Bass management & LFE interaction

[LDS]

Originally Posted by BScout

Though slightly unclear, seems to be showing a 2 subwoofer/LFE setup and you can see two LPF being applied for both subs in series one pre and post summation.
(It is unclear, though. Not the way I'd draw it as there should be variable symbol and distinction on which LPF filters are which.)

Yeah, it's not the most enlightening diagram. It seems to imply that the LFE signal and filter is being passed through the bass management filter, which doesn't make any sense.

I wonder if someone from DAD is floating around and can clarify what is happening in the signal path?

[ntp-jl]

I will post a detailed explanation tomorrow.

[ntp-jl]

So, I am venturing into the lion's den with a detailed explanation. Before going into all the technical details I will start off with something which is hopefully common knowledge for all, but still worth refreshing.

1. The SPQ function is a speaker EQ, i.e. it is part of the "speaker system".
2. Bass Management (BM) is a function in a speaker system which is used to "distribute" low-frequency content to one or more subs in order to get a "truthful" reproduction of the audio signal.
3. The LFE channel is not a function in the speaker system, but is a channel used by the sound engineer for creative purposes.
4. The technical requirements (mainly frequency range) for BM and the LFE channel may differ. The LFE channel goes up to 120 Hz, whereas the crossover frequency for the BM is decided by the speakers.

The block diagramme of the SPQ function on our website is, as mentioned by others, not entirely accurate. I have made a simplified block diagramme with a focus on BM and LFE, see "Slide 1", based on a 2.1 set-up.

Filter A is a highpass filter which removes the low-frequency content of a main channel before sending it out to a main speaker.
Filter B is a lowpass filter which removes the high-frequency content of a main channel and passes it on the BM summing.
Filters A and B should obviously be adapted to the crossover frequencies of the main speakers and the sub. It is possible to set cut-off frequency, slope and filter type indivifually for each main channel and filter.
It is also possible to remove filters A and B so you can send a fullrange signal to the main speakers, or even send a fullrange signal to the BM summing, if you desire so.

The low-frequency content of the main channels is then summed with the LFE channel (with or without a 10 dB boost). There is no filter on the LFE channel.
There is a lowpass filter, C, on the output of the BM/LFE summing.
You can of course set the cut-off frequency, slope and filter type for this filter.
It is also possible to remove this filter.

If you look at the low-frequency signal path from a main channel to the sub output, the signal will first pass through filter B and then filter C. As the two filters are in series, the resulting frequency response will depend on the cut-off frequency and slope of each filter. If the cut-off frequency of the two filters is close to each other then the impact will be bigger than if they are far from each other.

I think the big question is whether it is relevant to put a lowpass filter on the LFE channel, or not. Filter B will obviously remove high-frequency content so there will not be any high-frequency content coming from the main channels.

The LFE channel is as mentioned a "creative tool" for the sound engineer to provide low frequency effects. There should therefore not be any content above 120 Hz in that channel.

It should therefore not be necessary to apply filter C as all sources to that channel are strictly low-frequency. And in any case, most subs have a lowpass filter so filter C is effectively redundant.

"Slide 2" shows where the link between the BM presets in DADman and the respective filters in case you want to play around with it. It is also possible to manually remove or change any filter in the monitor profile.

Finally, the SPQ function supports up to 8 subs in a speaker configuration. Each sub can be configured to be either BM only, LFE only or a combination of BM and LFE. It is also possible to direct individual main channels to different subs in the BM, for example if you want a sub for the left side and a sub for the right side.

I hope this clarifies most questions, and please feel free to enlighten me if you have other experiences, knowledge or opinions.

[darrancummins]

Hi Jan,

thanks for your detailed response. I am assuming a single subwoofer, handling both Bass Management and LFE - this is far and away the most common setup I've come across.

I think that DADman's implementation here is confusing, and does mix up (for the user) the LFE with the subwoofer; (e.g. "LFE" Low Pass Filter in Bass Management window, vs "Level Sub 1" in Groups window. I know it's been confusing to me!

Originally Posted by ntp-jl

It should therefore not be necessary to apply filter C as all sources to that channel are strictly low-frequency. And in any case, most subs have a lowpass filter so filter C is effectively redundant.

Filter C is automatically applied as part of the default bass management, and appears at first glance (at least to me) to be the standard Dolby LFE 120Hz cutoff - I would expect to see this applied to LFE channel only. (See for example Wikipedia bass management signal flow - https://en.wikipedia.org/wiki/Bass_management - this is how I've always understood it, and how it is applied by Genelec, Grace Design's m908, Trinnov, etc).

Several subwoofer models extend above 120Hz (e.g. Dynaudio = 175 - 230Hz), so I would assume this 120Hz LPF would indeed need to be applied somewhere.

I did a little more measuring here today of the filter responses - see graphs below. (All measured direct from MTRX Studio input to output, no actual speakers or subs involved).

The basic bass management (main channel HPF and LPF) looks perfect - with the LFE filter type set to "None", these sum together to a flat response - all good:



However, adding back in the default (120Hz) LFE Low Pass Filter causes quite a significant dip in the response - it's actually a lot larger than I had previously thought. This is the effect of just the standard "LFE" low-pass filter on all main (bass-managed) channels, using the default bass-management preset. I would consider this to be quite severe:



I then tried a few things to mitigate this effect, e.g. changing the LFE filter type to Butterworth (BW), pushing it's cutoff frequency higher, and also moving the main channel BM crossover lower to 70Hz. These help slightly, but I don't think any are ideal:



Looking at these graphs, I would say that if bass management is used within DADman (with only a single subwoofer, to carry LFE as well - as is usually the case), then this "LFE" filter should be turned off, and that any LFE filtering would have to be done elsewhere (e.g. in DAW).

[ric982]

Hi,
Old thread but I thought I'd chime in.

I have two subs, but they only have one input each and I'm driving them from a MTRX Studio with separate outputs with different EQ, but both sourced from same LFE input and BM from satellites.

I modeled the transfer functions of the filters in a spread sheet. With an 80 Hz LR-2 (24dB) for the crossovers and a 150 Hz Butterworth-4 (24 dB). The graphs are attached below. If I didn't screw things up, the dip in the combined response (Crossover HPF + (Crossover*LFE) LPF) occurs at about 150 Hz and the peak deviation is on the order of -0.10 dB. I think you could add some EQ to get it back, but for that small an error, I'm not sure its worth the trouble (e.g. you could boost the sub here and normal satellites + BM would be +0.0 db and LFE would be +0.1 dB).

Also, you can remove the LFE LPF on the subwoofers in DADman and you can add it back to the LFE somewhere else in the monitoring chain if you want to get read of this problem all together.

[darrancummins]

Hi ric982,

thanks for the input. I would indeed have thought that any interaction between the LFE and BM filters being in series would probably have been pretty small, like your graphs suggest. (Are these theoretical / calculated filter curves, or did you actually measure the MTRX hardware itself?)

However, I found much more severe effects - both in the field (at a client's studio, measuring the speakers' responses in the room), and measuring just the filters themselves, under more "laboratory" conditions with our demo system. Both using MTRX Studio hardware, with DADman etc. Given the severity of the dips, it looks to me like some kind of additional phase cancellation, or something else very weird, is happening here. (I setup and made these measurements three separate times because I couldn't believe the results, and was sure I'd made a mistake - but got the same results every time).

For the client studio, we removed the LFE filter from DADman altogether, and inserted it in the DAW instead. This gave us back our proper, flat bass-management response for all the main speaker channels, and the system is now nicely calibrated and all sounds great.

But I still think this whole DADman bass management function seems very odd, and misleading.

[ric982]

Originally Posted by darrancummins

Hi ric982,

Are these theoretical / calculated filter curves, or did you actually measure the MTRX hardware itself?

...

For the client studio, we removed the LFE filter from DADman altogether, and inserted it in the DAW instead. This gave us back our proper, flat bass-management response for all the main speaker channels, and the system is now nicely calibrated and all sounds great.

But I still think this whole DADman bass management function seems very odd, and misleading.

No measurements. The curves were generated in Excel using its complex math functions to model the transfer functions a of 4 pole Linkwitz-Reily HPF/LPF and a 4-pole Butterworth LPF (you can find details on Wikipedia). A Linkwitz-Riely 4 pole is two butterworth 2-pole filters in series, so you can just square the complex output of the 2 pole Butterworth fitler to get the Linkwitz. Basically, you need to model the transfer functions of Butterworth 2 pole and 4 pole filters, which have very specific definition, to generate the curves.

One thing you might check - MTRX SPQ has both a 2-pole and a 4-pole Butterworth. Use the 4 pole for the LFE - it gets flatter faster as you move away from the cutoff. You can see this in the DADman, at least quantitatively. If you the expand the EQ on the Monitor Output that has the LFE and bass management, you can switch between 2 pole Butterworth, 4 pole Butterworth, and 4 pole Linkwitz, and see which is flattest.

I agree that it's not ideal that the SPQ has the BM going through the LFE filter. Most subwoofers do this as you expected. Moving it to the DAW avoids the issue.

One other possibility, I think. It's a kludge maybe, but you could send the LFE input to MTRX through a separate mono monitor and add the LFE LPF in the EQ. Then wrap that output back to a MTRX input and send that to your ATMOS monitor. It would burn an EQ channel, and might need a jumper cable somewhere. If your using DANTE, I've found the Dante Controller can sometimes help with virtual wraps. For example, send the LFE mono monitor output to a Dante output channel out of the MTRX and use the Dante Controller to wrap it back to a Dante input channel to the MTRX that can then be routed to the ATMOS monitor LFE channel (without the LFE LPF) to sum it with the BM signal.