0dBFS & Bass Management of DVD / Blu-ray Players

We've recently bench tested the HDMI and analog outputs of several Blu-ray players (aka. BD players) and noticed some interesting results.  This article focuses on the analog outputs of BD players and how they deal with 0dBFS input levels from program material.  Apparently not all DVD/BD source devices handle 0dBFS in a similar matter.  Before continuing, it's important to identify what 0dBFS means.  We have written several articles on this topic which can be seen here:

• Issues with 0dBFS in Digital Playback Systems

• The Case for Not Exceeding 0dBFS

In a nutshell, 0dBFS represents digital full scale or the maximum level a source device (ie. BD/DVD/CD player) should ever be presented by source material.  We've found some BD players output at 2Vrms when presented a 0dBFS signal while other outputs are at 1Vrms.  That in itself is not the issue so much as when the particular player cannot properly handle such a high input level and results in a clipped or distorted output. 

When we initially tested the analog outputs of the Oppo BDP-83SE we concluded that it could not properly handle 0dBFS input signals since it was distorting the summed subwoofer output with all speakers set to "small".  This wasn’t entirely accurate as we will explain with the mathematics after we first identify how bass management works in most DVD/BD players players.

When the subwoofer channel is fed bass from the five or seven main channels all set to "small", the summed output must be attenuated to avoid overloading the subwoofer channel as can be seen in the above bass management block diagram courtesy of the Dolby Digital Decoding Guidelines specifications.  The summed output can then be boosted in the analog domain to properly match the other channels.  Early DVD-A / SACD player models didn't properly handle this and often resulted in anemic bass response.  Denon was the first receiver manufacturer to address this via their analog inputs by offering a +10/+15dB bass boost for the subwoofer channel.  Others soon followed suite by either adding the gain back in their players to resolve this issue or offering the boost option in their complimentary A/V receivers external multi channel analog inputs. 

Both Oppo Digital and I re-tested the BDP-83SE player to verify its bass management system adhered to the functionality shown in the above block diagram.

Test Conditions

Oppo BDP-83SE Frequency Response (all channels "large")

 All speakers set "Large", test signal 192kHz TrueHD, 0dBFS, all channel trims set to 0.0dB

The output level was 7.7dBV (2.43Vrms) for the 5 main channels and 2.7dBV (1.36Vrms) for the subwoofer channel.  This result indicated that all channels, including the LFE channel were recorded at 0dBFS in the test signal.  The player adds 5dB attenuation to the LFE channel so the subwoofer output is 5dB lower than the main channels.

Oppo BDP-83SE Frequency Response (all channels "small")

 All speakers set to "Small", test signal 192kHz TrueHD, 0dBFS, all channel trims set to 0.0dB

Recall that Oppo's implementation of 0dBFS corresponds to a 2.43Vrms output level.  So by setting all channels to small and summing the combined response plus LFE channel, we achieve the following output level to the subwoofer channel:

• LFE - 5dB = 1.36V
• Left - 15dB = 7.7dBV - 15dBV = -7.3dBV = 0.43Vrms
• Right - 15dB = 7.7dBV - 15dBV = -7.3dBV = 0.43Vrms
• Center - 15dB = 7.7dBV - 15dBV = -7.3dBV = 0.43Vrms
• Surround Right - 15dB = 7.7dBV - 15dBV = -7.3dBV = 0.43Vrms
• Surround Left - 15dB = 7.7dBV - 15dBV = -7.3dBV = 0.43Vrms

Since the bass for each channel can be considered a correlated audio source, we simply sum them to get the equivalent subwoofer output as follows:

• 1.36V + 5(0.43) = 3.51Vrms or 10.9dBV  for 5 channels and
• 1.36V + 7(0.43) = 4.37Vrms or 12.8dBV for 7 channels

Clearly both the 5 channel and 7 channel cases exceed Oppo's 2.43V 0dBFs reference signal level.  Thus in reality the BDP-83SE does pass 0dBFS levels but not 0dBFS+ levels which result when all of the channels are set to "small" and a 0dBFS test signal is fed into the player.  This was something we did not take into consideration when initially measuring the BDP-83SE.  Incidentally we did measure other BD players that could properly pass a 0dBFS signal with all channels set to small but also noticed in one particular case their 0dBFS reference was only 1Vrms instead of the 2.43Vrms of the BDP-83SE.  It's quite possible the manufacturer in this particular case decided to scale back the incoming signal level in the DSP prior to mixing and converting to analog for this very reason. 

What we learned is that there isn’t an industry consensus for what output voltage 0dBFS should represent in digital playback systems but some manufacturers are better than others at allowing the user to compensate in the post summing amplifier stages via the level trims.  As a reference, most CD players analog outputs are between 2 to 2.2Vrms but they are not dealing with 4 or 7 channels either.  So it 2Vrms really the appropriate 0dBFS reference level for DVD/BD  players?

Is 0dBFS a realistic recording level for Multi channel?

We learned in the past that intersample peaks in two channel PCM often approached and/or exceeded 0dBFS levels for CD playback systems.  However for real world multi channel music and movies, continuous 0dBFS signal levels shouldn’t exist.  According to Dolby’s encoding guideline, the reference recording level for music and film should be -20dBFS.  As previously mentioned, common practice now encodes LFE -10dB lower than the main channels to leave enough headroom for line-level audio interconnect, and the 10dB is compensated back in A/V receiver or processor.  This is an issue we plan on more closely monitoring with BD source material to determine if it rings true for Blu-ray source material.  Could it be that BD movies typically have better dynamic headroom than the majority of music recorded today?  This wouldn’t be surprising given the current trends in the music industry that we outlined in the following article:

• The Dumbing Down of Audio

So at this point you're probably wondering why we spent so much ink about this 0dBFS stuff?  The reason is because when we initially tested the BDP-83SE bass management circuitry to document the crossover points, we did so with the subwoofer channel clipping and thus caused us to yield inaccurate test results.  We aren’t really sure how common 0dBFS levels are reached in modern multi channel recordings so we will test in adherence to more realistic test conditions that Dolby recommend of -20dBFS when testing the bass management circuitry and only use 0dBFS test signals for checking SNR, Dynamic Range, THD+N and output levels on a per channel basis.  These measurements are typically made relative to 0 dBFS.

Re-Test of Oppo BDP-83SE Bass Management

Oppo BDP-83SE Frequency Response

Using a -20dBFS 192kHz Dolby TrueHD test signal and all channels set to "small" with trims all set to 0dB, we measured the above response curve.  Channel 1 is the Main L channel while channel 2 is the subwoofer channel.  As you can see the -3dB point for both the HPF and LPF's is 100Hz which agrees with Oppo's findings as well.  We asked them to implement a "THX" mode that offers the following alternative filter response which they claim to be looking into at this moment:

• HPF: -3dB @ 80Hz, 12dB/octave
• LPF: -6dB @ 80Hz, 24dB/octave

Oppo BDP-83SE THD + N

With a -20dBFS test signal, the BDP-83 was able to reproduce playback levels at impressively low distortion levels (0.03% THD + N for the subwoofer channel and 0.015% THD + N for the main channels). 

THX Perspective

Thank you for sending this to us to review. Your findings represent the exact reason why THX does not certify multi-channel analog outputs of source devices. It’s too unpredictable. In addition to the output voltage clipping, there are other issues at stake. Even if the source device was capable of outputting the correct level, the multi-channel input of the AVR could very well clip as well.

So, as an alternative, you can start to lower levels across the board (closer to the noise floor), or do the -15dB/-5dB split and then boost it back in the AVR as you report (again, closer to the noise floor).

Where the bass management is done is critical. You don’t want to double up the bass management by using it in both devices. It should be done in either the source device or the AVR.

If done in the source device, the bass managed signals are then passed to the AVR. Some AVRs have internal bass management capability on the multi-channel input and some do not. If the AVR does have it, then hopefully the user has disabled the bass management scheme for just that input. If it does not, then the signals go straight to the volume control. Unfortunately, in this case most products are not able to apply speaker time alignment. Which means the speaker alignment feature is now missing along with any other post processing features.

If bass management is done in the AVR, the source device should be set to “all main channels Large, Sub-Yes”. But again, the user needs to know that their AVR has the ability to perform this function. If not, the signals will be incorrect.

In both of the options mentioned above, it can be difficult for the user to tell whether the level of the boosted channels is correct.

As you conclude in your article, there are no standards for multichannel analog outputs and how they are dealt with. Some AVRs can do bass management and sub level adjusting on the multichannel input and some can not. Source devices may or may not have good bass management options. Some source devices have higher output level capability than others and achieve the 10dB offset in different ways.

In addition to having the technical issues on the playback devices, there are still inconsistencies with how content is produced. The movie industry abides by strict standards with 5.1 movies needing a 10dB LFE offset. However, multi-channel music varies.

Therefore, back in the early days of DVD-A/SACD, THX decided that multi-channel decoding to analog in source devices would be incredibly confusing for the end user. Instead, we recommend that a digital carrier to the AVR would yield the best results.

I hope this helps.

Warren Mansfield

Director of Consumer Technology

Conclusion

Our research thus far has revealed that there really is no standard 0dBFS reference level used in DVD / BD players.  Some players will scale a 0dBFS signal back before summing it so the subwoofer output will not distort under this test condition while others will not.  At this times its unclear whether or not 0dBFS levels are common for BD multi channel source material but according to THX there does seem to be more care taken with multi channel movie mixes than music mixes to avoid this problem.  At this point, we will adhere to the Dolby recommendation of testing at -20dBFS levels when evaluating the multi-channel performance of DVD / BD players analog outputs.  We do plan on investigating this topic further both with recording engineers and our own empirical measurements of source material to get a better picture on how this affects real world applications.