Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Hilton]

Hi just a question.

Have any changes been made to the DSP engine, specifically with JRSS and/or Room Correction?

I recalibrated my system today and a couple things struck me as weird..

1. When running the Voxengo SPAN analyzer I noticed I was getting clipping with MC DSP Output set to 2.1 routing and checked for stereo source only mix to 2.1 and sub set to silent use room correction. (basically just using DSP to do bass routing for stereo only sources to the Sub - Is that the right way to do it? Maybe I've been doing it wrong all along?)

I don't ever remember getting clipping before with all the same DSP settings I had before. ( I reloaded my saved DSP settings and they were clipping too)

2. I've had to reduce the sub level via DSP by about 10db in both 2.1 and 7.1 zones.
Previously I had the system calibrated so that 7.1 had -10db for the sub and I had 0db for the 2.1 zone.

My amp doesn't do -10db for multichannel analogue input so I've got the SUB amps set so that room correction takes out -10 db to calibrate Sub for 7.1 multichannel - this means that normal stereo and 2.1 sources the sub isn't to low - hence my original -10db settings for 7.1

Has any of the math or routing changed in the DSP engine or is it just me? or am I just crazy?

[Hilton]

After all these years I'm still learning stuff... haven't solved my problem/question but did learn a lot more about phasing issues and crossovers with LFE/Music bass redirection.

Interesting stuff

https://film-mixing.com/2016/01/25/the-lfe-explanation/

[Hilton]

Here's my usual setup and calibration process from a previous post if it helps.

http://yabb.jriver.com/interact/index.php/topic,102429.msg719078.html#msg719078

[mattkhan]

My amp doesn't do -10db for multichannel analogue input so I've got the SUB amps set so that room correction takes out -10 db to calibrate Sub for 7.1 multichannel - this means that normal stereo and 2.1 sources the sub isn't to low - hence my original -10db settings for 7.1

what do you mean by "your amp does do -10dB for multichannel analogue input"? AIUI the "room correction" block does bass management correctly when there is an LFE channel in place so your amp should see the correct signal levels on each channel. 

[mojave]

My amp doesn't do -10db for multichannel analogue input so I've got the SUB amps set so that room correction takes out -10 db to calibrate Sub for 7.1 multichannel - this means that normal stereo and 2.1 sources the sub isn't to low - hence my original -10db settings for 7.1

The LFE channel is encoded at the same level as all the other channels at close to 0 dBFS. However, it is meant to be played back 10 dB higher. Use Tools > Advanced Tools > Audio Calibration > Volume Calibration for 7.1 to check your levels. The LFE channel on Volume Calibration is purposely -10 dB and should playback at the same SPL level as all other channels if levels are set correctly.

You can use "Set levels from decibel meter" in Room Correction and JRiver will automatically set levels bases on inputted SPL.

Has any of the math or routing changed in the DSP engine or is it just me? or am I just crazy?

I saw some strange issues during testing about a month ago, but they went away on reboot.

[Hilton]

Thank for input Mojave appreciate it.  I was up till 4am playing around last night and I've got it sorted.

I could have sworn I used to always have -10db sub PEQ applied to the 2 channel zone but it seems that is no longer required or my calibrations were out previously some how.

Anyway have worked out:

If you use MC room correction in either a 2.1 zone or 7.1 zone the LFE is definitely output at -10db down. (which it should be)
The only reason to apply a -10db to LFE/Sub PEQ is if your not using room correction in a zone and you cant do the correction in your AMP or receiver.

---

To clarify for Mattkahn - in my case the Sony STR DA5200ES AVR is quite an old one now and it doesn't do the -10db correction for the LFE channel when connecting via analogue inputs (RCA inputs) - it does however apply the -10db LFE via HDMI for dolby or DTS decoding in the AVR.  That's why I used to have to apply -10db to a PEQ for 2 channel sources when doing processing in MC and connection via analogue to the AMP.

In any case all back to normal and sounding great.  I did learn more about phase cancellation with crossovers and low-pass filtering however as part of the process, so ultimately very much worth the trouble.  System sounds better than ever.

Crossovers and filtering between LFE and redirected bass can make a real mess and cause lots of phase cancellations (which I had a few issues with) which is why I think my calibrations were out. I guess I just had a bad combo of filters and crossover points were causing me some conflicting readings.

Thanks again - love the crew here and your willingness to help - happy new year to you all!

[Hilton]

OK - So after many hours of playing test tones of various types for 2.1 and 7.1 I finally have learnt a few things about JRSS and wanted to share what I've come up with for my system that sounds far better for me than the system ever has. (and it measures way better too)

Bass management is a messy business that's for sure.  In any case I also wanted to check if there's any better ways to achieve what I now have setup???

First off - Linkwitz Riley manual 24db crossover slopes and manual bass management are the bees knees now I know how to do them (I think they're right!)

For 2.1 upmix to take advantage of the sub for stereo content I worked out that Output must be set to 2.1 or rerouting channels to the sub wont work.

For manual bass management and crossover setup I've disable JRSS and have setup custom PEQ's to reroute channels and create the cross overs.

2.1 bass management by Hilton, on Flickr

The order of these is important and I think I've got them optimised here.
Here I've setup the copy left and right channels to the sub and taken 10db out to allow for the +10db which is set in my receiver/amp. (analog inputs)

Also stacked 2x high-pass 12db crossovers for left/right @ 70hz to create a 24db Linkwitz Riley crossover. My speakers are about 5db down at 40hz and 60 hz crossover was too low, 80hz was too high for my taste.  (this sums to 0 db at crossover and is also in phase at crossover) (note I had lots of phase issues using JRSS and the built in cross overs - this manual approach has fixed a few holes I used to have and boominess around the crossover because of phase cancellation issues and bass rerouting logic/choices made in JRSS. - more on JRSS in a moment)

I've also stacked 2x low-pass 12db crossovers for the sub @70 to create the other side of the crossover for the subs. (symmetrical sounds way better to me with zero summing at the crossover and completely in phase)

Can I encourage matt to change to or make LWR crossovers default or at least optional for room correction? Everything else about the room correction is pretty much spot on.

I've also taken out 5db at 160hz because of a room mode there, other than that the room is actually pretty good from my main seating position. (I believe in minimal correction)
Last bit is taking 3db out of the sub to get overall levels right with calibration test tones. I could possibly do this in the receiver too, but calibrating the receiver for other audio sources outside JRiver meant that I ended up with 3db more sub level than when using the receivers built in bass management. This allows me to switch between any source without having to fiddle anything.

2.1 bass management-2 by Hilton, on Flickr

So manual 2.1 bass management is actually pretty straight forward and the results are definitely worth it.  The bass in my system is so much tighter, full and flat.

Here's a typical 80's pop song with that characteristic bass roll-off on the kick drum.  You can see the crossover doing its thing and levels are all good without having to tweak final output levels - even with very hot mixes ie mastered or normalised to 0db (why engineers do that today is beyond me - but I digress.)

span - 2.1 bass management-2 by Hilton, on Flickr

Multi-channel however was a lot more involved......... and I'm still not sure I've done it the best way.

Here I've set output 7.1 and decided it was too hard to try do all the rerouting manually so I've let JRSS create the upmix channels and levels from 5.1 to 7.1.
But this is where it gets weird.. more on that shortly. 
I'm using silent redirect to room correction, however I'm not actually using room correction at all. All the bass management and room correction is again done manually in PEQs.

7.1 bass management by Hilton, on Flickr


Here's the manual PEQs for bass management. Same deal as for 2.1 crossovers and left/right/centre bass management, except you can see I've had to do quite a lot of jiggery with the JRSS levels to get them to measure and sound right.
-12db instead of -10db
another -7db on the sub (also have a question why I need to do this when it was calibrated for 2.1 I only needed -3db to measure correctly - that's another -4db down ontop of the -12db I've already done)
and overall another -5db across everything to bring back below clipping at 0db input signals.

This last bit that was troubling me earlier and prompted the original post.
JRSS does something weird here I don't think it should be doing. 
Using Dire Straits 6ch 96/24 brothers in arms (and others I've tested) I get between 2.7db and 5db clipping (overs) on left/right/centre even though all I'm doing is reducing levels and taking bass away from the mains with bass management.   Now I know the final output from MC is limited or compressed to stop the overs, but 5db??? that's an awful lot. Even 3db is a 50% reduction in power output level, and it's noticeable and sounds like compression limiting kicking in. It takes away quite a bit of the impact and dynamic range of some songs. Sure you don't notice it on everything, but it's been a huge change to the sound I'm getting from my system.   Surely JRSS shouldn't be boosting everything by 3 to 5db to upmix from 5.1 to 7.1. with just bass management invoked - I can clearly hear and see JRSS increasing the levels when turned on???!!??.


7.1 bass management2 by Hilton, on Flickr

Here's corrected -5db on final output
span - 7.1 bass management-2 by Hilton, on Flickr

Here's the overs with the final -5db PEQ turned off. So what's the deal guys? Is that normal?
span - 7.1 bass management-clipping by Hilton, on Flickr


So what are your thoughts? Is this normal? Is it the best way to do manual LWR crossovers and bass management for upmix to 7.1?

[Hilton]

I've done some more investigation and more thorough calibration with REW and ye old ECM 8000 mic. (with calibrations files)
(BTW it sounds much better than using JRSS bass management - bass always sounds over-cooked with JRSS and the room correction filter choices available (butterworth vs Linkwitz Riley) and I'm sure when I get around to measuring it will show it too)

Doing various tones, sweeps and analysis with LFE only, LFE + Bass management I've settled on measurements and settings that agree with the dolby digital standard for bass management. (see below)

That is;
-15db for all bass managed channels for redirected bass
-5db for LFE
+15db for LFE in analog stage

I'm doing the analogue boost through +10db on AVR LFE channel + another +5db from turning up the Sub amps 5 more db
And I've also taken another -2db from the sub channel in MC in the final tuning. (-7db sub instead of dolby recommended -5db) probably because my Sub AMP levels are a tad higher than they should be.

dolby digital bass management by Hilton, on Flickr


dobly digital manual bass management in MC by Hilton, on Flickr


and REW room measurement with the above settings
Dark purple is the measured response
Light purple phase
Black solid and dashed are calibration files for mic and soundcard
2.1 and 7.1 zones have identical bass management settings in MC DSP. (apart from JRSS doing upmixing to 7.1.- and for 2.1 JRSS is completely disabled)

mc-calibrated by Hilton, on Flickr

[Hilton]

Wow - awesome Group Delay. I always knew these speakers and crossovers sounded great.. now I know why. 

Minimum Phase system 

group delay by Hilton, on Flickr

[mwillems]

That's some pretty solid group delay!

I'm glad somebody else is taking up the call to switch bass correction LR by default;p the current filters really do introduce irregularities at the crossover and I've been complaining about the "butterworth bump" for years. 

BTW am I reading your FR graph right? Do you really have bass playing down to 2Hz or is it a measurement artifact?  I don't usually start my sweeps that low so I don't know what happens down there.

[jjazdk]

You have almost 40dB of variation in your bass output, according to your REW measurement!?!

It doesn't look like a proper measurement for a couple of reasons:
1. Your input level is way too low, averaging at 60dBSPL with a dip to 40dB and up to 80dB max. If your microphone calibration is correct, part of your measurement is down in the noise floor, and that does not give valid results.
2. Your graph has 20dB pr. division, thereby masking most of the variation in the curves.
3. Smoothing with 1/6. octave in the bass hides too much of what is going on.

A few advices/hints:
Raise your level to something like 80dB SPL on average.
Adjust your graph such that the entire range is from +40 to +100dB, instead of -80 to +160dB.
Remove all smoothing for pure bass measurements, or use a maximum of 1/12.

As mwillems points out, having full output from 2Hz to 20hz is a clear indication that something is wrong with the measurement :-)

mc-calibrated by Hilton, on Flickr

[mwillems]

You have almost 40dB of variation in your bass output, according to your REW measurement!?!

I noticed that myself.  I thought it might be potentially accurate as the 40dB variation is all below 30Hz or so (from 40Hz up everything looks smooth), with the rise only coming below 20Hz in the subsonic region.  I could imagine that being the case if he's running bass shakers or some other "special effects" type sub-subs, so I'm open to the possibility that the rise from 20Hz down some distance might be  real. 

Seeing the unsmoothed graph would make it clearer, but I agree that it's unlikely that the section below 10Hz is entirely accurate unless he's got one of those fan subwoofers that can play down to DC.

[mattkhan]

fwiw you also need to click "generate minimum phase" to see the minimum and excess phase on the graph

[Hilton]

Hi - There's no meaningful measurement below about 25hz, that's all artifacts and noise from the cheap phantom power supply in the mixer and self noise from the mic.  I'm looking at a different mixer, mic, and/or dedicated phantom supply to fix that.
The slot ported Subs don't generate anything below 25hz which is why you see the roll-off at 30hz.

The Group Delay is smoothed 1/6th Octave so it's probably not a perfect representation but GD is very good none the less.
All room measurements were done with 85db - I've moved stuff around on the graph.

Can assure you the system sounds like it measures. Have a mate with perfect pitch hearing come and sanity check it and he can hear the peaks and dips with the sweeps.
Worst ones are dip @220 that I left and peak at 160hz that I pulled back with eq and another dip 85hz I've not yet treated but may look at that one. The rest is +/- 3db with a gentle roll-off from 4k to about -6db at 20khz (natural sound signature of the speakers - which just happens to be Bob Kats and many recommended curve - these speakers were built with that natural response.

I'm measuring at 3M distance and at seating height.

[Hilton]

I bought a new Audient iD14 digital interface for it's excellent Mic pre-amps (Class-A) and replaced the 15 year old ECM 8000 with a new one.  Noise problems fixed! Really nice piece of kit.

Here's the updated measurement with 1/24th smoothing and 18 manual PEQ bands in MC. Sounding much better. About +/- 3db mostly.

Pretty happy with that until I checkout Audiolense XO and Acourate - did a quick comparison on MC PEQ vs Audiolense in 2ch trial mode and there's definitely some differences probably worth investing in. 

1-24th-smoothing by Hilton, on Flickr