Meridian Unplugged - Powered by MrTechGuy.co.uk Forums Discussion Meridian - Systems +3dB "LevelPerfect" Tweak Performance Improvement Poll

No, I'm saying it should not sound worse.

Once you get that base line right flip on Trifield and hold on.

Correct. It was the HD621 that tipped me off something was not right.

With HD621 it sounded the same as direct wired. No better, no worse, no different.

But with G68 it lost something. What's different?

Speaker calibration.

So I effectively defeated speaker calibration in front.

Now g68 is no better, no worse, no different in direct DSP mode than wired. Just like the HD621 (which does much of what a G68 does but has no speaker calibration).

You need to get this right before thinking about what upsampling, resolution enhancement, apodizing, jitter reduction does. This is more important than any of that imho.

Understand that Meridian operates (processes) around the 16-18th bits. You don't muck with bits higher. And you can muck with bits lower all you want but you won't hear it. First thing Meridian does on a ht processor calibrated by the book is throw out one of those bits !!

Can you tweak the speaker output level at each speaker? I dont think so.

I think what you are doing makes sense George, but my V6 should be here soon and I want to leave my unit stock so I can do my comparison.

With my tool no, but I may add that. But I have enough options that you can make it do almost anything want and keep some sense of order.

In my opinion if you don't apply this tweak you are hiding the true performance of both the V6 and V4 (with DSP Speakers) so as you'll hear less difference without the tweak than with.

It's like comparing two sports cars on all season tires.

You need every freak'n bit if your comparing V4 against V6.

George and Nick - thanks for your replies. It seems strange that you can't have an uncoloured pre set given all of the other options available.

I have been struggling to get all of my testing done - I have a new SB Touch too which changes things as well. I implemented the +3DB today and it did seem to make a difference (like you say more noticeable then the 5200 apodizing).

I can see that I'm going to have to bunk a day off work soon!

Cheers
Tom

Just wanted to add that I am running Trifield center eq flat. It has sounded more natural to me since the beginning of listening to it.

I was as well before this.

George, can you please comment my readingsof the measured Trifield EQ curves are correct:

The graph is showing as explained in the text

The 4 top lines are the different EQ�s
top EQ3 then EQ2 then EQ1 and EQ flat.
Then follows the left channel top is direct mode(no center in layout)
and superimposed (same level) the left channel (still no center in layout) in Trifield mode.
So with no center speaker the mains are the same(Trifield or direct DSP mode)

When a center is in the layout and Trifield is selected,
then the highest frequency amplitude down to 8kHz are close the same for the L&R channels.

Then the level slowly start attenuated down to 1kHz and then approximately flat minus -3dB to 20 Hz.

Is that correct?

So when you do the +3dB "boost" on the mains, the are at "real 0 dB" only above 8kHz
and below that frequency you loose down to 3dB.
But that also means that without your +3dB thing(have the mains calibrated to 0dB),
that with Trifield mode enabled, the mains have a "lost" of 6dB !!.

Look at the EQ curves and Trifield selected with +3dB all front speakers calibrated,
then you have a problem with a +3dB saturation on the center when you select EQ flat and EQ1.
But this is only true when your recording have a headroom below 3 dB.
I believe that is extremly rare or happened not practically.
So you can adjust all fronts to +3dB and have no problem at all with EQ flat and EQ1.

With EQ2 and EQ3 you get saturated up to +6dB in some frequency ranges.
Still I think there is no recording with so less headroom,
but it could happen rarely that the DSP get saturated.
So EQ2 and EQ3 CAN have a problem.

Of coarse you can "adjust" the level for the center in the specific DSP presets,
but then you have not the correct level difference as it should be.

But that is only all true when the center makes the same SPL as the L/R speakers.
If the center needs for example a minus 2dB calibration adjustment to have the same SPL as the L/R DSP�s,
then you can still calibrate the center with +3dB.
To keep the correct level differences to the L/R speakers, you have to lower the center in the specific DSP preset over -2dB.
Then you are more save that the center not get saturated and you have guaranteed the correct level differences.

Doing so, you have all the advantages with nearly no disadvantages with the +3dB trick.

And you can do this similar to the rear and sides speakers in the DSP presets-different values of coarse.
You have to keep about the same differences to the fronts as without the +3dB trick.

Only when your center have a number closer to zero(bigger),
you have to take care about saturation of the music under seldom circumstandes.

What do you mean George-correct ?

Robert

Yes you interpreted the EQ Curves correctly.

The curves show there is even more gain than you'd expect on specific frquencies. Keep in mind thoses were measured in ANALOG.

What is most accurate about them is the relative differences and not the absolute values. The dB scale may be off (I'm almost sure it's off).

I later confirmed DIGITALLY that EQ3 does indeed need exactly 3dB headroom and Flat does not need any headroom. So +3dB and Flat is safe. And +3dB and EQ3,EQ2,EQ1 is not safe.

I posted these recordings of the center channel but I have not made screen shots of them yet.

http://softronix.com/pictures/levelperfect/

Also keep in mind if you calibrated your system "normally" to 0dB if you ran your Center +3dB in Trifield preset you get exactly the same result as Calibrating your system +3dB and setting Preset to 0dB. So even under Meridian's normal constraints you could make things clip.

We just want to be better aware of what things increase gain where when running closer to Full Scale.

This is why I added this feature to my tool. It will summerize risk of clipping and resolution loss on every channel and every preset whether you choose to calibrate to +3dB, 0dB or negative (to reach standard 75db pink noise).

Sorry George, do ask you again.

Iam confused now.

Look at your Trifield EQ curves,
all EQ have a +3dB boost from 20 to 200Hz,
then Q2 and Q3 further boost at higher frequencies about +3dB more,
means up to total +6dB compared to the L/R (when they are full scale).
In Trifield the R/L are only on high frequencies (above 8kHz) at the same level as with direct mode.

So when you boost all three speakers(all 0 db calibrated) with your +3dB,
you shift all front DSP to real full scale (on M terms +3dB).
As (from the EQ curves) the EQ are minimum 3dB higher all EQ INCLUDE flat EQ can get saturated with +3dB(when the recording is full range).
And EQ2 and EQ3 get even higher up to +6dB above full scale.

So why you then writing on your gain optimizer tool

---- Trifield DSP Preset Centre level Center EQ NOT FLAT adds 3.0dB yields 6,0dB (Clip risk 3,0dB)--

only clip risk 3,0 dB, its a 6,0dB risk.
Only for EQ flat and maybe EQ1(as is just a touch greater-0,5dB ?- as EQ flat) this 3,0 dB risk is correct.
For EQ1 and EQ2 the risk is here + 6,0 dB.

You wrote

"I later confirmed DIGITALLY that EQ3 does indeed need exactly 3dB headroom and Flat does not need any headroom. So +3dB and Flat is safe. And +3dB and EQ3,EQ2,EQ1 is not safe."

So thats fit not to the EQ curves.

Or when the curves measured digitally -as you wrote the curves are measured analog-
are the shifts are different?
When your statement in your programm is correct,
then all the EQ curves below 200 Hz are NOT shifted compared to the mains.
So only in frequencies higher the 200 Hz are the +3dB shift?

When that is true, it explain your writing.

So actual who much the EQ are shifted digitally ?


Another point to make the not full scale from M better:

There are level differences or offsets in the DSP presets.
This are coming from the listening tests and experience of M.
So I mean, these level differences should be left the same when the gain of the DSP are shifted.
If not, its sound different as M "designed" the sound.

So a solution could be

1) to setup each speaker to +3,0dB(M) means the scale of M.
In reality its real full scale or 0dB.
So we have the best resolution and have no lost here.

2)Then start the standard speaker calibration so every speaker is calibrated to the same SPL.
So you get for every speaker a different minus number.
Then you know how much offset is necassary to get the same SPL for each speaker.


3) For EACH DSP preset, there is a maximum signal level.
for example EQ3 have a level 6dB above L/R and maybe 26dB above rears.

4)Now the prozessor take care in EVERY preset,
that the maximum level doesnt go above 0dB(real dB).
In Trifield this is the center speaker, the prozessor knows how much every speakerlevel have to shifted from maximum.

5)Further the prozessor know, how much the speakerlevels have to be shifted to get the same SPL.

6)Then the prozessor calculate for the selected DSP preset, how much each speakerlevel have to be shifted.

7)Knowing all that, the prozessor adjust the VOLUME(not make the dB shift as now in calibrate) for each speaker to get the correct necassary SPL level.

8)What the difference to the regular procedure?

EACH speaker is then on full range(highest resolution) and produce the correct SPL with the volume adjustment- means the hybrid volume regulation from the DSP- and that mean not to loose resolution.

This procedure only sounds maybe complex when you first read it,but its simple.

And as M can set each volume in every speaker with a change in the communication sw,
it shouldnt be a big deal for them.

What do you mean?


Robert

No it's not 6dB. It's 3dB.

The scale on the EQ Curves is WRONG.

Just read those as EQ has more than FLAT (that's all you can read from it).

I measured the difference DIGITALLY (accurately), and it's 3dB, not 6dB when you turn on EQ3 (EQ2 and EQ1 are not all that much smaller increase).

The program is correct.

If you run full scale data with the +3dB tweak and run Center with EQ3 YOU will clip by approx 3dB. The wave file it uploaded and you see what it does yourself. I ran both Flat and EQ3 so you can see the difference.

Here is the difference between EQ3 and FLAT with +3dB