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

I don't have a 568 here to test but I tested things every which way I could imagine and with flat center EQ and I have yet to hear or measure anything clip with Trifield music. Some CD's already have clipping (some inaudible and some audible).

But if you are actually hearing distortion something is wrong. I even tried pushing it to +10dB to check the limits. Never heard any distortion but things would get brighter and harsher if pushed to far.

A couple things you might do/try.

Is try setting L/R to +3.0 and listen in Direct Preset. If that distorts something is wrong. If you could send me a snippet of that tune that distorts I'd like to look at it and listen on my setup which has center at +3dB. It could be the tune already pushes the limits and now you can just hear it better. Could be a problem with your center channel. You could also try Trifield with no Center Configured in the system and Front Left Right and rears adjusted accordingly. You should be able to set every channel to +3dB with Flat EQ and not hear any ADDED distortion or clipping. You could also try swapping one the fronts the center with settings as is to verify it's not the center channel. Oh one other test. Try hooking up source direct to center DSP speaker (also 8K's while your at it). That's exactly the SAME level +3dB tweak is trying to normalize too.

Remember your only pushing things up to full scale. Nothing should clip with your settings. If you go over +3dB you might clip, but it's extremely gradual and it does not sound like distortion unless it was severe clipping.

I cannot personally vouch for a 568 but it appears from others it works fine.

I'm hoping it's just the tune is already distorted and the +3dB tweak is just bringing it out. A marginal tweeter can sometimes sound distorted.

I noticed a centre channel distortion with a Melody Gardot CD using Trifield, flat EQ, MRC byspassed and the +3dB tweak. I tried the speaker direct to the HD621 and didn't encounter the same distortion.

Any thoughts ?

Avii

@ George & Avii

If you take the 3dB tweak out and change nothing else do you still hear the same ?

cheers

I've reverted back to the speaker level pre the +3dB weak but keeping everything else the same. Using the same CD/track and Trifield DSP I don't hear the distortion.

Avii

PS The CD is "Melody Gardot Worrisome Heart" and I'm using track 5 "Some Lessons" to compare.

Which processor?

G61R non SL.

My response to Avii

I do hear the distortion right at 3:12s on the Trifield center.

I also hear it in Direct Preset mode which is exactly Full Scale 0dB with +3dB tweak, it won`t make something clip that isn`t clipping/distorting already. It`s exactly the same as if you connected your source direct DSP Speakers.

In Trifield I can`t guarantee center won`t clip (even with EQ Flat) but in practice and in lot of measuring I have yet to see Full Scale Stereo add up on the center such that it makes the center clip.

I also looked at the track in Audacity. Around 3:12 there is an extremely short and small clip in the original track. Too small to hear, no biggy.

Between 3:12 and 3:16 there is tons of headroom (this is right where distortion is for a couple seconds).

At 3:16 there is significant clipping, but I have seen worse and you usually don`t hear it something like that.

So my conclusion is there is real distortion (not clipping) and the +3dB Tweak is now letting HEAR what was always there.

The distortion may be amplified by Trifield processing because it is harder to hear in Direct Preset, but I can easily hear in Direct. I`m sure I`d hear it too if I wired Transporter to DSP Speakers.

I will try recording (when I find some time) the Center to see if Trifield with +3dB some how caused more clipping, but by what I have heard and seen so far I believe +3dB is showing it self off in how well it can now resolve things. The original track IS distorted. Surprising, on an otherwise clean recording. I`ll try it with headphones at work too.

I listened to "08 Love Me Like A River Does" again using computer, TASCAM MKII (running ASIO) and headphones. You can hear "distortion" all through that track. It happens again at 0:52 to 1:00, 1:48 to 1:58 and 3:12 to 3:16 (no where near clipping). I doubt it's a bad rip because it does the same thing in same situation each time during the song. One instrument was recorded too loud, probably too close to Mic and clipped at the recording. CD is not clipping badly and is not the cause of what your hearing nor is +3dB tweak. +3dB Tweak brought out every flaw in this recording.

Thanks for the info.

I'd be interested in your findings from the above.

Avii

After listening with headphones on a non Meridian system and clearly hearing distortion in the pure original track with 0 tweaking. There is no point in testing further on Trifield.

To put it bluntly.

Garbage in equals garbage out.

Mr. Meridian (the other George) sent me his sample file that was distorting.

This one was completely different than Avii's.

It turns out that for this track a womans voice (which was probably recorded mono) *IS* right at 0dB. When Trifield extracts it from the L/R channel it ADDs them together going up to +3dB. So in some rare cases Trifield Center does need 3dB of headroom.

I did a lot of testing including sweeps etc. to verify a problem like this did not exist. And I was amazed it popped up now after all this time. I don't know how the internals of Trifield work but if you had the Center EQ on and played this track with out +3dB it may also clip.

I've always thought Meridian's 3dB of headroom was NOT a guarantee of not clipping either. They just hedged their bet it would be rare. So with +3dB it less rare but still rare and in my opinion well worth it.

What was also amazing is, I have seen plenty of CD tracks that had much more clipping than what Trifield added here and it was not audible. This track is clearly audible.

What makes it very audible is the woman has a very sustained voice that mixed right to 0dB. So even though the clipping is not large, it's very regular. It has to be regular to hear it.

I personally have no intention of reverting back to -3dB, system sounds just way to good to go back.

Most well recorded tracks (and this is a well recorded track) just occasionally peak at 0dB and it's often a symbol or drum or something. I believe Trifield focuses on vocal range. So I don't believe this will happen often enough. I also don't know if how L/R are correlated effect how things Add.

It is a shame Meridian does not fix this problem. The problem is the same and the fix is a best compromise until Meridian fixes it right. This is just another example that the +3dB Tweak is not perfect. I know it's not perfect but I think it's way better than losing resolution all the time.

One work around folks could try is what ever your center channel is above 0dB then in Trifield subtract that amount so center ends up at 0dB. I would only do this if you are parnoid but like the +3dB Tweak. Of course this will compromise Trifield some, but before the +3dB I actually had my center backed off some because I wanted to give 6K's a little more priority (poor mans WIDTH parameter). So backing off center JUST in Trifield you may actually like regardless of the +3dB tweak and avoid any clipping.

This problem does not effect MOVIES.

So you have to decide if this is critical to you. If you have not heard it yet and like the +3dB tweak results, stick with it.

If Meridian does not come up with a fix there are other ways with simple hardware to fix it right that one of us may attempt to do.

Even with the "perfect fix" Trifield would still need headroom on the center channel and you'd still have to give up 3dB of resolution for the rare case of it adding to above 0dB (as it did with this case). The "perfect" Meridian fix should at least only add this headroom when needed on Trifield and not apply it to movies. Center channel can run full scale on movies.

The most important channels to get up to 0dB (+3dB in Processor) is the Main / Left/Right. The second most important is the Center (and a VERY close second for Trifield and Movies). I feel my center is SO much better with the +3dB tweak.

Just as in the 518, no one Digital Gain setting was perfect for all music. But on average 0dB (+3dB in processor) is the best target. Remember we are not artfically boosting it, we are removing an artificial attention the processor does.

If you only have a couple rare tracks with this problem and you don't want to give up +3dB tweak you could attenuate the track (or album) by 3dB.

Thanks for Mr. Meridian for catching this !!

Perhaps if I said, lucky folks that run Analog Speakers don't have this problem, it might raise an eyebrow at Meridian to FIX it.

George,

with all respects fro your +3dB work here what is really a great advantage,
you should not "throw the child out with the bath water"
and dont get mad with problems that are not related to your great +3dB trick.

Just to bring my thoughts here:

EQ in Trifield:

Trifield as you measured
http://www.meridianunplugged.com/wiki/index.php?pagename=TrifieldCenterEQCurves

the values "measured" on the screen from your graph

direct left is 92.2dB(purple line) and EQ-flat is yellow line with maximum 95.6dB- its a add of 3.4dB
with EQ1 96dB the add is 3.8dB
EQ2 is 97dB adds 4.8dB
EQ3 is 98.2dB adds 6dB

But look at Bobs article
THE PSYCHOACOUSTICS OF MULTICHANNEL AUDIO J. ROBERT STUART
http://www.Meridian-audio.com/w_paper/multips3.pdf
on page 8 Fig.19
a similar curve like your measured EQ1 is here-BUT not 3.8dB ABOVE the 0 dB maximum -
its ONLY 1dB above 0dB.

>>>>>>
George where I can find the original EQ graphs from M ?
Links to richardkroll homepage are broken other links link that
http://www.softronix.com/ubb/Forum1/HTML/000557.html
doesnt work..
>>>>>

So you write on the EQ measurements that the RELATIVE levels are correct-but where is the real or exact 0 dB level ?
That IS THE POINT to know.
I assume as you measured analog and you never have limited/clipped measurements-
that you have not measured at full scale.
Summary of EQ
Flat add 3.4dB
EQ1 add 3.8dB
EQ2 add 4.8dB
EQ3 add 6.0 dB
Absolute level is unknown here, but is "way" below full scale or 0dB,
as otherwise a distortion have to heard on the center when playing Trifield.
BUT that also means the main R/L signals is at least 6dB below full scale**(more see below)
with digital and analog systems !!
Or nobody heard that or no full scale CD is on the markt..


Coming to the +3dB trick

You measured clearly that full scale is -3dB below real full scale-
http://www.meridianunplugged.com/wiki/index.php?pagename=How%20to%20obtain%20level%20perfect
so you dit make the +3dB trick
And that is only with processors not without - means digital sources just to the DSP.

So thinking about the summary from Trifield EQ above-
the L/R signals have to be at least 6dB below full scale otherwise it get distorted with EQ2 and EQ3

and have to be at 3dB below full scale when EQ2 and EQ3 are not used(or get distorted)-
and then EQ1 and EQflat get only distorted partial(EQ1 with 0.8dB and EQflat with 0.4dB).

So doing the +3dB should be a good possibility to earn more resultion and have a very small risk to get distorted
-only with full scale signals that are very rare in reality (in theory).

George,
now when you have some CD that are clearly have signals at 0dB or allready clipped
the +3dB is still excellent-
through that wrong recorded CD away !!!
ITS A FAULTY PRODUCT (when above 0dB !!)

So dont go mad when FAULTY products are making your idea "Bad"

One possibility MAYBE
GEORGE when you please try it out

Take such a CD with full scale but not above !!

Play it with the +3dB trick and just lower 0.5 with EQ flat and 1dB with EQ1 the center volume during playing the CD.
Is the signal still distorted as BEFORE( I dont mean the 0.8 or 0.4dB from the +3dB with full scale CD) ??

If it not or less distorted, then M make at that point the center level adjustment digitally
as otherwise you have to have the distortion as before but just at lower volume(0.5 or 1dB from center level decrease).

So that test is important to know as then you can allways use the +3dB trick
and only on real fullscale CD you have a custom preset for Trifield with a lower center level and its done.

Processor +3dB trick

So thinking of the summary of the Trifield EQ offsets and there is no difference for analogue and digital here-
that means that also the "analog full scale" have to be 6.0 dB below zero othwerwise it get distorted with Trifield EQ2 and EQ3.
So I think also in the processor the +3dB trick have to done !!
Its simple logic, the signal(EQ2 EQ3) havent to get distorted-
so the maximum signal level at L/R have to be also at below 0dB or full scale-
means have to be at minus 6dB !!
So why not making the same on the processor?

But how?
Look at the config file there is no such thing as "front level" where to play with..
Only a 518 can help here to gain here digital so the D/A converter get the gained out(+3dB) output.
But that is different to gain after the processor or inside the processor..

Whats your thought on that?
Can you try out the 0dB CD and adjust the center level in preset or during play-
is there a difference and does have the signal the same distortion or not or just the volume level of distortion is decreased?

George, can make a file ( with some very short full scale signals)
for download and the HH can burn a CD then and more people can test that out?

Robert