Dirac Live via. MiniDSP DDRC-24 - Question about 2.1 setup

[Deleted member 48726]

Hello there, fellow scientists

I've been experimenting with miniDSP 2x4 HD in combination with REW and RePhase. 100 h. in I still didn't find the SQ and effect that I find satisfactory. So I bought the Dirac Live upgrade for the minidsp. And I must say that I'm initially very positive with the results. A handful measurements and 12 test profiles in, I have the initial frequency response that I like. -A magnitude better than I've achieved with REW and RePhase. It sounds better than ever.

But I have a question. -Right now I have my single SVS sub connected to the two RCA out (3+4) on the DDRC-24. And I wonder if the mixed phase corrections Dirac makes for L+R (1+2) will interfere with the signals to the sub and smear the outcome? Is it perhaps better to only use one RCA out to the sub or does it sum the signal identically either way? You can see the routing matrix below.

 

[sejarzo]

I had the same question, as I have yet to implement my DDRC-24 with a Hsu sub with both L/R inputs. Unsure if this is in the online manual for the DDRC-24, but this section in their online application advice is to configure it as a single sub. Apparently Dirac Live recognizes this to be a single sub and then develops the optimal correction for a 2.1 configuration.

https://www.minidsp.com/applications/subwoofer-tuning/sub-integration-dirac

 

[Deleted member 48726]

I realised that if the crossover and delay is done correctly, it doesn't matter if Dirac knows if there is a sub or not, as the signal below xover point only (mostly) goes to the sub. I have tried setting up as steep a filter as possible to avoid too much overlap. delay is measured with a single point dirac measurement main R in out 1, sub in out 2. Export to minidsp, note delay to main R. Measurement with main L in out 1 sub kn out 2, note delay for main L. Then setup minidsp with the delays, setup of crossover, then do the dirac corrections.

 

[mdsimon2]

In general bass frequencies are mono so it should not matter if you sum L+R to the sub. It is important to note that if L + R bass are identical you are effectively raising the level being sent to the sub by +6 dB which may mean you will clip the DSP output at higher volume levels.

Where is your sub located? Is it equidistant between your left and right speakers? A 2.1 setup will always be a bit tricky as you cannot achieve phase alignment to both the right and left speakers if they are different distances to the sub. I haven't run a 2.1 setup in a long time but I think depending on sub location this is how I would do it.

1) Sub equidistant to L + R: sum L+ R to sub
2) Sub located next to L: only send L to sub
3) Sub located next to R: only send R to sub
4) Sub located away from both L and R: sum L + R to sub

I am highly skeptical that Dirac makes any meaningful phase corrections at bass frequencies given the limited FIR taps of the DDRC-24 but the above guidance might help you achieve better sub / main phase alignment prior to Dirac.

Michael

 

[sejarzo]

I realised that if the crossover and delay is done correctly, it doesn't matter if Dirac knows if there is a sub or not, as the signal below xover point only (mostly) goes to the sub. I have tried setting up as steep a filter as possible to avoid too much overlap. delay is measured with a single point dirac measurement main R in out 1, sub in out 2. Export to minidsp, note delay to main R. Measurement with main L in out 1 sub kn out 2, note delay for main L. Then setup minidsp with the delays, setup of crossover, then do the dirac corrections.

In my situation, the mains are exactly equidistant from the listening position and closer to the listening position than the sub. Therefore, the delay should be the same for both main channels.

 

[sejarzo]

In general bass frequencies are mono so it should not matter if you sum L+R to the sub. It is important to note that if L + R bass are identical you are effectively raising the level being sent to the sub by +6 dB which may mean you will clip the DSP output at higher volume levels.

...

That assumes the recording was recorded/mixed and mastered with equal levels of bass in both channels. In your example situation (2), if a recording were to have a bass instrument panned to the right channel, that would result in lower levels of content below the crossover frequency being fed to the subwoofer versus that same recording with that instrument panned to the left channel, wouldn't it?

 

[mdsimon2]

That assumes the recording was recorded/mixed and mastered with equal levels of bass in both channels. In your example situation (2), if a recording were to have a bass instrument panned to the right channel, that would result in lower levels of content below the crossover frequency being fed to the subwoofer versus that same recording with that instrument panned to the left channel, wouldn't it?

Sure, that is why I said in general bass frequencies are mono and one of the reasons I use stereo subwoofers. A 2.1 system will always be a compromise. When I've done it in the past I've used a sub centered between L + R with summed L + R to the sub but not everyone's room can accommodate that.

Michael

 

[Deleted member 48726]

In general bass frequencies are mono so it should not matter if you sum L+R to the sub. It is important to note that if L + R bass are identical you are effectively raising the level being sent to the sub by +6 dB which may mean you will clip the DSP output at higher volume levels.

Where is your sub located? Is it equidistant between your left and right speakers? A 2.1 setup will always be a bit tricky as you cannot achieve phase alignment to both the right and left speakers if they are different distances to the sub. I haven't run a 2.1 setup in a long time but I think depending on sub location this is how I would do it.

1) Sub equidistant to L + R: sum L+ R to sub
2) Sub located next to L: only send L to sub
3) Sub located next to R: only send R to sub
4) Sub located away from both L and R: sum L + R to sub

I am highly skeptical that Dirac makes any meaningful phase corrections at bass frequencies given the limited FIR taps of the DDRC-24 but the above guidance might help you achieve better sub / main phase alignment prior to Dirac.

Michael

Reg. phase and delay is shouldn't matter if the sub is placed far or close to the main speakers as long as its driver is approx. at the same parallel line to the main speakers.
This is because you and the mic is placed at the listening position so when you are measuring delay from e.g. L main and sub its the same as measuring R main and sub even if the sub is placed in the corner. Delay and phase as measured end heard is similar and thus can be corrected by Dirac just fine.

I would never just run a L or R to the sub.

When a Dirac profile is applied, I made it a good habit to run 0dB tones at various frequencies but particular at frequencies which Dirac may have boosted (without amps on) at full volume and adjust output and master attenuation in the minidsp so clipping is avoided.

 

[mdsimon2]

Reg. phase and delay is shouldn't matter if the sub is placed far or close to the main speakers as long as its driver is approx. at the same parallel line to the main speakers.
This is because you and the mic is placed at the listening position so when you are measuring delay from e.g. L main and sub its the same as measuring R main and sub even if the sub is placed in the corner. Delay and phase as measured end heard is similar and thus can be corrected by Dirac just fine.

Good point on the parallel line. Not sure what I was thinking yesterday but I agree that putting the sub on a parallel line with L and R will give the same relative path length difference between the sub / L and sub / R from the listening position. As a result you can apply an equal delay to L and R in order to correct for the relative path length difference to the sub. For example taking the drawing below applying ~8 ms delay to both L and R would correct for the relative path length difference to the sub.

Where I disagree is I don't think Dirac can do much to correct the path length difference as it does not have the ability to delay individual output channels. As a result this delay should be set before running Dirac. And again, the DDRC-24 has rather limited FIR taps so it won't be able to do any significant phase correction at bass frequencies.

When a Dirac profile is applied, I made it a good habit to run 0dB tones at various frequencies but particular at frequencies which Dirac may have boosted (without amps on) at full volume and adjust output and master attenuation in the minidsp so clipping is avoided.

You can also measure the exact correction applied by Dirac in REW using the DDRC-24. Just set the DDRC-24 as your playback and capture device in REW and run a frequency sweep with and without correction.

Michael

 

[MCH]

Good point on the parallel line. Not sure what I was thinking yesterday but I agree that putting the sub on a parallel line with L and R will give the same relative path length difference between the sub / L and sub / R from the listening position. As a result you can apply an equal delay to L and R in order to correct for the relative path length difference to the sub. For example taking the drawing below applying ~8 ms delay to both L and R would correct for the relative path length difference to the sub.

?
while the listening position is equidistant to L and R (as it normally is), the difference in distance with the sub from the LP is always the same by definition, no matter where the sub is, or am i missunderstanding something?

parallel line or collinear?

(honest question, sub placement and setup is of great interest to me)

 

[Deleted member 48726]

Good point on the parallel line. Not sure what I was thinking yesterday but I agree that putting the sub on a parallel line with L and R will give the same relative path length difference between the sub / L and sub / R from the listening position. As a result you can apply an equal delay to L and R in order to correct for the relative path length difference to the sub. For example taking the drawing below applying ~8 ms delay to both L and R would correct for the relative path length difference to the sub.

View attachment 260067

Where I disagree is I don't think Dirac can do much to correct the path length difference as it does not have the ability to delay individual output channels. As a result this delay should be set before running Dirac. And again, the DDRC-24 has rather limited FIR taps so it won't be able to do any significant phase correction at bass frequencies.



You can also measure the exact correction applied by Dirac in REW using the DDRC-24. Just set the DDRC-24 as your playback and capture device in REW and run a frequency sweep with and without correction.

Michael

Delay I measure as described further up in the thread. I use a single dirac measurement and a sub instead of one main and go back to minidsp plugin to set it.

 

[mdsimon2]

?
while the listening position is equidistant to L and R (as it normally is), the difference in distance with the sub from the LP is always the same by definition, no matter where the sub is, or am i missunderstanding something?

parallel line or collinear?

(honest question, sub placement and setup is of great interest to me)

I might be misunderstanding your question but I should start by saying I was wrong (again). There is no need for the sub to be on a parallel line with the mains. As you note with a centered LP you will always be able to correct the path length difference with a delay.

However, if you set the delay based on a centered LP and move off center now the delay is no longer valid as the path lengths from L and R are no longer equal (but this is true of any sub positioning).

Michael

 

[sejarzo]

As the primary question was configuration, I didn't copy/paste the next step from the app note page on the minidsp site after the channel settings.

EDIT for clarity: According to the app note, this step is to be completed in the DDRC-24 configuration before running Dirac Live correction, as Michael noted in #9 above.

I find the final paragraph a bit odd, in that nowhere else on said page does it define how a small error in delay would prevent "good integration" or how one might adjust the delay to achieve that.

 

[mdsimon2]

As the primary question was configuration, I didn't copy/paste the next step from the app note page on the minidsp site after the channel settings.

I find the final paragraph a bit odd, in that nowhere else on said page does it define how a small error in delay would prevent "good integration" or how one might adjust the delay to achieve that.

View attachment 260092

I think they are saying that if you set a delay solely based on path length difference you may still have issues if your sub for example has additional delay (many active subs have additional delay).

Delay directly impacts phase which impacts how your crossover will sum. Linkwitz quantified the effect of small delays on anechoic woofer / subwoofer response here -> https://www.linkwitzlab.com/frontiers_5.htm. Unfortunately as he notes actually measuring these differences in room is pretty much impossible as the room response will dominate.

Michael

 

[phoenixdogfan]

In general bass frequencies are mono so it should not matter if you sum L+R to the sub. It is important to note that if L + R bass are identical you are effectively raising the level being sent to the sub by +6 dB which may mean you will clip the DSP output at higher volume levels.

Where is your sub located? Is it equidistant between your left and right speakers? A 2.1 setup will always be a bit tricky as you cannot achieve phase alignment to both the right and left speakers if they are different distances to the sub. I haven't run a 2.1 setup in a long time but I think depending on sub location this is how I would do it.

1) Sub equidistant to L + R: sum L+ R to sub
2) Sub located next to L: only send L to sub
3) Sub located next to R: only send R to sub
4) Sub located away from both L and R: sum L + R to sub

I am highly skeptical that Dirac makes any meaningful phase corrections at bass frequencies given the limited FIR taps of the DDRC-24 but the above guidance might help you achieve better sub / main phase alignment prior to Dirac.

Michael

So how many FIR taps will be meaningful. Let's assume one sub and a crossover at 100 hz to the mains? Is there a guidline or rule of thumb?

 

[sejarzo]

Thanks for that explanation. I didn't think about active subs with additional controls.

And yeah, it makes complete sense that measuring those small differences would be effectively impossible due to reflections, etc.

 

[DWPress]

So how many FIR taps will be meaningful. Let's assume one sub and a crossover at 100 hz to the mains? Is there a guidline or rule of thumb?

It really depends on how much correction you need below 100Hz. If it's just a simple XO then very few taps but add in a few amplitude corrections with phase considerations and they add up quickly. When I was using my miniDSP 4x10 I always took my corrections into rePhase to test which will alert you if you don't have enough taps to resolve a filter. There's probably a better way to do it with math but this worked for me.

 

[mdsimon2]

So how many FIR taps will be meaningful. Let's assume one sub and a crossover at 100 hz to the mains? Is there a guidline or rule of thumb?

Probably best to download rephase and see what you can do with 4096 taps/ch at 48 kHz. I think you will find it a struggle to do much below 80 Hz, especially if you are dealing with steeper slopes.

You can also measure the output electrically and see how different the phase response is from minimum phase to get an idea of how much phase correction Dirac is actually doing. It's been a while since I've used Dirac or made these measurements but my memory is that there is very little phase correction.

Michael

 

[Deleted member 48726]

I think they are saying that if you set a delay solely based on path length difference you may still have issues if your sub for example has additional delay (many active subs have additional delay).

Delay directly impacts phase which impacts how your crossover will sum. Linkwitz quantified the effect of small delays on anechoic woofer / subwoofer response here -> https://www.linkwitzlab.com/frontiers_5.htm. Unfortunately as he notes actually measuring these differences in room is pretty much impossible as the room response will dominate.

Michael

Exactly. There is DSP built in into active subwoofers which always has processing delay. You can verify that by measuring delay and calculate which distance the delay corresponds to. It is always "further away" than the actual listening pos.
It's a known fact that AVRs during calibration sets the sub to be further away than it is. This is why.

 

[Deleted member 48726]

It really depends on how much correction you need below 100Hz. If it's just a simple XO then very few taps but add in a few amplitude corrections with phase considerations and they add up quickly. When I was using my miniDSP 4x10 I always took my corrections into rePhase to test which will alert you if you don't have enough taps to resolve a filter. There's probably a better way to do it with math but this worked for me.

Or you can just run Dirac correction without consideration and then do some sweeps in REW afterwards to check if it looks satisfying. That's what I do but only on Dirac profiles I find satisfying with me ears first and foremost. Some I don't even bother check. I do the correction, save the profile, adjust master vol and output with test tones, make profile backup, listen to a handfuld of known tracks. If it sounds better it sounds better. Sometimes I run some REW sweeps but keep in mind that Dirac has an algorithm it follows to sum all the measurements at different positions you've done. You can not diractly* compare your REW sweeps and Dirac profiles but use it to check for phase, distortion, GD, obvious mistakes.