Confused by REW and LFE channel - help appreciated

Xyrric

Active Member
Hi All,

With a bit of time on my hands I've been tinkering with with my home cinema system and am confused by some of the results I've been getting with REW. Any light you can shine on things would be a great help.

A bit of background

The system is 7.1.4 in a large living room with a Lyngdorf MP-50 processor. The room is not rectangular but has a slight L shape to it in the right rear corner when sat facing the screen. Recently I came across this Home Theatre Gurus video



and was enthused enough to get out my UMIK-1 and have a go with REW. I'm definitely a novice with REW but the video is very good at patiently walking you through all the steps. Since then REW sweeps seem to have taken over my life.

Over the years I've acquired three different subwoofers and having watched the video twice decided to see if I could integrate them all. Now the recommendation by Lyngdorf if you have non-matching subs is to use the LFE out on the processor as the other option of using the aux channels for subs is recommended for when you have matching subs as front or rear pairs. So my signal chain for the subs is

MP-50 LFE out to input 1 on miniDSP and then outputs 1,2 and 3 sending signals to the subs.

The MP-50 does this thing where it resets all speaker distances to zero whenever you change the crossover. After getting fed up with entering speaker distances every time I wanted to test a different crossover I ended up leaving all the distances at zero while aligning the subs. Additionally the MP-50 was in BYPASS mode meaning no room correction was being applied (as far as I understand) and the voicing was NEUTRAL. Lyngdorf's voicings are equalisation settings that raise the bass or the treble etc depending on preference.

The miniDSP was set up so that all the xovers on outputs 1,2 and 3 were bypassed and there was no PEQ of any kind.

Measuring

All measuring of the subs was done without my front three speakers being on. They all have their own amps which weren't turned on. I have a separate 5 channel amp powering the side surrounds and rear and this was on to enable my right rear speaker to be used as a timing reference. The XOVER in the MP-50 was set at 100hz. All measurements were done with the volume on the MP-50 at -20. All measurements in REW for integrating the subs were done using channel 1.4 with 1.6 as the timing reference.

1) I gain matched the subs as per the video above. All subs were set with their own gain dials at 50% and their xovers dials at their highest settings or bypassed. The subs were gain matched by adding the required gain on the output channels of the miniDSP.

2) Then I spent a few nights moving the subs around the available free space in the room before ending up with the F113 in the front left corner, the CAP2400 in the rear left corner and the SPL1200R front centre right beside the centre speaker. Below are the three individual responses and the summed response.

1602169299878.png


As you can see there was some negative summation between 35-55hz and I played about with the delays on the three subs to try and improve this. It was a bit swings and roundabouts to be honest but I settled on settings that improved things a bit between 20-70hz while losing a bit between 70-115hz.

1602170211395.png


For greater clarity the difference between the two summed curves can be seen here

1602170477985.png


3) REW filters

Having achieved the pink curve above I generated filters in REW for export to the miniDSP as follows. (Pretty much copying from the home theatre gurus video above)

1602170891303.png


This was actually my second attempt as the first at a target level of 85dB gave me bass that was simply overpowering. So after all of the above with the coefficients imported from the saved file to the miniDSP the output of all three subs gives me the result below.

1602172543046.png


Okay so far nothing out of the unusual.

Having got this far I then decided to look at the response of each of the front three speakers and the LFE channel combined in REW by for example measuring channel 1.1 along with secondary output channel 1.4 when looking at the FL+subs combination. The result is below

1602173029050.png


Confused by the result I then looked at the FL, FR and C alone. That is measuring channels 1.1, 1.2 and 1.3 individually without the secondary channel. These results are added below.

1602173190275.png


Further confused I then took the same measurements of the front three again but this time with the sub outputs turned off in the miniDSP (even though the second input channel wasn't being used which I thought meant no signal should be going to the subs) and I got the following

1602173548837.png


So a bit confused really as to what these measurements are telling me. Obviously my knowledge of how the miniDSP and REW interact is lacking and i'm hoping some of you fine folks can put me straight.

Thanks in advance

X
 

Conrad

Moderator
I've read this a couple of times and I'm a little confused.

LFE is a discreet channel that REW can generate signals through. It's played back only by the subwoofer(s) (in this set-up). It'll run 10dB hotter than the measurement on any of the mains channels and it won't have a crossover in place so it'll roll off later than the sub alone on a mains channel.

Usual practice is pretty close to what you've done so you're definitely on the right track and it looks like you have potential to get a great response in room.

What I normally do is:

- check each sub placement in room, determine the best combination from the available places, if any, and set the subs there.
- time align the subs using LFE to get the best alignment. From this point on you treat all subs as a single sub. There are exceptions to this, but they're uncommon.
- if using an automated EQ process I run it here.
- EQ subs to flat/taste, a house curve is common which you have.
- check integration and adjust distances to integrate well with LCR

It sounds like you've found the best placement and you've time aligned the subs.

Your EQ appears to be very aggressive and you're giving up about 10dB of headroom there. I'd worry less about getting ruler flat and maintain some of that volume. Try setting the target to 85dB instead of 75.

You probably only need two filters as well, 38Hz and 68Hz. Tame those two bumps and you're good to go. Fewer filters tends to sound better, but try either/both.

Once you've got that, measure all subs and confirm the EQ is doing what you expect. Then measure each main with no subs and with a consistent timing reference. I use surround back left.

In REW you can then use the alignment tool to check the integration between each main and the subs, you can adjust the delay on the subs to get the smoothest integration through the crossover. It's a bit trial and error but it's easier than measuring, adjusting, measuring again. Just find some thing that works for the Subs+L, then have a look and see what it looks like for C and R. If it's rubbish have a look at what's good for C and R and you quickly get a sense for what works well. Apply those distances in the minidsp and measure subs+main for the front three and you should be done.
 

Xyrric

Active Member
Hi @Conrad.

Thank you very much for your reply. I have seen your contributions on other threads and have always been appreciative of your constructive and knowledgeable help and the time you give to other members on the forum.

I would like to respond to your suggestions one by one and with as much detail as possible because I worry I may have made a noob mistake that once you spot it would explain some of my concerns.

- check each sub placement in room, determine the best combination from the available places, if any, and set the subs there.
My wife is a good'un bless her and has let me put three subs wherever I want. The only restrictions have been some immovable bits of decor so I am very happy with the final placements of each sub.
- time align the subs using LFE to get the best alignment. From this point on you treat all subs as a single sub. There are exceptions to this, but they're uncommon.
I have done this as best I can but really more by luck than judgement. In all honestly I don't understand what I should be aiming for with the time alignment tool. I first looked at getting the F113 and the CAP2400 aligned and to my untutored eye adding a delay of 2ms to the F113 looked about right. Then I looked at the combined response of these two to the SPL1200R and judged the combined response of the new virtual sub needed to be delayed 21.5ms. To achieve this I ended up with the following final delays. F113 = 23.5, CAP2400 = 21.5, SPL1200R = 0. If you would be so kind I'm attaching the raw sub data for you to look at in the hope that you can confirm my assumptions or point me in a better direction.

- if using an automated EQ process I run it here.
I was wondering about this. Would there be any benefit of giving the room correction in the Lyngdorf (RoomPerfect) the final corrected sub response (achieved by generating filters in REW and then importing those into the miniDSP to give a very flat response) as a starting point for it to apply its own filters?

- EQ subs to flat/taste, a house curve is common which you have.
When to do this obviously depends on the route taken above.
- check integration and adjust distances to integrate well with LCR
How do I do this? In the first instance REW sweeps with a timing reference give relative delays. I always use my right rear speaker as this is my nearest speaker to the microphone. This is the only distance I physically measure and then time align all the other ground level speakers using the delays suggested by the REW sweeps. But how do I integrate the new "virtual" sub to the mains?

Your EQ appears to be very aggressive and you're giving up about 10dB of headroom there. I'd worry less about getting ruler flat and maintain some of that volume. Try setting the target to 85dB instead of 75.
Okay so you mean don't add the 10dB house curve from 100hz to 30hz but just EQ it flat like this
1602188012798.png


You probably only need two filters as well, 38Hz and 68Hz. Tame those two bumps and you're good to go. Fewer filters tends to sound better, but try either/both.

Sorry for being dense but I don't follow you. Do you mean add these filters in the miniDSP PEQ?

I am also attaching the raw data for my front three speakers for you to look at. My real confusion comes from the following. Why do the measurements of the FL, FR and C look so different whether the subs are on or off in the miniDSP? For example when I use channel 1.1 on REW nothing is going to the subs so why am I getting the totally different curves?

Sorry for another long post. I just want to make sure I understand all the minutiae.

All the best,

X
 

Attachments

  • 7 raw measurements.zip
    9.4 MB · Views: 51
  • FL FR C measurements.zip
    7 MB · Views: 46
  • Fronts with subs combined.zip
    3.5 MB · Views: 45

Conrad

Moderator
Thank you very much for your reply. I have seen your contributions on other threads and have always been appreciative of your constructive and knowledgeable help and the time you give to other members on the forum.

Not a problem, I might add that to my linkedin :)

- check each sub placement in room, determine the best combination from the available places, if any, and set the subs there.
My wife is a good'un bless her and has let me put three subs wherever I want. The only restrictions have been some immovable bits of decor so I am very happy with the final placements of each sub.

That's great, it means you're starting with the best chance of success out of the gate.

- time align the subs using LFE to get the best alignment. From this point on you treat all subs as a single sub. There are exceptions to this, but they're uncommon.
I have done this as best I can but really more by luck than judgement. In all honestly I don't understand what I should be aiming for with the time alignment tool. I first looked at getting the F113 and the CAP2400 aligned and to my untutored eye adding a delay of 2ms to the F113 looked about right. Then I looked at the combined response of these two to the SPL1200R and judged the combined response of the new virtual sub needed to be delayed 21.5ms. To achieve this I ended up with the following final delays. F113 = 23.5, CAP2400 = 21.5, SPL1200R = 0. If you would be so kind I'm attaching the raw sub data for you to look at in the hope that you can confirm my assumptions or point me in a better direction.

With time alignment you're looking for the fewest nulls, or nulls in the best places (way above crossover, for example). In a perfect world you'll have a combination of responses that constructively sum to fill dips and destructively sum to flatten peaks. In reality that's rarely the case so we look to fill the nulls as they can't be EQd and peaks can.

I'll take a look at the MDAT files in the morning but it looks like you're getting a really good combined response already so unless there's something very strange going on I don't see any need to change. As long as it sounds good too that is.

- if using an automated EQ process I run it here.
I was wondering about this. Would there be any benefit of giving the room correction in the Lyngdorf (RoomPerfect) the final corrected sub response (achieved by generating filters in REW and then importing those into the miniDSP to give a very flat response) as a starting point for it to apply its own filters?

In my experience, EQing before auto EQ causes issues. If you apply EQ and then have auto EQ process that response, if you don't like the outcome it's then very difficult to change things. You'll have to take the response and then adjust your already applied EQ to make the compensation you want to. You're compounding EQs and that can sometimes sound bad.

I've found that it's easier to give auto EQ the best response possible using just time alignment, show the processor one sub and have it apply it's EQ to that. Then you have the minidsp to apply adjustments to try and improve either the response or the integration. That said, this is usually applied to EQ systems that have less control than RP. I use it with Audyssey XT32 and I really a house curve because XT32 cuts my subs flat and there's nothing I can do about that, except for correct after the fact with a low shelf.

I'll be honest, I'm quite excited about this, I've never had the chance to play with RP measurements. Most people that buy Lyngdorf stuff don't want to sit and tweak after the fact, or it's so good that they don't need to. I'm considering an MP60 as my next processor so this is very interesting to me.

- EQ subs to flat/taste, a house curve is common which you have.
When to do this obviously depends on the route taken above.

- check integration and adjust distances to integrate well with LCR
How do I do this? In the first instance REW sweeps with a timing reference give relative delays. I always use my right rear speaker as this is my nearest speaker to the microphone. This is the only distance I physically measure and then time align all the other ground level speakers using the delays suggested by the REW sweeps. But how do I integrate the new "virtual" sub to the mains?

I imagine that your LCR are already time aligned. If they're not, do this first. You can do this with the timing reference you mention, the front right. It doesn't matter which speaker it is as long as you don't change it, and don't change the distance or level of that speaker. Get them all aligned, then take the following measurements:

- all subs on a mains channel, no main, timing reference
- left speaker, no sub, timing reference
- center speaker, no sub, timing reference
- right speaker, no sub, timing reference

Once you have those, open up the time alignment window. I select the sub as one measurement and I only move the sub. Choose a main in other dropdown. Drag the sub's delay slider and get a sense for how the phase loops, you'll see the nulls move across the measurement and then repeat. You'll see that as you move the delay you go through a few points that look ok. Ideally you want small delays so start with a delay that's close to zero. There's no hard and fast rule so if you've got one that's ok at 2ms and one that's great but it's 10ms, it's up to you. Great and 50ms is probably too much though.

Once you find a decent delay, switch the speaker to another main but leave the sub. You'll now see that speaker's integration with the sub with that specific delay. If it's bad then move it and get a sense of where the good ranges lie. With any luck you'll find a good response that lines up with a good response of the other mains. It's really trial and error. You can probably look at the phase and work it out, but it takes as long to use the sliders as it does for me to look up the calculations.

Your EQ appears to be very aggressive and you're giving up about 10dB of headroom there. I'd worry less about getting ruler flat and maintain some of that volume. Try setting the target to 85dB instead of 75.
Okay so you mean don't add the 10dB house curve from 100hz to 30hz but just EQ it flat like this
View attachment 1378388

No, a house curve is fine. You're targetting 75dB but the response has a 28Hz peak at 105dB. That means you're cutting 30dB from your response. If you click the EQ filters button you'll see that the cuts are huge. Move the target closer to the response. There's a Calculate Level link, click that and it'll move the target for you. You want it to be a lot closer to the initial response. I would use the slope you used before and adjust the level so the target hits 90dB at 20Hz.

REW will EQ down (and up) to where the response crosses the target for the last time. If you set the target line to hit 100dB on the image above it'll EQ down to about 26Hz as that's where the response the crosses the target and never comes back above it. You want to EQ all way down to 20Hz so adjust the target line accordingly.

REW shouldn't generate a filter of more than about -6dB with that response.

You probably only need two filters as well, 38Hz and 68Hz. Tame those two bumps and you're good to go. Fewer filters tends to sound better, but try either/both.
Sorry for being dense but I don't follow you. Do you mean add these filters in the miniDSP PEQ?

What I mean is that, with that response and a properly set target REW should only generate you a couple of filters. It's only use 6 now which isn't too many, but I think you can get a flat line in fewer.

I am also attaching the raw data for my front three speakers for you to look at. My real confusion comes from the following. Why do the measurements of the FL, FR and C look so different whether the subs are on or off in the miniDSP? For example when I use channel 1.1 on REW nothing is going to the subs so why am I getting the totally different curves?

Sorry for another long post. I just want to make sure I understand all the minutiae.

All the best,

X

I'll take a look at the files when I get a chance tomorrow.

A question for you though, what's channel 1.1? How are connecting to the processor? If it's over HDMI using FlexAsio (which is recommended) then the channels should all light up and be properly named.
 

Conrad

Moderator
I couldn't resist, I had a quick look.
For differences with the subs on and off, do you mean this?

comparison.jpg

If so, that's an expected response. Red is with the subs on and so has output <100Hz, orange has the crossover in place and so it doesn't. Above the crossover they're the same.
 

Xyrric

Active Member
Morning Conrad,

Just to be belt and braces about this so that I don't make a fundamental set up error that means a whole days worth of REW measurement aren't quite what they should be, let me fill you in on the way I have things set up.

Laptop connected by HDMI to MP-50. Just in case it makes any kind of difference I will mention that my HP laptop doesn't have a HDMI in but a displayport in so at the laptop end of the HDMI cable I have the HDMI to displayport adaptor below.

Amazon product

In setting up REW I followed Austin Jerry's guide so have ASIO4ALL installed.
1602225752345.png


Are there any benefits to changing to flexasio? I am a little leery about changing because it took me ages to get the ASIO side of things to play nice when I first set things up and at least at present it's working.

You asked about channel 1.1. When I take a measurement the following screen pops up

1602226116584.png


On the right you can see
Output - above displaying 1.3 (This is the Centre speaker)
Secondary output - greyed out above
Ref Output - above displaying 1.6 (This is the right rear)

When I click on the Output dropdown I get all the following options

1602226393664.png


The display audio outputs listed are as follows
miniDSP 2x4HD 1
miniDSP 2x4HD 2
Display audio output 1.1 - Front Left Speaker
Display audio output 1.2 - Front Right Speaker
Display audio output 1.3 - Centre Speaker
Display audio output 1.4 - Subs
Display audio output 1.5 - Rear Left Speaker
Display audio output 1.6 - Rear Right Speaker
Display audio output 1.7 - Surround Left Speaker
Display audio output 1.8 - Surround Right Speaker

So this goes back to my confusion about the results I mentioned getting previously. The four measurements below were taken in the following manner.

Red - This was output 1.4 and at this point in time none of the front three speakers were on. This measurement was for me to verify the combined response of the three subs after they had been gain matched, aligned and had the REW generated filters imported into the miniDSP.

Brown - This was output 1.1 plus the secondary output active and set to channel 1.4 allowing me to measure the front left and the subs from the red measurement above. I was expecting the addition of the front left into the measurement to add more difference to the end result below 100Hz but the curve we get pretty much mirrors the red one but at higher SPL

Blue - This was as above but only with output 1.1 and the secondary output channel tick box unchecked so that the secondary channel was greyed out. To my mind this was me measuring the response of just the front left. But something isn't right because the sweep is a mirror of the brown one.

Pink - All settings in REW are as for the blue sweep but this time I went into the miniDSP and in the routing turned all the subs off. This result is what I was expecting to see when I did the blue sweep. So what exactly is going on with the blue measurement?


1602229247222.png


As for your other points just a few questions

- all subs on a mains channel, no main, timing reference When you say all subs do you wish me to take measurements of each sub individually or of the combined virtual sub that we have created by time aligning the three individual subs. Taking the measurements of the subs on a mains channel (for example front left) eliminates the potential problem with the brown/blue/pink sweeps above but it brings the XOVERs into play. This is another point where I become confused. As mentioned in my original post, for my setup Lyngdorf recommend that the subs be connected from the LFE out. The LFE out has no XOVER as I understand it so does it matter at what frequency I set the XOVER in the Lyngdorf other than perhaps to have it as high as possible? Alternatively would it help to take each of these measurements at different XOVERS say 60Hz, 70Hz, 80Hz, 90Hz, 100Hz, 110Hz and 120Hz? (Did you know that in the Lyngdorf you can set a custom XOVER as high as you want. If you have their own subs even as high as 800Hz). Also I have noted your point about EQing an already EQ'd curve so I'm assuming that all these sub measurements should be with the gain matched, time aligned settings but WITHOUT the REW EQ imported into the miniDSP. (So many things to remember - I'm going to need an airline pilot check list before I take each measurement:))
- left speaker, no sub, timing reference
- center speaker, no sub, timing reference
- right speaker, no sub, timing reference


There's a Calculate Level link, click that and it'll move the target for you.

Further to the discussion above I won't actually need to do this will I because we don't want to EQ the subs before the Lyngdorf runs RoomPerfect?

As a point of interest when I click on "Calculate target level from response" it gives me a target level all the way down at 62.1dB

1602231143880.png


Thanks once again for your feedback

X
 

Attachments

  • 1602227151109.png
    1602227151109.png
    155 KB · Views: 107

Conrad

Moderator
Ok, I think I see what's up - it's that second output.

Displayport showing 1.n is fine, that's the standard channel numbering it's just an unusual way of referencing it. That's probably due to the limited data that displayport sends and receives. If ASIO4All is working for you then there's no need to move away. FlexAsio is much easier to set up and it's much more stable though. If it starts giving you any issues I suggest you switch.

I don't use the second output, just make sure you've got the correct speakers connected/amps turned on.

So for subs only, turn off the power amps to L and/or R and measure on the L or R channel. That'll give you subs only, with the crossover engaged.
For mains only, measure LCR and mute the subs in the minidsp as you are doing.
Then you can integrate those measurements.

What you're measuring is the combination of LFE and mains, which means your sub response isn't crossed over and is 10dB higher.

When you measure 1.1 the signal is sent from REW out to the processor from, say, 10Hz up. The processor gets that and says, ok, 10Hz, send that to the subs. Anything below the crossover is sent to the subs (but it's not a wall, so there's a slope either side of 80Hz as the subs roll off and the mains kick in).

You can use 1.4 to measure the integration between the subs, but for integration with the mains use 1.1, 1.2, and 1.3 and turn the subs off. You need to take at least one measurement on 1.1, 1.2, or 1.3 without the main which will give you your crossed over sub response.

Here's mine as an example:
1602261615424.png

Orange is my subs (all of them playing together) output on the left channel, with the amp powering the main turned off.
Blue is the left main on the left channel with the subs muted in the minidsp.
Green is the integration between the two, the left channel with amp on and subs unmuted.

Does that make sense?

As for the target level, that's odd. Have you got a calibration file loaded for your mic?
When you did the EQ with the house curve in the initial post what were your filters?

I have had it where REW gets a bit confused and sets the target low. If your filters are large (-10 or above) then try moving the target line manually.
 

Xyrric

Active Member
Hi Conrad,

Okay that is good to know about the 1.4 channel.

Did you have a chance to look at the mdat of the way I aligned the three subs to create the overall sub response? Would you suggest any improvements before I do some measurements this weekend?

In terms of measurements I will do the following. Please correct me if I am making any false assumptions.

The Lyngdorf will be set to bypass (no room correction) and neutral (no eq)
The miniDSP - no PEQ on the sub outputs just the gains and delays used to align the subs.

Then measurements
- all subs on a mains channel, no main, timing reference. BY THIS I'M ASSUMING YOU MEAN FOR ME TO MEASURE THE RESPONSE OF THE COMBINED SUB AND NOT THE INDIVIDUAL SUBS
- left speaker, no sub, timing reference - UNDERSTOOD
- center speaker, no sub, timing reference - UNDERSTOOD
- right speaker, no sub, timing reference - UNDERSTOOD

The above are thus four measurements and I'm happy to take them in sets of four with Xovers in the MP-50 of 60,70,80,90,100,110 and 120Hz. At each xover frequency I will enter the distances for the front left, front right and centre speaker before measuring but i'm still not clear on what distance I should enter for the sub. Should I just leave it at zero?

As for the calibration file it is entered as below in the mic cal files slots. I'm assuming I don't need to do anything in the sound card boxes?
1602273227364.png
 

Conrad

Moderator
I didn't have a chance to look properly, work's been crazy this week. Looking at the combined response though I think you've done a good job, there's no nulls and a great response even before EQ. Your "pre eq" response is better than a lot of people's "after eq" response.

If you want, while you've got everything out and the EQ off, measure each sub on a mains channel with no main and include that in your next MDAT.

So, the measurements I'd suggest are (always with the same timing reference):

each sub, no main, on the mains channel
then combine them as you have and measure the combined output (all three subs) on the mains channel with no main
I'd measure each main too, just for reference.

Then apply RP. If you're going to run with it in the long term then you might as well set it how you like it.

Then re-measure the combined subs on a main with no main
And measure each main
Throw in each main full range as well, so all subs and L, all subs and R, all subs and C as three measurements. They'll be the combination of the subs measurement and the mains measurements, but it's good to have to confirm.

With that we'll be able to see if we can improve on the response and integration over and above room perfect. I've never seen what RP does to a response so this is very interesting for me.

One of the things I understand about RP is that it "removes the room" from the response. I guess this is done by looking for common traits in the response from different position and then removing anything that's not common as a room artefact, leaving the native response of the speaker. This means we'll probably need to be quite careful not to over EQ and ruin that processing. In your case less is definitely going to be more. In fact I would only EQ through the minidsp below the crossover.

I also believe that's it supposed to be quite good with integrating multiple subs. In that case, if you don't have great integration with this round we might present two subs, or more, and see what it can do when it has complete control over each one.

For crossovers, stick with large jumps. 60, 80, 100, 120. Only if it looks like you need to is it worth doing 10Hz increments. The differences are very small anyway. Given you're taking a lot of measurements you might even want to just stick with 80 and 100.

For some reason your images come through quite small and I can't enlarge them. If you've loaded the 90 degree cal file for your mic, that should be all you have to do. It's fine if it's a little bit out, we can work with that. We always take real measurements to confirm anyway.
 

Xyrric

Active Member
Hi @Conrad.

Sorry for the lack of communication. We had a bit of a family emergency and I've been away for two weeks. All is good now and I should find some time from the middle of this week.
 

Conrad

Moderator
No worries. Family comes first, especially given the current situation. Glad to hear things are better, I’m around whenever you’re ready.
 

Xyrric

Active Member
Hi @Conrad.

So I had a little time yesterday to take some measurements. First of all I wanted to double check that I had got my subs combining right. Attached are measurements of the subs on the LFE channel with a detailed description of all the systems settings in the description panel of each measurement. The starting point for all measurements was:

Roomperfect in BYPASS
Neutral Voicing
Postprocessing NONE
XOVER at 100Hz
All delays on the ouput channels of the miniDSP at zero.
The measurements are made in the LFE Channel.
The Front Right is used as the timing reference.
MP50 volume 17.2

Looking at these measurements and particularly the contribution of the SPL1200R I decided that that little sub was definitely trying to punch above its weight and not really contributing anything, so I have removed it leaving the JL F113 and the CAP 2400.
 

Attachments

  • 28 Oct subs on LFE channel single measurements.zip
    8.3 MB · Views: 39

Xyrric

Active Member
Carrying on I experimented with improving the combined response. The attached file shows the best results and these were achieved by delaying the CAP2400 and inverting it.

I eventually settled on a delay of 6.5 on the CAP2400 and inverting it. This gave a substantial improvement between 30-50Hz at the expense of some loss below 18Hz.
 

Attachments

  • 28 Oct CAP2400 INVERTED .zip
    7.1 MB · Views: 39

Xyrric

Active Member
One thing that RP doesn't do is time align multiple subwoofers so that was the next thing to do. To achieve this I took the following measurements with the CAP2400 delayed and inverted as above. All measurements use the FR as the timing reference.

Measurement 1 is signal sent to FL but FL disconnected, so measuring the subs on the LFE channel only.
Measurement 2 is signal sent to FL with the FL connected and the LFE disconnected, so measuring FL only.
Measurement 3 is as measurement 2 above but only measuring from 0-200Hz to focus on bass frequencies.
Measurement 4 is signal sent to FL with the FL connected and the LFE also connected, so measuring the combined response of FL and LFE all together
Measurement 5 is as 4 above but only from 0-200Hz.


Looking at the results we see the following
Measurement 1 the subs only this is showing a delay relative to the timing ref of 61mm.
Measurement 3 FL only ,no subs 0-200Hz, is showing a delay relative to the timing ref of -805mm

So relative to the acoustic timing ref I the LFE should be delayed by a 61 - (-805) = 866mm

The current distance of the Sub LFE in the MP50 is 750cm so this should be increased to 836cm.

I always get confused with how adding delay/increasing distances works and worry that i'll get it wrong so I ran all the above measurements past a very helpful fellow enthusiast who can be found on these forums and particularly on the Lyngdorf threads. As always he was very supportive and full of good advice. He preferred adding just 34cm to the sub distance when considering the balance between phase and timing. He also recommended reducing the level of the LFE output by 10dB and then running RP. I am more than happy to follow his lead. So doing the above we end up with the following as our STARTING point before we run RP.

1603979992779.png
 

Attachments

  • 28 Oct Time aligning subs.zip
    3.7 MB · Views: 44

Conrad

Moderator
It looks like you've done a great job, is my overall impression.

A few thoughts:

- You might want to try running the velodyne as a VNF sub. If it's right behind or next to you you can EQ it quite aggressively as the gain is low. You can probably get it ruler flat and just get the impact from having it VNF.

- If you're measuring the subs on the LFE channel, isn't that 10dB louder? I know my processors add 10dB when they're given an LFE signal. I like to EQ on the LFE channel to get the response flat and then check integration with the sub(s) measured on a mains channel, with that main off. If you're genuinely measuring on the LFE channel and you've removed 10dB you might find things a bit anaemic. But I'm not a Lyngdorf user so I don't know for sure. It's certainly how it works for Classe, Marantz, Sony, Trinnov, etc.

- I wouldn't bother time aligning with the mains before running RP. I would only time align the subs with each other and then present them as a single sub to RP. That way RP can do what it does and, if you don't like the result, you can alter it more easily. Granted, that's easier to change after auto EQ than actually EQing the subs before auto EQ, but still. If we think that RP can do a great job with a single sub, then give it that chance.

An example:

You add 5ms to one sub to get good time alignment between the two. You then add 2ms to the subs to align with the mains. Then you run RP. RP might not do anything, or it might, we don't know.

Either way, if you want to affect the outcome you might need to add another 2ms on to get back to where you were. Now you're in a position where you're adding 2ms and RP is removing them and you're adding them back on. If you didn't adjust initially RP wouldn't have done anything and you'd add 2ms. Same net result but you're doing less processing and, I think, overall your speakers would be more closely aligned.

It's a bit of a contrived example that might never happen, but it might.

My principle is to do as little processing that any auto EQ can undo as possible before running auto EQ. Time align the subs, sure, but any EQ and integration should be done after auto eq.
 

Xyrric

Active Member
Hi @Conrad.

Thank you for your considered thoughts.

1) VNF sub.
Great idea! This is something I've often wondered about down the years. However with the present layout of the room there just isn't the room as most of the amplifiers are at the back of the room behind the sofa. Maybe in the future if things permit. As an alternative i'm working on a BOSS platform under the sofa at the moment.

2) LFE Channel 10dB boost
Noted. This is a valid question and worth checking with a few measurements before running RP. However how I have reached this point has been an ongoing process.
It started when my other half wanted to move the furniture in the room necessitating relocating front speakers and subs. This was good from my point of view as I could bring my FL and FR away from the side walls and give them some room to breathe. I experimented with the newly available positions for the subs and the best of those was with the F113 corner loaded in the front left corner of the room and the CAP2400 against the rear wall facing the left of the room and 35cm from the left wall.
One thing that became immediately apparent though in the new layout with both subs effectively corner loaded was how overpowering the bass was in a very bad way. Watching movies and particularly listening to music the bass was way out of balance. The MP-50 has a "set bass level" feature if you plug in it's supplied microphone. This is used before you run RP and the MP-50 indicates if the sub LFE level is too high to run RP optimally. Doing this I had to reduce the sub level by -24dB on input 1 on the miniDSP.
The measurements I sent to you were all done with the volume raised back to -7dB so that the measurement sweeps for each sub measured individually were peaking at approx 95dB and the combined response was hitting 100dB at 30Hz.
Now reducing the volume from this point by a further 10dB will lower the volume on input 1 of the miniDSP to -17db. Later on I will try and run RP if I have the time and I think RP will find this volume too high and recommend I lower it to around -24 again. We'll see.

Help me clear up something about the subs and LFE.
The subwoofers are playing back two elements as I understand things.
1) The re-directed bass
2) The .1 LFE channel which is a distinct channel on a bluray in exactly the same way that the FL is a distinct channel. The special point of interest with the .1 LFE channel is that it is mastered 10dB hot compared to the other regular channels.

Is my understanding correct? If it is then if RP does it's thing and blends the subs to the other speakers with regard to the re-directed bass won't the LFE channel still be +10dB?

3)Time aligning

This has been thoroughly discussed over on the Lyngdorf thread. One thing that RP doesn't do is time align the subs to the fronts. Hence why the recommended procedure on the Lyngdorf thread is to time align subs and speakers before running RP. In the discussion below Rob Sinden talks about this from the point of view of a single sub. With REW and multiple subs we obviously have other ways of achieving this. By the way Rob starts his reply with "...Ron here" which is a running joke on the thread :)
………..Ron here

If RoomPerfect could measure the timing of speakers and subs it would be a great addition, but it looks like the feedback Flemming has give explains why they don’t do it. Its just not accurate enough….

Retailers and distributors have different areas of expertise and I have seen lots of things written about RoomPerfect etc. that are clearly a result of consumers being told the wrong thing or not having sufficient information.

That’s what myself and others users here are trying to help with. Provided you follow the measurement process correctly with RoomPerfect, I find results incredibly consistent. Adding a sub into any system is one of the biggest improvements available and the way RoomPerfect integrates the speakers and subs is brilliant, however the timing issues is something you have to understand and do correctly or the end result won’t be right.

Every sub with an internal amp I know of will add a delay the signal. So, for example if you sub measure 352cm from you, you may need to add 120cm to this distance so the output from the sub arrives at the same time as the sound from your speakers. So you need this info from you sub manufacturer. Bear in mind when you place a sub against a wall, its acoustic footprint is where the sub sits against the wall…so you would need to input the distance to your sub as 382 (as this is the distance to the wall behind your sub), plus 120cm for the delay the plate amp and its processing makes.

If anyone has tried this and isn’t happy with the results or would like more info, please post here and we and try and fill in the blanks.....
 

Conrad

Moderator
If you take two subs that are EQd for a specific location and then move them (especially from not in a corner to in a corner) then you're changing the relationship between the subs, room, and listening position so it's not surprising they sounded bad. There's a chance you could improve things, but the odds are that it'll make things worse, which is what you found. It's recoverable though, you just re-eq.

The leveling function is a common function between auto eq systems. Basically they all want all speakers to be within a certain range so that they can effectively manage the relative volumes. If they only have +-10dB adjustments, having a sub that's 24dB hot means that they might not be able to bring it in line with the other speakers. It makes sense, but it's also common to boost the sub after the fact.

As for the LFE, what you've described isn't my understanding. I believe that the track is mastered in a way that expects the processor to add 10dB to the signal on the LFE channel. You can test this by playing the same tone at, say, 30Hz from REW into channel 1 and then into channel 4. Channel 4 will be 10dB higher without any other change. This means that the LFE channel is having 10dB added to it by the processor.

If you level match multiple subs then it makes no difference which channel you use, as long as you use the same channel for both (or at least, LFE for both or not LFE for both). It's when you level match between subs and mains that you need to be careful. For that you need to use not LFE for both. If you use LFE the redirected bass will be 10dB light.

I didn't realise RP didn't time align the mains and subs at all. When you said "RP doesn't time align multiple subs" I thought that was the limitation. If it doesn't affect the timing between mains and subs then sure, feel free to adjust the timing to get the best integration before running RP.
 

Xyrric

Active Member
Hi @Conrad.

I'm trying to learn as I go along here so will shortly do the 30Hz test you have suggested in channel 1 and in channel 4.

Looking at some dolby information I came across the following


It states

"The signal in the LFE channel is calibrated during soundtrack production (my emphasis) to be able to contribute 10 dB higher SPL than the same bass signal from any one of the screen (front) channels."

Does this not mean that the 10dB boost is done on the disc independent of the AVR? Things are confusing because other documents state exactly the opposite and that the 10dB boost is handled in the AVR
 

Conrad

Moderator
I think that holds up and is consistent with my understanding.

A sound editor knows that the LFE track will be boosted by 10dB when played through a dolby processor. So if they want the signal to be played at the same volume as the mains, they'll master it 10dB lower. What that affords them though is an additional 10dB of headroom. So when they want a big boom they push the level to 0 and, relative to the mains, that's now +10 at playback time.
 

Xyrric

Active Member
Ok @conrad now I am just confused.

Attached are four measurements

Measurement 1 is the last measurement from 2 days ago. This is measuring channel 1.1 but with the FL disconnected from the MP-50 and so the response of the subs on the LFE channel is being measured.

Measurement 2 is the same measurement today and agreeably they match

Measurement 3 is exactly above but measuring channel 1.4 instead of 1.1

Measurement 4 is done by disconnecting the subs from the LFE out and connecting them to the FL.

As for the tone sweeps maybe I did something wrong. The methodology was to open the generator and the SPL meter. In the generator I selected a sine tone, the level was -12.00dBFSand the display output at the bottom of the generator was set to 1.1. I played the tone and took the reading that the SPL meter generated. I then repeated the above but with output 1.4 on the generator.

Please see below for an example
1604075681459.png


I took a number of measurements and the results are as below

Tone Frequency Hz 100 90 80 70 60 50 40 30 20
Channel 1.1 83.8 86.2 88.0 86.2 84.2 88.1 90.1 94.1 81.8
Channel 1.4 89.3 90.4 91.1 88.5 85.8 89.1 91.1 95.5 82.5

I could do with some enlightenment as i'm not sure what we have learned
 

Attachments

  • 30 Oct LFE test.zip
    4.8 MB · Views: 37

Xyrric

Active Member
Sorry but the table doesn't show up the way I formatted it

Tone Frequency Hz 100_____ 90_____ 80_____ 70_____ 60_____ 50_____ 40_____ 30_____ 20
Channel 1.1________ 83.8___ 86.2___ 88.0____ 86.2___ 84.2___ 88.1____ 90.1___ 94.1___ 81.8
Channel 1.4________ 89.3___ 90.4___ 91.1____ 88.5___ 85.8___ 89.1____ 91.1___ 95.5___ 82.5
 

Conrad

Moderator
I don't understand why you would disconnect and reconnect the subs.
The processor has an LFE out, but I think that's slightly misleading, that's a sub out.

that output should play the redirected bass from the main AND the LFE channel when present. And the LFE channel should have 10dB added to it.

If you connect one sub to that connection and play a test tone from REW on channel 1 and then the same test tone on channel 4, you should get a 10dB difference.

Maybe this is a question for the Lyngdorf thread? I don't have a Lyngdorf processor so it might be doing something non-standard.

When you play the 100Hz tones, where are the tones coming from? And is that with the main channel connected or disconnected?

Here's mine as a reference. Same REW sweep settings, but one is on the L channel and one is the LFE. Exactly 10dB difference until the crossover kicks in (which doesn't affect the LFE)

main vs lfe.jpg
 

Xyrric

Active Member
Thanks @Conrad. with your patience in walking me through this, its appreciated.

You said above
"If you connect one sub to that connection and play a test tone from REW on channel 1 and then the same test tone on channel 4, you should get a 10dB difference."

I'm pretty sure that is what I've done with measurements 2 and 3 above.
Measurement 2 is front left disconnected and the combined subs connected in the LFE out on the Lyngdorf. In REW the sweep is sent through channel 1.1

Measurement 3 is exactly as above but the sweep is sent through channel 1.4 in REW.

I have just measured them again and get exactly the same results as previously.
 

The latest video from AVForums

Are the TCL MiniLED TVs better than OLED? TCL Interview with Marek Maciejewski | AVForums Podcast
Subscribe to our YouTube channel

Latest News

iFi Audio launches NEO Stream music streamer with DAC
  • By Ian Collen
  • Published
Musical Fidelity launches M6x 250.7 multi-channel power amplifier
  • By Ian Collen
  • Published
Is Google set to rival Dolby Vision and Atmos?
  • By Ian Collen
  • Published
'Tizen TV by Samsung' brands coming to UK?
  • By Ian Collen
  • Published
Denon announces five new 8K AV receivers
  • By Ian Collen
  • Published

Full fat HDMI teeshirts

Support AVForums with Patreon

Top Bottom