Jump to content

Recommended Posts

  • Replies 118
  • Created
  • Last Reply

I ran some tests on the analogue outputs to check what was going on. The RME FireFace is, unsurprisingly, clean. The Marantz AV7005 seems to have up to 16dB available to the SW input before it clips.

Test setup was a 0dBFS sine wave at 60Hz, MV = 0 and SW trim = 0. SW output feed through a voltage divider (~28dB) to my audio interface and into audacity for a poor mans oscilloscope :)

The 16dB headroom is available across the fixed multichannel analogue input gains (0/+5/+10/+15 selectable) and the channel trim, i.e. if I apply +15dB gain on the multichannel input then I can get to +1 on the SW trim before clipping, if I apply +10dB on the mc input gain then I can get to +6 on the SW trim and so on.

This also made me realise I was running the output on the RME FireFace too hot, it was set to a +4dBu setting (which confusingly yields +13dBu at 0dBFS) rather than -10dBV setting (which yields +2dBV at 0dBFS). At the +4dBu setting, the Marantz was clipping like crazy with practically any signal so that's good to find out. It's v annoying that there are no good specs on input sensitivity for these things.

I think this means that the device can theoretically provide a max of 121dB, i.e. I was supplying a 0dBFS sinewave, which in film reference terms is 105dB, so +16dB is therefore 10 for the LFE and 6 for bass management. This is consistent with that dolby patent or the Oppo world view that says "~5dB is required for bass management". It will be interesting to see if this is different with a new D&M prepro, I'm sceptical that they will be any different but we'll see. 

To give an example, this with +21dB added in the marantz, +15dB on the MC inputs and +6 SW trim.

post-1440-0-03727500-1425830711_thumb.png

Link to comment
Share on other sites

The 16dB headroom is available across the fixed multichannel analogue input gains (0/+5/+10/+15 selectable) and the channel trim, i.e. if I apply +15dB gain on the multichannel input then I can get to +1 on the SW trim before clipping, if I apply +10dB on the mc input gain then I can get to +6 on the SW trim and so on.

 

It's v annoying that there are no good specs on input sensitivity for these things.

Is the 0-5-10-15dB a selectable thing in jriver?  If so, how would you know how that compares in level with a player's HDMI out to your Marantz?  What are you using to generate the sine waves?  Are you putting one in the LFE only or can you put tones in other channels too to re-direct and sum with the LFE? 

 

I hear you on how dumb it is that they never tell you what the max input is.  I usually shoot for 1.2 peak Volts at the most.  I go to REW's RTA and do the distortion analysis.  Play a sine wave and slowly bump it up until you see the THD go up and then measure your voltage at that point and that's your input's limit. 

Link to comment
Share on other sites

 

balloons_confetti.jpg

12751-dmpbd85img1big.jpg

WCS WINNER DMP-BD85

0dB trim.  No master volume so output is full level.  In fact, 0dB is the maximum adjustment for the sub out and all other channels.  It looks like someone actually planned for a WCS in their design!

 

Ok, dumb question alert...

 

 

... but if it has no Master Volume and is used as a pre/pro straight into the sub/speaker amps, how do you control the volume?  Is it a case of set it up and then be at the mercy of however hot the mix is (or isn't)??

Link to comment
Share on other sites

Is the 0-5-10-15dB a selectable thing in jriver?  What are you using to generate the sine waves?  

the 0/5/10/15 gain is in the Marantz not in jriver, I think it's a fairly common option in prepro's with multichannel analogue inputs. For reference from the manual

 

post-1440-0-76077600-1425903423_thumb.png
 
My approach was to take jriver out of the equation entirely. I used REW as the signal generator and selected channel 4 as the output which is routed to the SW analogue input on the Marantz. I then set the signal level to -3 and verified that the RME mixer saw that as a 0dBFS signal. I then measured the voltage coming out of the Marantz (at MV=0, SW trim = 0) and read ~6.9V with the SW input gain at 15dB and ~1.2V with the SW input gain at 0dB, i.e. there was ~15dB difference & the output was still a clean sine wave.
 
The RME was set to produce +2dBV at 0dBFS (i.e. ~1.25V) so this seems to be consistent with the Marantz operating at unity gain on the multichannel analogue input.
 
Therefore we can see that the Marantz can accept 1.2V in and produce ~7V out without any issues. This is plenty strong enough to drive my amp (and probably pretty much any amp) so then it becomes a question of how do you use that capacity to reproduce any content cleanly. I think the remaining question is what is causing the limit I was seeing, the capability of the (pre)amplifier itself or something else? 
 

Are you putting one in the LFE only or can you put tones in other channels too to re-direct and sum with the LFE? 

 
 
I haven't run a WCS test since I made these tweaks at the weekend, I think I'd currently pass at >15Hz and fail at <15Hz. Here's my reasoning....
 
At the moment, I do it basically as per the traditional spec so I have DSP blocks in jriver configured to do the following;
 
- attenuate main channels by 15dB 
- attenuate LFE channel by 5dB
- apply correction filter 
- sum low passes into SW output channel
- add 15dB back to main channels
 
If you ignore my correction filter then it means my SW output has 5dB headroom for the summed signal & I think we need ~8dB for WCS don't we? 
 
However I do have a correction filter (as shown in http://data-bass.ipbhost.com/index.php?/topic/379-maxmercys-wcs-test-disc-beta-and-an-o-scope/?p=6418)and this filter cuts 1dB at 10Hz, 5dB at 20Hz and much more than that in the rest of the passband). This means I have 6dB headroom at 10Hz (still fail), 10dB headroom at 20Hz (a passing score). I think breakeven is at ~15Hz when the filter is cutting ~3dB.
 
I need to revisit gain structure again though as a result of all this so I might attentuate further in jriver but then recover that gain via the SW amp itself (which is not maxed out atm). 
 

 If so, how would you know how that compares in level with a player's HDMI out to your Marantz? 

 
I think that depends how the DAC/preamp are configured doesn't it? at the moment, I am giving it 1.25V at 0dBFS which seems to pass through unaltered. The specs say the unbalanced preouts produce 1.2V at "rated output" but doesn't define what "rated output" means. 
 
I suppose I can switch the AV7005 to accept audio from the PC via HDMI instead and compare the resulting preout voltage. The output capability of the amp doesn't change either way though, it just becomes a Q of how the signal processing within the amp works. 
 

I hear you on how dumb it is that they never tell you what the max input is.  I usually shoot for 1.2 peak Volts at the most.  I go to REW's RTA and do the distortion analysis.  Play a sine wave and slowly bump it up until you see the THD go up and then measure your voltage at that point and that's your input's limit. 

 

I filed a support query with Marantz to find out what the actual specified max voltages is. They say it should be able to cope with about 5V in the input side & that a digital input will produce 1.2V at 0dBFS down the unbalanced preout.  

 

Thanks for the tip on how to measure this, am I right in thinking that you set MV on the low side so as to be certain it is the input stage that is clipping?

Link to comment
Share on other sites

Ok, dumb question alert...

 

 

... but if it has no Master Volume and is used as a pre/pro straight into the sub/speaker amps, how do you control the volume?  Is it a case of set it up and then be at the mercy of however hot the mix is (or isn't)??

 

 

You would not use this player as a pre-amp. The Oppo BDP-105 is unique in being usable as a pre/pro with some limitations like no auto EQ, no manual PEQ, global crossover, global crossover slopes.

 

Paul found a similarly high quality analog outs player from Cambridge (Azur 752BD). It's the same bass management scheme as the Oppo but uses Wolfson DACs and a proprietary TF2 upsampler and also has 3 filter selections to sort the digits, according to Cambridge.

 

The fact that the Oppo shits the bed with SW summing makes it useless. F***ing Hilarious how the Oppo fanbois jumped on Paul in the AVS BR players forum saying no one is interested in redirected bass and similar such bovine excrement. I wish some of you would have jumped in to help expose the travesty of that thread (wherein they have the model # of the Oppo player wrong to dodge search engines and dump all Oppo complaints). I almost reactivated my account just to bark a little to vent some frustration.

 

AVS has become a useless ad repository ruled by groupies and shills.

 

I'm looking at the Marantz pre/pro for decoding, DACs, bass management, auto EQ and level control and just getting any old player to bitstream the 1s and 0s to the Marantz. Hoping for a flat FR and no clipping in WCS discs.

 

The Oppo "FLAGSHIT" will go up for sale for anyone who would love to pay a lot of money for analog circuitry they plan to never use and for those who do plan to use it but don't mind grotesque distortion at whisper playback levels.

Link to comment
Share on other sites

Man,

 

I rarely visit AVS, no time for the idiocy.  I only check in to see what PassingInterest has done recently, he's really the only one I learn from there.  Nice of them to change the model number on the thread...clowns.   Had I known, I could have given support, but probably would have just gotten banned. 

 

Just like Gracchus said in Gladiator: "Rome (AVS) is the mob".

 

 

JSS

Link to comment
Share on other sites

I filed a support query with Marantz to find out what the actual specified max voltages is. They say it should be able to cope with about 5V in the input side & that a digital input will produce 1.2V at 0dBFS down the unbalanced preout.  

FWIW I verified this quickly & I would say this spec is accurate as the same signal from the PC produced about 0.5-1dB less SPL when sent via HDMI vs the analogue inputs. This is consistent with the slightly larger voltage the RME is set to produce at 0dBFS. Therefore it's seems clear that the Marantz AV7005 has about 16-16.5dB headroom available to handle LFE which means 6-6.5dB for intersample peaks and bass management, i.e. not enough.

 

Their support wouldn't provide a number for the max clean output voltage btw, they said they don't have that data (seems odd to me).

 

EDIT: those readings were absolutely wrong but I think correct in spirit. I revisited this after measuring the voltages and it seems the RME specs are also somewhat questionable as the -10dBV setting produces 0.8V at 0dBFS not +2dBV (1.25V). This meant my channel trims were off the mark. Anyway I think the 1.2V output is likely to be accurate for the digital input as, with the trims set correctly, I find that doing nothing but switching to a digital input produced about +4-5dB on the mains and -5dB on the LFE input. The voltage difference is about 4dB on the mains (~0.8V vs 1.2V) so that's about right, the 5dB on the LFE shows it is applying the usual +10dB rather than the +15dB I apply to the analogue input.  In the end then it looks like the device has the voltage headroom but whether it has the software to manage a WCS signal correctly is unknown.

Link to comment
Share on other sites

Looks like a Marantz I have here delivers 4.0Vrms clean on pre outputs.

When sub out trim level is set to -12dB, it will handle a all-channel 0dBFS signal up to 0dB MV.

 

The only situation that can make it clip is if there are low-frequency square-wave signals on any of the channels that gets summed and filtered.

The filtering will change the waveform so that the peak value increases, up to +3dB.

How this is handled internally I don't know, it actually did not cross my mind to check when I verified this.

If it clips digitally I really don't see a disaster anyway, as the clipping then will only preserve what was originally there from the soundtrack.

You can forget everything about nice sound, accurate sound, in fact, sound quality at all, when such waveforms are placed on a soundtrack, so it does not really matter.

 

If more headroom is needed, for the square waves, you will have to lower the ref level to somewhere below 0dB MV.

 

The 4.0Vrms limit is not a practical problem, as any subwoofer system should be able to get max output with 4.0Vrms or less.

As long as it is possible to reduce output level using trim or MV to preserve the waveform and prevent distrorion and clipping, you only need enough voltage to ensure the subwoofer amplifiers can be driver to clipping.

 

 

---------------

And thanks again for starting this thread, really an eye-opener and very valuable information for anyone looking for a new pre/pro or avr.

 

Link to comment
Share on other sites

At the moment, I do it basically as per the traditional spec so I have DSP blocks in jriver configured to do the following;
 
- attenuate main channels by 15dB 
- attenuate LFE channel by 5dB
- apply correction filter 
- sum low passes into SW output channel
- add 15dB back to main channels

 

As you are using jriver i thought i'd chime in.

I see you are attenuating main channels by 15dB and adding 15dB back after bass management.

I thought about that too but, if my understanding of things are correct that would mean 15dB of lost dynamic range that can't be recovered.

Adding 15dB would then just raise everything including noisefloor by 15dB.

 

I searched for alternative options and stumbled upon a vst plugin called Voxengo BMS.

It is an 8ch bass management plugin where you can set a separate mixgain, gain for the lowpassed signals that is to be summed to LFE.

There is a LFE master gain too for the summed signal but that is set at 0dB, see no reason to change it...

That way I can atennuate LFE before, to account for the summed signals, and by mixgain atennuate the lowpassed signals, and not have to atennuate the whole main channel.

That way I wont lose dynamics of the main channels, if it is an issue...

I do however atennuate them as I mix center to front for phantom center and mix together side and back surround, but that is another matter...

 

 

When I calculated needed atennuation I found that about 10.2dB headroom was needed to sum 7 105dB signals with a 115dB signal.

So the way I have it setup is to atennuate LFE by 10.2dB before bass management.

Voxengo BMS after, crossover at 80Hz and mixgain of 20.2 (10dB lower than LFE) and the filter set to move lowpassed signals from all main channels to channel 4 (sub).

Other signal processing needed is done in the second parametric eq section placed after Voxengo BMS.

 

My main gripe with the plugin is that there is no setting of individual crossovers for each channel, nor a setting for high and lowpass slope separately.

I think I've read somewhere that the highpass is set to 12dB and you can set lowpass filter separately, but not sure,

could also be that the slope setting is for high and lowpass both.

A loopback measurement of the filtered signals would reveal that though...

 

 

If you want you can test it out, would be fun to know how it stacks up against manually doing each step in jriver peq section.

I have not set up a system to monitor the signals as a poor mans o-scope, but if you want to then please test it and see how it fairs in summing worst case signals and also just how it sounds.

If you can hear the difference of not atennuating main channels 15dB and then adding back.

If there is no audiable difference then I think i'll might switch to just using jriver filters as they are more customizable...

There are probably other filterplugins too but ive not found any though...

Link to comment
Share on other sites

I see you are attenuating main channels by 15dB and adding 15dB back after bass management.

I thought about that too but, if my understanding of things are correct that would mean 15dB of lost dynamic range that can't be recovered.

Adding 15dB would then just raise everything including noisefloor by 15dB.

jriver's internal processing is 64bit float so there is no effect at all on dynamic range by dropping 15dB and then adding it back in. A 64bit number is just enormous basically relative to the audio signal, it provides about 385dB in dynamic range terms

 

i.e. using jriver to do this is totally safe and has no impact at all on SQ.

Link to comment
Share on other sites

jriver's internal processing is 64bit float so there is no effect at all on dynamic range by dropping 15dB and then adding it back in. A 64bit number is just enormous basically relative to the audio signal, it provides about 385dB in dynamic range terms

 

i.e. using jriver to do this is totally safe and has no impact at all on SQ.

 

In that case I think I'll switch then, to better customize the crossovers...

Link to comment
Share on other sites

the 0/5/10/15 gain is in the Marantz not in jriver, I think it's a fairly common option in prepro's with multichannel analogue inputs. For reference from the manual

 

So you are assigning channels to the RME interface and sending that into your Marantz's multi-ch analog ins?  I would figure your Marantz would have better DAC's in it by a notch than the RME does.  Have you ever A/B'd the RME vs Marantz for DAC sound?  If your RME will pass a spdif out in surround, that would be ideal compared to a video card's HDMI or motherboard's light pipe. 

Link to comment
Share on other sites

So you are assigning channels to the RME interface and sending that into your Marantz's multi-ch analog ins?  I would figure your Marantz would have better DAC's in it by a notch than the RME does.  Have you ever A/B'd the RME vs Marantz for DAC sound?  If your RME will pass a spdif out in surround, that would be ideal compared to a video card's HDMI or motherboard's light pipe. 

yes that's right re the analogue inputs. The RME, on paper, has the better DACs by some distance IIRC (AK4396 vs the AK4358 in the Marantz) though whether the whole package is superior is hard to say. Certainly I felt that the SQ improved a notch when I switched to the analogue inputs & that was with a lower spec focusrite saffire in the role of the RME. This was not a blind, double or otherwise, test though so take with a suitably large pinch of salt :) The ergonomics of performing such a blind test are woeful though, I don't see how it is feasible to do given the way the Marantz works. 

Link to comment
Share on other sites

The 4396 is better than the 4358 on paper, but it certainly seems the 4358 is good enough to be transparent.

If you see the amplfier-sound thread I started, the test loop includes the DAC in the Marantz, and at least so far I am not able to detect an audible difference.

That is for a test comparing orignal to 5x looped-through.

Link to comment
Share on other sites

When I calculated needed atennuation I found that about 10.2dB headroom was needed to sum 7 105dB signals with a 115dB signal.

So the way I have it setup is to atennuate LFE by 10.2dB before bass management.

as far as I can see (as in "based on my testing"), I need to attenuate each channel, in a 5.1 setup, by an extra 2dB to pass a WCS signal cleanly. This is a lot less than the above calculation so which is technically/theoretically correct? 

Link to comment
Share on other sites

as far as I can see (as in "based on my testing"), I need to attenuate each channel, in a 5.1 setup, by an extra 2dB to pass a WCS signal cleanly. This is a lot less than the above calculation so which is technically/theoretically correct? 

 

My calculation was for a 7.1 setup,  2 extra channels of potentially 105dB is going to be more demanding. 

Also you are using Jriver own bass management right? 

I'm not sure how that does in regards of atennuation and summing on its own,  so I don't know how much extra atennuation would be needed for each  channel to pass WCS signal over Jriver own bass management algorithm... 

 

I got my calculations from a coherent/in phase source dB summing calculator site. 

I summed 105dB together 7 times,  and then took that number and summed it by 115dB.

If my memory serves I got a result of 125.2dB in total,  meaning for the channel summing I would have to attenuate LFE by 10.2dB to make room for the additional 10.2dB from summation and main channels to be attenuated by 20.2 dB to make them 10dB below LFE. 

 

If this calculation is correct I don't know,  I would prefer to have it calculated in -dBFS as that is how the signal processing would be set up instead of in dBSPL as I've calculated above. 

But I could not wrap my head around how to do the calculations using negative dB so I gave up on that in the end. 

 

I have not checked this with oscope or anything,  but I have tested it with a 0dBFS sinewave track and copied it to all channels in jriver before any atennuation or summing steps. 

I then disabled all other processing apart from the ones involved in bass summing, clip protection disengaged and look at peak volume,  think it got to about 96 %,  never any overflow. 

That test made me think I was on the right track at least. 

 

However,  I read your(?)  thread at the jriver forum and some other threads. 

Somewhere there was said that volume adjustments in Jriver peq  adjust volume  might be relative and not absolute atennuations. 

If that is correct then it might upset the whole setup as it is based upon fixed absolute reductions of every channel for it to work as intended. 

 

But as I said,  my worst case scenario test was held below clipping as measured by Jriver peak meter and measurement by microphone indicates redirected bass from main channels to be 10dB below that of equal signal being sent to LFE,  so it seems right at least. 

 

 

Another thing I got to thinking about lately was if these atennuation margins would change if one has square waves instead of sine. 

But I don't think so as the peak value would be equal for 0dBFS sine and square right? 

Just that the rms of sine is - 3dBFS and rms of square is equal to its peak, right? 

 

 

Please correct me if I'm wrong in any of this,  so to not spread misinformation and so I can correct the signal chain! 

 

Edit: Damned autocorrect messing with my words,  typing on a phone is no fun :(     Hope the message gets through though.

Link to comment
Share on other sites

Was the calculator this one? http://www.sengpielaudio.com/calculator-coherentsources.htm

 

I think the values here agree with what I've seen elsewhere, namely that 5.1 asks for +8.25dB and 7.1 asks for +10.14dB. I went back and checked my test and realised that I was only running WCS on 4 main channels not 5, fixing that so there are all 5 in use mean that I need to apply -8dB to get it to report as clean in jriver (and verified that by recording the digital output). I think this means your figures were correct and it was just my test error that was confusing me. 

 

I don't use jriver room correction btw but I think it works in the same way, i.e. attenuate then mix then raise the levels back up. 

 

I don't think a square wave makes a difference, the peak is still at 0 so as long as the peak is passed cleanly then you're good to go. I might test that later to verify (aka satisfy the OCD)

Link to comment
Share on other sites

I use a Denon 3313CI, and I believe it behaves similar to the Marantz as far as being capable of putting out 4V and able to play 0 dBFS in 7.1 channel with master volume at "0" and sub level at "-12" (the minimum).  Of course, Audyssey can boost output up to 9 dB, so one must take that into account if using Audyssey.  One can also adjust a source trim (either analog or digital) if need be to reduce levels even further to provide gain for > "0" master volume or playback with heavy Audyssey corrections enabled.

 

Also, FWIW, I made these determinations using my MiniDSP hardware (a 2x4 and OpenDRC-AN) and its specs, some in-line attenuators, and the REW tone generator (produces output to FL and FR at the same time) coupled with adjustments to the source level trim in my AVR.  I've also tested with programmatically synthesized 5.1 channel FLAC files with digital full-scale wave forms.

Link to comment
Share on other sites

BDP-103D

oppo_bdp103d.jpg

A friend of mine is getting rid of his 103 because apparently it puts around a 150-200ms delay on it's HDMI inputs.  This becomes unbearable when you use a computer as the source.  Before he put it on ebay, I yoinked it and ran some WCS on it to see if it responded like it's big brother, the 105. 

 

It replicated the BDP-105's problem with clipping the re-directed bass on the sub out exactly.  It's noise floor on the sub output was around -83.7dB, which isn't very good for people who have high powered sub rigs.  I also found that the 103 was able to cycle through it's menu options faster than the 105.

 

OPPO = POOP

post-1247-0-93088700-1426730448_thumb.jpg

post-1247-0-18994200-1426730495_thumb.jpg

post-1247-0-58729800-1426730501_thumb.jpg

post-1247-0-24060000-1426730521_thumb.jpg

post-1247-0-28205700-1426730530_thumb.jpg

post-1247-0-91617400-1426730535_thumb.jpg

Link to comment
Share on other sites

A friend of mine is getting rid of his 103 because apparently it puts around a 150-200ms delay on it's HDMI inputs.  This becomes unbearable when you use a computer as the source

Is he gaming with the computer and routing through the OPPO? That's the only reason I can think of being worried by the latency. I've been thinking about doing the opposite and getting a 103 just to test routing through the HDMI in on my HTPC.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...