G3P Elevated BM & LFE Issue

[ttocs]

Here is a comparison of Large Front Left represented by the blue dotted line, and BM for the SL going to the Green Large Fronts (pair), Red Large Center by itself, and just the Blue Large Front Left by itself ( I disconnected the Large Front Right so BM goes to only one speaker ). With only the Solid Blue Large Front Left active, the BM going to it is about the same level as the Red Large Center.

So, BM going to one large speaker is about 8-9dB elevated. BM going to both Large Fronts is about 13-15dB elevated.

[p4t]

Based on number 5 of your test emotivalounge.proboards.com/attachment/download/43287 , I change to large center instead of large fronts, so center channel set as large, and all the rest is small, left sub=mono, center sub= LFE, will the result center channel will also elevated 7Db?

[ttocs]

210723-01
FW2.3
Center Sub LFE
Only one active Large Speaker in each plot
The BM in each is +8dB

Large Left Speaker providing BM for Small SL
 

Large Center Speaker providing BM for Small SL

[muslhead]

Jul 22, 2021 23:52:39 GMT -5 ttocs said:

Here is a comparison of Large Front Left represented by the blue dotted line, and BM for the SL going to the Green Large Fronts (pair), Red Large Center by itself, and just the Blue Large Front Left by itself ( I disconnected the Large Front Right so BM goes to only one speaker ). With only the Solid Blue Large Front Left active, the BM going to it is about the same level as the Red Large Center.

So, BM going to one large speaker is about 8-9dB elevated. BM going to both Large Fronts is about 13-15dB elevated.

View Attachment

This is a far cry from Emotiva's admission they see the elevation at no more than 6db.

[ttocs]

Jul 23, 2021 7:11:40 GMT -5 p4t said:

Based on number 5 of your test emotivalounge.proboards.com/attachment/download/43287 , I change to large center instead of large fronts, so center channel set as large, and all the rest is small, left sub=mono, center sub= LFE, will the result center channel will also elevated 7Db?

Depends. I did those tests without adjusting the levels of any channels after measuring. In other words, I leveled in Processor Levels, then ran the tests and found that the Levels needed adjusting using a 3rd party measuring tool, like REW.

I am finding as I go that the Levels Tool in the processor is being fooled by total output of the speaker being leveled, at least when using the User EQ because there is no automatic correction happening. I don't recall which post it was, maybe several posts, but I showed what happens after using REW to calibrate the levels of the channels. Also, I did not correct anything in the post you quote, but last night I did make some manual corrections to flatten the curve so there are minimal bumps and dips that might affect the measurement, and this is why I chose to keep the Large Channels in each plot for reference and probably should continue to do so.

The tests I ran yesterday and today are with corrected User EQ response curves.
The tests I ran previously were did not have corrected User EQ response curves, and I believe that the peaks and dips affected some observations, but the principle issue is valid, that the BM is elevated.

[ttocs]

Jul 23, 2021 7:35:38 GMT -5 muslhead said:

Jul 22, 2021 23:52:39 GMT -5 ttocs said:

Here is a comparison of Large Front Left represented by the blue dotted line, and BM for the SL going to the Green Large Fronts (pair), Red Large Center by itself, and just the Blue Large Front Left by itself ( I disconnected the Large Front Right so BM goes to only one speaker ). With only the Solid Blue Large Front Left active, the BM going to it is about the same level as the Red Large Center.

So, BM going to one large speaker is about 8-9dB elevated. BM going to both Large Fronts is about 13-15dB elevated.

View Attachment

This is a far cry from Emotiva's admission they see the elevation at no more than 6db.

I'm certain they were quoting the BM being sent to a subwoofer channel vs a Large Speaker Channel. If sent to a single subwoofer channel the BM is elevated about 4dB, if sent to two subwoofer channels then there's an additional 1-1.5dB. I haven't tested 3 subwoofer channels. But even if each additional sub channel from the processor adds 1dB, that's pretty minimal overall assuming that there's no elevated bass to begin with. The issue with +1dB per sub channel would really come into play for RMC users adding subwoofer modules.

[marcl]

I think the testing is important to give Emotiva a lot of information to look for places where the levels are being affected, but they maybe wouldn't have expected. But ultimately this needs a fresh look, specifications and an internal design and code review when the fix is implemented. I know Emotiva knows this. But I'm just saying they will have to take that approach, and not seek to react to measurements until the fix has been tested to verify to the spec.

[ttocs]

FW2.3

I setup the processor to output the Left, Center, and Right channels to miniDSP, then output one channel to an amp and out to One Large Speaker. One Small Speaker channel will then have its BM sent to the One Large Speaker.

Center Sub is still setup as LFE so BM will go to the One Large Speaker.

With only one Large Speaker active, there won't be any summing in-room from multiple speakers.

Single Large Speaker fed by Left, Center, and Right channels from the XMC-2, and a single Small Speaker.
The Large modes tested are:
Large Center, so 1 channel is active and BM is played by the Large Speaker.
Large Fronts, so 2 channels are active and BM is played by the Large Speaker.
Large LCR (Left, Center, Right), so 3 channels are active and BM is played by the Large Speaker.
In every case the BM is hugely elevated above just the Large Speaker playing full range.
edit: changed smoothing to match other plot below.


Match the color of the dotted line to same color solid line for measuring.
Large Center Speaker: BM is +8dB above just the Large Center speaker normal bass.
Large Fronts: BM is +13dB above the Large speaker normal bass.
Large LCR: BM is +13dB above the Large speaker normal bass.

Since it's just one speaker making bass, it's easy to measure without accounting for multiple speaker differences.

[p4t]

Jul 23, 2021 13:04:29 GMT -5 ttocs said:

I setup the processor to output the Left, Center, and Right channels to miniDSP, then output one channel to an amp and out to One Large Speaker. One Small Speaker channel will then have its BM sent to the One Large Speaker.

Center Sub is still setup as LFE so BM will go to the One Large Speaker.

With only one Large Speaker active, there won't be any summing in-room from multiple speakers.

Single Large Speaker fed by Left, Center, and Right channels from the XMC-2, and a single Small Speaker.
The Large modes tested are:
Large Center, so 1 channel is active and BM is played by the Large Speaker.
Large Fronts, so 2 channels are active and BM is played by the Large Speaker.
Large LCR (Left, Center, Right), so 3 channels are active and BM is played by the Large Speaker.
In every case the BM is hugely elevated above just the Large Speaker playing full range.
View Attachment

Match the color of the dotted line to same color solid line for measuring.
Large Center Speaker: BM is +8dB above just the Large Center speaker normal bass.
Large Fronts: BM is +14dB above the Large speaker normal bass.
Large LCR: BM is +14dB above the Large speaker normal bass.

Since it's just one speaker making bass, it's easy to measure without accounting for multiple speaker differences.

Based on your test, center channel get elevated around half of front channel. Center channel get elevated around  7-8 Db. I think to solve this you have to run LCR using minidsp, and make any necessary adjustment using minidsp. By the I ordered dayton 408 dsp, do you know if this anygood? Compare to minidsp? 

[ttocs]

Jul 24, 2021 8:17:18 GMT -5 p4t said:

Jul 23, 2021 13:04:29 GMT -5 ttocs said:

I setup the processor to output the Left, Center, and Right channels to miniDSP, then output one channel to an amp and out to One Large Speaker. One Small Speaker channel will then have its BM sent to the One Large Speaker.

Center Sub is still setup as LFE so BM will go to the One Large Speaker.

With only one Large Speaker active, there won't be any summing in-room from multiple speakers.

Single Large Speaker fed by Left, Center, and Right channels from the XMC-2, and a single Small Speaker.
The Large modes tested are:
Large Center, so 1 channel is active and BM is played by the Large Speaker.
Large Fronts, so 2 channels are active and BM is played by the Large Speaker.
Large LCR (Left, Center, Right), so 3 channels are active and BM is played by the Large Speaker.
In every case the BM is hugely elevated above just the Large Speaker playing full range.
View Attachment

Match the color of the dotted line to same color solid line for measuring.
Large Center Speaker: BM is +8dB above just the Large Center speaker normal bass.
Large Fronts: BM is +14dB above the Large speaker normal bass.
Large LCR: BM is +14dB above the Large speaker normal bass.

Since it's just one speaker making bass, it's easy to measure without accounting for multiple speaker differences.

Based on your test, center channel get elevated around half of front channel. Center channel get elevated around  7-8 Db. I think to solve this you have to run LCR using minidsp, and make any necessary adjustment using minidsp. By the I ordered dayton 408 dsp, do you know if this anygood? Compare to minidsp? 

I'm not familiar with the Dayton, so I'm no help with that.

The BM going to Large Channels is not "fixable" by using a workaround. If I were to lower what I "think" the BM level is, that won't address the exact crossover setting being used, and further, I use multiple crossover settings. No, this needs to be fixed at the source. I don't recall this being at this level in older firmware, so I might test this to find out. I use FW2.0 as a baseline comparison for things like this.

I stopped using Large Channels for BM when I found a good solution to all the BM stuff using the Left Sub Output, I bought a bunch of subs. So I use L Sub for BM, C Sub for LFE, run both channels to miniDSP and output both to the same 3 subwoofers, then I make the LCR all Large and the rest Small, and the Center Speaker gets its own subwoofer right up front under the speaker. With this setup I can individually level BM and LFE to be perfect, at least for Surround, but I can't do anything about All Stereo.

[p4t]

Jul 24, 2021 8:49:15 GMT -5 ttocs said:

Jul 24, 2021 8:17:18 GMT -5 p4t said:

Based on your test, center channel get elevated around half of front channel. Center channel get elevated around  7-8 Db. I think to solve this you have to run LCR using minidsp, and make any necessary adjustment using minidsp. By the I ordered dayton 408 dsp, do you know if this anygood? Compare to minidsp? 

I'm not familiar with the Dayton, so I'm no help with that.

The BM going to Large Channels is not "fixable" by using a workaround. If I were to lower what I "think" the BM level is, that won't address the exact crossover setting being used, and further, I use multiple crossover settings. No, this needs to be fixed at the source. I don't recall this being at this level in older firmware, so I might test this to find out. I use FW2.0 as a baseline comparison for things like this.

I stopped using Large Channels for BM when I found a good solution to all the BM stuff using the Left Sub Output, I bought a bunch of subs. So I use L Sub for BM, C Sub for LFE, run both channels to miniDSP and output both to the same 3 subwoofers, then I make the LCR all Large and the rest Small, and the Center Speaker gets its own subwoofer right up front under the speaker. With this setup I can individually level BM and LFE to be perfect, at least for Surround, but I can't do anything about All Stereo.

Please later on update your best setting, as I am already ready with a bunch of subwoofers, and aside of minidsp 2X4 I already owned and another minidsp 2x4HD and Dayton 408 on the way. ;-)

[ttocs]

Jul 24, 2021 10:55:05 GMT -5 p4t said:

Jul 24, 2021 8:49:15 GMT -5 ttocs said:

I'm not familiar with the Dayton, so I'm no help with that.

The BM going to Large Channels is not "fixable" by using a workaround. If I were to lower what I "think" the BM level is, that won't address the exact crossover setting being used, and further, I use multiple crossover settings. No, this needs to be fixed at the source. I don't recall this being at this level in older firmware, so I might test this to find out. I use FW2.0 as a baseline comparison for things like this.

I stopped using Large Channels for BM when I found a good solution to all the BM stuff using the Left Sub Output, I bought a bunch of subs. So I use L Sub for BM, C Sub for LFE, run both channels to miniDSP and output both to the same 3 subwoofers, then I make the LCR all Large and the rest Small, and the Center Speaker gets its own subwoofer right up front under the speaker. With this setup I can individually level BM and LFE to be perfect, at least for Surround, but I can't do anything about All Stereo.

Please later on update your best setting, as I am already ready with a bunch of subwoofers, and aside of minidsp 2X4 I already owned and another minidsp 2x4HD and Dayton 408 on the way. ;-)

Take a look at the thread Finding Subwoofer for this. We can discuss setups more freely there so we can keep this thread only for Elevated BM. The second post always has my current setup.

[ttocs]

Reverted back to FW2.0 for this.

Single Large Speaker fed by Left, Center, and Right channels from the XMC-2, and a single Small Speaker.
The Large modes tested are:
Large Center, so 1 channel is active and BM is played by the Large Speaker.
Large Fronts, so 2 channels are active and BM is played by the Large Speaker.
Large LCR (Left, Center, Right), so 3 channels are active and BM is played by the Large Speaker.
In every case the BM is hugely elevated above just the Large Speaker playing full range.


This seems logical. Still needs to be fixed, but it's logical.

[p4t]

Jul 24, 2021 11:33:16 GMT -5 ttocs said:

Jul 24, 2021 10:55:05 GMT -5 p4t said:

Please later on update your best setting, as I am already ready with a bunch of subwoofers, and aside of minidsp 2X4 I already owned and another minidsp 2x4HD and Dayton 408 on the way. ;-)

Take a look at the thread Finding Subwoofer for this. We can discuss setups more freely there so we can keep this thread only for Elevated BM. The second post always has my current setup.

Got it, ttocs.

[ttocs]

Having gone back and forth with firmwares a few times this week, I was starting to doubt the REW plot from a couple days ago, but just reconfirmed the previous results with new identical results. It just isn't logical what the results are, but they are correct.

Also, I saved settings in each firmware version and tried to use the Levels "as saved", but couldn't. The different channels are simply at different levels. Not all the same. Each channel is unique and needs to be checked. We cannot rely on updating firmware and using saved settings and expecting the levels to be the same as before. They are not the same. And it was driving me nuts that I'd have to re-level each time. But after doing it four times in a row I saw the pattern.

This is a comparison of FW2.0 and FW2.3 of the Single Large-Single Small Channel Test. Or, as Kramer would say, "The SLSSCT".
FW2.0


FW2.3


The differences between them would be noticeably different even with the lights off. FW2.0 looks logical, with each additional channel added for BM the overall BM level goes up, even with only one speaker making sound. But with FW2.3, one channel is elevated quite a bit, two and three channels are the same highest elevated.

[marcl]

Jul 24, 2021 13:36:46 GMT -5 ttocs said:

Having gone back and forth with firmwares a few times this week, I was starting to doubt the REW plot from a couple days ago, but just reconfirmed the previous results with new identical results. It just isn't logical what the results are, but they are correct.

Also, I saved settings in each firmware version and tried to use the Levels "as saved", but couldn't. The different channels are simply at different levels. Not all the same. Each channel is unique and needs to be checked. We cannot rely on updating firmware and using saved settings and expecting the levels to be the same as before. They are not the same. And it was driving me nuts that I'd have to re-level each time. But after doing it four times in a row I saw the pattern.

This is a comparison of FW2.0 and FW2.3 of the Single Large-Single Small Channel Test. Or, as Kramer would say, "The SLSSCT".
FW2.0


FW2.3


The differences between them would be noticeably different even with the lights off. FW2.0 looks logical, with each additional channel added for BM the overall BM level goes up, even with only one speaker making sound. But with FW2.3, one channel is elevated quite a bit, two and three channels are the same highest elevated.

So are you saying the whole thing is complicated by the fact that - all else being equal - levels change with each different firmware version?   I think we always suspected, irrespective of the bass management issues.

[ttocs]

Jul 24, 2021 14:04:34 GMT -5 marcl said:

So are you saying the whole thing is complicated by the fact that - all else being equal - levels change with each different firmware version?   I think we always suspected, irrespective of the bass management issues.

Yes.

I saved settings from the "previous" firmware I was going "from", so I could use it in the next version I went "to", but it wouldn't work. Each channel was off by a different amount. Keep in mind that the LCR were all the same SPL, but after loading the different firmware version I had to lower the Left and Right channels by -4dB to match the Center Channel, but the SL was the same SPL as the Center. As Jodi Foster said in Alice Doesn't Live Here Anymore "It's the weird capital of America".

I always thought the difference in level was to the "overall" volume level and all channels were the same amount of up or down. But no. Each channel can be different, at least with FW2.3. It may be that 2.3 is the oddball here. FW2.0 and 2.1 are very similar with respect to BM, but I didn't go through the effort to measure overall levels of 2.1, just saying that I used 2.0 purposely as it's the earliest firmware with Dirac should I want to compare things related to Dirac.

[markc]

Jul 24, 2021 8:49:15 GMT -5 ttocs said:

it may be a solution but that most definitely is not a good solution!

I am not buying a bunch of subs and an external device like your miniDSP to do the (simple, protocol driven) bass management that my £3000 processor should do.

Bass management principles are very simple and the process established for many years.

How can companies fail to implement this after all these years.

While still not ideal, I have pretty good surround and ceiling mounted Top speakers so the better workaround for me is to set the speakers to large for the time being.
After redoing the channel level matching, the obvious thing was how clear the spoken audio was compared with before, using all speakers set to small, and this with the XMC-2 main volume being turned down.

presumably this is because the overamplified diverted bass plus LFE signal was being attenuated to avoid clipping along with the remaining audio in the extracted bass small speaker outputs.

[ttocs]

Jul 26, 2021 9:17:33 GMT -5 markc said:

Jul 24, 2021 8:49:15 GMT -5 ttocs said:

it may be a solution but that most definitely is not a good solution!

I am not buying a bunch of subs and an external device like your miniDSP to do the (simple, protocol driven) bass management that my £3000 processor should do.

Bass management principles are very simple and the process established for many years.

How can companies fail to implement this after all these years.

While still not ideal, I have pretty good surround and ceiling mounted Top speakers so the better workaround for me is to set the speakers to large for the time being.
After redoing the channel level matching, the obvious thing was how clear the spoken audio was compared with before, using all speakers set to small, and this with the XMC-2 main volume being turned down.

presumably this is because the overamplified diverted bass plus LFE signal was being attenuated to avoid clipping along with the remaining audio in the extracted bass small speaker outputs.

My setup is only a solution for Surround. Nothing I can think of will allow All Stereo to function as it used to with a mix of Large and Small channels.

Having All Large Channels is the best one of all. As I found this weekend the Large Channels are unaffected by different firmware versions when using Surround. I haven't tested All Stereo with All Large Channels but I suspect it would act the same as Surround because the most affected channels are Small.


edit: I just checked and it seems that when any channel is Large, it is not affected like the BM for Small channels. So, after testing Surround, DD Surround, DTS, and All Stereo, I would say that All Large Channels and Center Sub LFE seems like it would be the best scenario and be unaffected by whatever is going on with the firmware affecting Small Channels' BM.

For my system, which has 8 speakers with small woofers for surrounds and tops, I'd need to use something like a miniDSP 10x10 to get the 8 small speakers protected and send the bass to the subs controlled by other miniDSP's. This would then allow Dirac to see each channel as full range and calibrate full range. And since my subs are aligned as a group, which means there's no need for DLBC - even though I'd love to use it - then it's just whatever little tweaks might be needed for each speaker's crossover point.

[markc]

eek.
All Large for me it is, even though I'd rather not.

Good work ttocs.

This really is a shambles though!
Even though I have fairly capable speakers, the Dolby specs and recommendations still mean that bass management and bass redirection to the subwoofer is best practice.
My in ceiling Atmos Tops and the wall mounted surrounds may be OK on paper, but if they do get a full range signal which includes bass that hasn't already been extracted to the .1 in the source, they are not going to be that great!