamp.