Jump to content

Measuring Equipment for 7.1 ULF potential...


maxmercy

Recommended Posts

I have recently gotten the capability to send digital signals via HDMI to pieces of audio equipment and measure the rolloff inherent, and the ability to measure how well AVRs and pre-pros can handle a 7.1 channel LF worst case scenario, tracking the waveform by using my soundcard as an oscilloscope with a properly built DIY probe circuit to prevent from frying my soundcard.  There are some very big surprises.

 

Let's put it this way: No AVR or pre-pro I have put to the test so far (7.1 channels each encoded with -0.1dBFS sinewaves from 1-15Hz) can spit out a clean sinewave at -0dBMV with speaker trims set at -0dB, regardless of inherent rolloff.  They have all failed.  Two of the units are considered 'high end' from 'reputable' manufacturers.

 

More later.

 

 

JSS

Link to comment
Share on other sites

  • Replies 86
  • Created
  • Last Reply

Busy building bass traps with the available temps and sunshine today and tomorrow.  The short version: so far the failures have been (with different levels of 'failure')

 

Outlaw 975

Sherbourn PT-7030

Denon AVR2809ci

 

More details as I get time.

 

JSS

Link to comment
Share on other sites

Did a similar analysis not long ago, and came to the conclusion that mv+trim <=-12dB is required to ensure full headroom on the processors/recivers I have here, both Denon, Marantz. 

 

With trim level closer to 0dB, and trying to play at mv ref 0dB, you will easily clip the lfe+rerouted low frequencies digitally, before it even enters the DAC.

 

Don't remember if I posted any of this on data-bass, I have a spreadsheet with levels and headroom data if anyone is interested in the actual numbers.

Link to comment
Share on other sites

The clipping is not analog, as Kvasvoll said.  It must happen before the DAC as all the bass-managed signals are combined, as the waveforms are completely destroyed at the lowest freqs, and are not the typical 'flat tops' of analog cipping.  Once above 10Hz, distortion is still evident, but improved, and the waveforms look more like sinewaves.

 

JSS

Link to comment
Share on other sites

So far, not that I have encountered yet.  HHTYD comes to mind, EoT intro, but usually around -5 or -6dBMaxPossible for clean tracks.  TF2 has a huge peak, but most systems will highpass it away as it occurs at 120Hz.  But, if I am paying $$ for a piece of equipment, it should AT LEAST be able to handle the spec, especially if it a 'boutique' item that has been bench-tested to provide 9V unclipped output out of the SW out, but the digital summation craps out beforehand (Outlaw 975).  I would not have bought the 975 were it not for that bench test figure that came so highly touted.  Oh well... 

 

I do realize we are the freaks of the audio world, but some engineer out there had to know the summation headroom was inadequate.  

 

 

JSS

Link to comment
Share on other sites

The worst-case scenario aside, the real question is where does the processor lose it's shit. Trouble with consumer gear is there are never any indicator lights so users always believe everything is fine. I don't believe any AVR can output 9V cleanly, nor have I ever seen a spec that claims it could.

 

When you say bench test, you mean an independent test of some sort?

Link to comment
Share on other sites

 No AVR or pre-pro I have put to the test so far (7.1 channels each encoded with -0.1dBFS sinewaves from 1-15Hz) can spit out a clean sinewave at -0dBMV with speaker trims set at -0dB, regarless of inherent rolloff.

 

By spit out do you mean out of the analog outs?  Even if the digital signal did not clip and each channel was rated for a maximum peak output at 2V, maxing out 8 channels would sum to 16V output on just one channel.  The voltage rails would at least have to supply 18V for the analog side of things to even have a CHANCE of not going into clipping and 18V rails on the preamp side of things is rare.  In this hypothetical case even if the analog outs did not clip, there are few amps that would deal with that high of a voltage on their inputs well.  If that output was sent on XLR that voltage would be doubled to 32V!

 

I am curious of the peak voltage out in your tests because if you are using a soundcard as an o-scope and you need a special probe to attenuate the voltage to not fry your input, surely you are pushing the analog outs to their limits.  Is there a way for you to send the digital back out of the AVR or pre-pro you are testing and analyze it from there? 

Link to comment
Share on other sites

The worst-case scenario aside, the real question is where does the processor lose it's shit. Trouble with consumer gear is there are never any indicator lights so users always believe everything is fine. I don't believe any AVR can output 9V cleanly, nor have I ever seen a spec that claims it could.

 

When you say bench test, you mean an independent test of some sort?

 

Sorry, 7.2V:

 

http://www.soundandvision.com/content/test-report-outlaw-audio-model-975-preampprocessor-page-3

 

JSS

Link to comment
Share on other sites

 

Put in 115 for one, 105 for the other 7, and choose 'in phase'.  What you get is 125dB and change.  The way we calibrate, we calibrate with a signal that is 10% of full modulation (-20dBFS) and it should measure 105 or 115dB with a C-weighted SPL meter (there is much deabte on this topic, but this is the best paper I have run across, and use it as my standard)

 

http://www.meyersound.com/pdf/cinema_technical_papers/cinema_calibration_tech_report.pdf

 

So a proper PN file for calibration will actually be -23dBFS RMS (using sinewave RMS as -3dB of peaks as the 'reference').   So the calibration will yield 105dB on an SPL meter for a sinewave played back with 0dBFS peaks.  the 3dB crest factor of the sinewave is the difference in 3dB you are seeing.  The system needs to be able to handle 128dB peaks to have enough headroom.

 

 

JSS 

Link to comment
Share on other sites

By spit out do you mean out of the analog outs?  Even if the digital signal did not clip and each channel was rated for a maximum peak output at 2V, maxing out 8 channels would sum to 16V output on just one channel.  The voltage rails would at least have to supply 18V for the analog side of things to even have a CHANCE of not going into clipping and 18V rails on the preamp side of things is rare.  In this hypothetical case even if the analog outs did not clip, there are few amps that would deal with that high of a voltage on their inputs well.  If that output was sent on XLR that voltage would be doubled to 32V!

 

I am curious of the peak voltage out in your tests because if you are using a soundcard as an o-scope and you need a special probe to attenuate the voltage to not fry your input, surely you are pushing the analog outs to their limits.  Is there a way for you to send the digital back out of the AVR or pre-pro you are testing and analyze it from there? 

 

Yes, the analog outs, as those are the outs that will feed our sub amps.  And I expect the AVR manufacturers to plan for this worst-case by making 0dB trim on the analog out with max digital signal to be much lower than 2V, to allow for the +12dB trims most receivers have, and then some room to spare for summation on the SW out.  

 

The SW out need not be a strict summation, and need not be 16V or 32V; it just needs to send an unclipped/undistorted signal out that can then be amplified to produce the proper SPL via amplifiers and transducers.  Some AVR manufacturers have increased the capability of the SW out to drive pro amps.  But in my case, I have 2 EP4ks, with their input sensitivity being less than 2V, so I do not need huge output from the SW out.  I need it to be undistorted.  With speaker trims set at 0dB, and a worst case 7.1 signal, NO AVR can do it that I have tested thus far. 

 

I am not measuring voltage from the SW out (still need to build a proper calibrator to do that).  I just am measuring waveform integrity.

 

I do not know of a way to send the digitally summed signal back out before the DAC.

 

I just started measuring this stuff when I started playing back 7.1 worst case scenarios on my equipment and heard obvious clipping despite very low MV levels, and started to investigate.  These measurements were initially not intended for public consumption, they were only a way for me to find a processor capable enough for my needs; but since no piece of equipment passed my tests, I though people should know.  If anyone has a problem with my findings, measure it for yourself and post about it.  I for one would like to know what pieces of equipment can handle the specification without clipping anywhere in the chain.

 

 

You will need the following: 

 

1. Ability to send all-channel 7.1LPCM signal through HDMI at full modulation at infrasonic frequencies to DUT.

2. Ability to look at waveform out from DUT, and measure distortion if needed.

 

Report your results.

 

 

JSS

Link to comment
Share on other sites

Your findings are interesting, maxmercy.

If the signal is destroyed before it enters the master volume level attenuator, there is no way to avoid the distortion, and it will be present even if you play at lower spl levels.

 

I concluded that it is sufficient to have headroom for up to -6dB (down -6dB from the 125dBrms/128dBpeak max) from the worst case, and with the Marantz and trim level at -6dB this is acceptable.

The system wll be limited by clipping on the dsp input at some frequencies, due to the eq in the dsp.

Overall the whole chain is able to max out the subwoofer amplifiers, and further headroom is not needed until more headroom from more capable subwoofers are added.

 

All processing needs to move in to the computer, that is the future, no more headroom-limited and noisy avr/processors with proprietary dsp software.

Then you will have full control of all signal processing, all eq and bass management done in one place, and you can add proper room correction if you want to use that.

Link to comment
Share on other sites

Just to make sure I'm understanding this corectly this is absolute worst case scenario with full scale in every channel from 1-15hz. That's also with every channel trim set to 0 too. So anyone, like Kvalsvoll above, who has their trims below that should be ok correct? 6db down should be good enough I'm assuming for those rare occasions this happens? Not to mention all the guys who have efficient speakers that aren't even close to 0 on the trim.

Link to comment
Share on other sites

The Denon I tested also does well at around -7dB with trims at 0dB.  The Sherbourn and Outlaw had trims at minimum and still had broken up waveform at -20dBMV for the Outlaw and '70' on the MV control on the Sherbourn, (whatever the hell that means on that unit).  I had to attenuate the digital signal with nanoAVR by 6dB on all channels to get a clean signal on the Outlaw, attenuate by 7dB on all channels for a clean signal on the Sherbourn.  I do not have time to go into more specifics, nor to set up everything again to take O-Scope captures if people don't believe my results, as I am not selling nor trying to convince anyone of them; I was just reporting something I found odd/annoying about equipment that is very highly praised.  The methodology and software is out there and most of it is free.  Report your results.

 

I would encourage everyone to fully test their equipment.  I wish I would have before the '30-day no hassle return' period was up....oh well.  If I find any piece of equipment that can handle a 7.1 worst case with speaker trims at 0dB, I'll post it here.

 

 

JSS 

Link to comment
Share on other sites

I have such a love/hate relationship with that Sherbourn. Without nanoAVR to do bass mgmt, it has the quietest noise floor I have ever heard. Amps at max, MV at 90, and I can barely make out a hiss with my ear IN the SEOS. Such potential squandered.

 

JSS

Link to comment
Share on other sites

Archived

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


×
×
  • Create New...