- Written by John E. Johnson, Jr.
- Published on 21 November 2012
On the Bench
All distortion measurements were made within an 80 kHz bandwidth, unless otherwise specified, and the test signals were generated at - 5 dB, unless otherwise specified. The XLR balanced outputs were used for all measurements (except for the headphone output). Note that manufacturers usually specify distortion within a 20 kHz bandwidth, since that is the limit of human hearing. We use an 80 kHz bandwidth, except where specified, because 24/192 sampling yields a frequency response out to 96 kHz, and in order to keep the playing field level, so to speak, we also use the 80 kHz bandwidth for 16/44.1 tests. As a result, distortion measurements in our lab may, in some cases, be higher than the manufacturer specification. What we have found is that reducing the bandwidth to 22 kHz (20 kHz is not an option in the Audio Precision), gives a distortion measurement 30% - 75% lower than the measurement with an 80 kHz bandwidth. So, our data represent the "real world" measurement of signal in a bandwidth that just about every hi-fi component responds to.
First, the test results using signals from CD and DVD-A.
CD, 16/44.1, 1 kHz, 0 dB (Full Scale), output was 4.4 volts RMS, and distortion was 0.002%. At 0 dB, this is the highest voltage you can get from the player's output.
CD, 16/44.1, 1 kHz, -5 dB, the output was 2.5 volts RMS, and distortion was marginally higher than it was at 0 dB, 0.0037%, which I suspect is really just a reflection of the fact that the signal peak is lower in relation to the noise floor, as no distortion peaks are visible in either graph.
CD, 16/44.1, 19 kHz, 20 kHz, 1.74 volts output, the B-A peak at 1 kHz was not visible.
CD, 16/44.1, 60 Hz, 7 kHz, 2 volts output, IMD was 0.005%.
Using signals form DVD-A, the results were identical for 24/96 and 24/192, so I will only be showing the 24/192 results.
DVD-A, 24/192, THD+N was an order of magnitude lower than with 16/44.1, at 0.0003%
DVD-A, 24/192, 19 kHz and 20 kHz. The B-A peak at 1 kHz was 120 dBV below the fundamentals.
IMD at 24/192 was very low, at 0.0006%
The measured Frequency Response for CD, 16/44.1, and DVD-A, 24/96, 24/192 is shown below. For CD, 16/44.1, the response rolls off sharply at 20 kHz, while for DVD-A, 24/96, the response rolls off sharply at 40 kHz, and for DVD-A, 24/192, the response begins rolling off slowly at 50 kHz, to be down 1 dB at 90 kHz.
Now we come to the beginning of the bench tests with the USB input. Technically, all the USB inputs on the BDP-105 are asynchronous, in that the bitstream is reclocked to the BDp-105's DAC and then decoded. This is accomplished by storing the bitstream in memory as it arrives, and then re-clocked for decoding. The "official" asynchronous USB input on the right hand side of the rear panel is designed to be connected to your computer through one of its USB ports. In any case, all the USB inputs for the BDP-105 remove the jitter that is inherent to USB connections.
Here is the first example, with a 1 kHz sine wave, at 16/44.1. The distortion - 0.0037% - was exactly the same as the measurement with the test signal being played on a CD (see graph above).
19 kHz and 20 kHz yielded the following graph. There was no B-A peak visible at 1 kHz.
IMD at 16/44.1 sampling through the USB port was 0.005%, again, the same as it was when playing the signal from a CD.
With 1 kHz at 24/192 sampling, through USB, distortion was 0.0003%, which is the same as the 24/192 signal played on the test DVD-A (see graph above).
Using 19 kHz and 20 kHz signals, the B-A peak at 1 kHz was 103 dBV below the fundamentals.
IMD at 24/192 through USB was 0.0005%.
The frequency response through USB is shown below. It is down 0.5 dB at 20 kHz, and begins a roll-off at 65 kHz.
So, we finally have a signal being played through a USB port that is as good as playing it on a CD or DVD-A. This means you can store all your music, including high sampling rate (up to 24/192) on a hard drive, and connect it to one of the BDP-105's USB ports, and play it with the same quality as if you were using a disc. The end-of-days for putting CDs and DVD-As into the player is closer than we imagined.
I also measured distortion at 1 kHz through the headphone output, with the signal being played on a CD. Below is the resulting spectrum. The output was 1.9 volts compared to 2.5 volts output through the XLR jack, and this lower output in relation to the noise floor resulted in a higher measured THD+N of 0.009% (compared to 0.003% with the CD test signal). The signal is very clean, and is why the sound was so good using those HiFiMAN HE-500 headphones.
SACD appears not to be a hugely successful format for the release of music in the US, but Dr. David Rich, a Secrets Senior Editor noted that, "In classical music, SACD is now a standard for releases from the UK, EU and Japan. The SFO continues to release in SACD. Nobody is dual issuing, so if want the recording, you get the SACD with about $2 added to the price of a CD from the same label. Not all of the labels have US importation, which makes them extraordinary expensive. Japan is reissuing older recordings as SACDs with crazy prices in the US based on what Amazon wants.
Polygram is not releasing in SACD, so in most cases, the SACDs are with second tier orchestras and conductors. With world class performances from 5 - 50 years ago going for $7.00 per disc (in some cases $2.00 disc in 50 disc packages), it is very hard to justify purchasing an SACD in this country. I think I purchased 100 CDs this year. Maybe three were SACDs. Naxos went with Blu-ray Audio a couple years ago. Those are more than twice the price of the Naxos CDs (some have a little more material). As far as I can tell, nobody else is following this method.
Most of the demand for SACDs appear to be from people who think DSD sounds better than red book in stereo. Very few people listen in multi-channel. The mixes I have heard have nothing but ambience in the center channel, as the engineers think most people will be using inexpensive speakers under the TV. Since 80% of the sound improvement comes with direct center information from the stage, the current SACDs sound no better than Dolby Pro Logic II, and in many cases worse (DPL II creates a real center). DPL II can be remarkably effective on the correct CD as was demonstrated at the CEDIA Learn to Listen class, but the demo was with seven identical speakers, all vertical at ear height."
In any case, some of us have a significant SACD library, and the files cannot (at this point) be transferred to a hard drive and played through the USB port. So, here are measurements from a test SACD. I tested the player's SACD capabilities with the SACD mode set to DSD as well as PCM. The internal DACs decode the SACD mode setting and output the analog signal to the player's analog output jacks as well as output the signal digitally through the HDMI port in that same mode setting. I limited the bandwidth to 22 kHz due to the large amount of noise in the out-of-audible-band high frequencies that is normal for SACD recordings.
The graphs are shown in pairs, with the first one in DSD mode, and the second one in PCM mode.*
At 1 kHz, THD+N was 0.17% in DSD mode, and 0.006% in PCM mode
The 19 kHz, 20 kHz test showed no clear evidence of a B-A peak at 1 kHz in either DSD mode or PCM mode. Notice, however, the truncation of the frequencies above 40 kHz in PCM mode (through filtration in the conversion process).
IMD was 0.13% in DSD mode, and 0.007% in PCM mode. There seems to be a bit less second harmonic at 14 kHz in PCM mode.
Notes from Dr. David Rich regarding DSD vs. PCM modes: "I noted in the OPPO BDP-105 review you discussed issues of connecting an AVR so it would send DSD to the DACs. It is important to understand this will bypass the DSPs (multiple ones in a high end AVR). The room correction is lost, as is the added delay for distance correction. Surround synthesis overlays are lost (5 to 7 channels etc.). Bass management is lost. I would let the OPPO do the DSD to PCM conversion to bring all this back.
One might consider sending DSD as analog from the OPPO to the AVR (if the AVR has a 6 channel input), but these also bypass the DSP.
For stereo direct, simple bass management may be included. An analog 2nd order HPF at 80 Hz is added in the stereo path. The LFE channel goes to the ADC and then goes into the DSP section with the 4th order LPF. The LFE DAC then converts this to analog post DSP.
You cannot do signal processing on DSD, and even delay is very difficult since the DSP instruction set is expecting PCM. All the DSD direct mode does is send the digital signal from the HDMI decoder that supports DSD to the DAC with a DSD input. In the DAC, all the digital filtering is bypassed. In the analog section of the DAC, the MSB bit is toggled (DSD is a one bit system). Some modification of the MSB section may be needed to get it to work at the very high DSD switching speed. Only a few DACs will take DSD in. and they cost more. This is why most AVRs do not support DSD.
Getting DSD from the HDMI chip to the DAC is a microscopic modification in the microcontroller which issues the command to disconnect the DSP from the DAC and connect direct to the DAC (a couple of NAND gates) to the HDMI chip.
Issues can occur with the analog post the DAC with DSD since DSD has so much out-of-band energy. SACD players have special analog designs to deal with this but no AVR designer would add the overhead for a feature 0.1% of the users would use.
With HDMI (PCM (HDMI 1.1), Dolby lossless or DTS lossless (HDMI 1.3) , SPDIF, or USB, the PCM goes direct to the DSP, and you get all the functionality. This is the standard mode of operation for an AVR or Pre/Pro.
The jitter problem is worse with DSD since high frequency jitter modulates with the high out-off-band noise in a DSD signal and moves it to baseband. PCM does not have this problem since the out-of-band noise floor is flat at a very low level, and also does not occur in DSD mode save the Sony ES when driven with the one Sony SACD audio only player that slaves to a clock in a compatible Sony AVR."
Below is the measured frequency response, in PCM mode. Roll-off begins at about 25 kHz, with a sharp decline beginning at 30 kHz. This is done to minimize the effects of the large noise hump above 30 kHz.
*Here is some correspondence that I had with Jason Liao from OPPO (12-5-12):
"The DSD/PCM setting in the player affects both the analog and HDMI output. However to ensure that the player indeed uses DSD and not fall back to PCM mode, please either completely disconnect the HDMI output, or connect the HDMI output to a DSD-compatible AV receiver, or set HDMI Audio to Off. If there is an HDMI device that does not support DSD and HDMI Audio is not set to Off, the player will force PCM so the HDMI output has sound. The “Track Type” shown on the TV screen will tell you whether it is played as PCM or DSD. - Jason Liao"
"OK, well, I published the SACD results today, and I suspect that it may all be PCM rather than DSD. I will have to set the test equipment up again in the next few days and unplug the TV monitor that I was using to select the SACD test tones. - John Johnson"
"What happens if I use one HDMI output that goes to the SSP and the second output directly to an HDTV (which does not support DSD)? - John Johnson"
"In that case, there is a “Dual HDMI Output” setting in the “Video Setup” section. It should be set to “Split A/V” mode. HDMI 1 will be for video only and its audio output will be disabled. HDMI 2 will be for audio output. If the SSP on HDMI 2 can support DSD, the player will honor the DSD setting for “SACD Output” and send DSD to both HDMI 2 and the internal DAC.
On the other hand, if you set “Dual HDMI Output” to “Dual Display”, the player will convert DSD to PCM so both HDMI ports can have sound. - Jason Liao"
I re-measured the distortion with SACD, making sure that I was in PCM mode, by unplugging the HDTV that was connected to the HDMI input. I noticed that I could make the output from the player change from DSD to PCM and back by plugging and unplugging the HDMI connection from the HDTV. This should not occur when using an HDMI connection to your SSP, unless it is incapable of decoding DSD.
The OPPO BDP-105 has coax digital and Toslink optical digital inputs. They are limited to 24/96 rather than 24/192 because they are designed for connecting digital out from your HDTV if it has one, and they decode Dolby Digital as well as DTS signals, unlike a standard outboard DAC.
Here are the bench test results (they were similar to the USB input bench test results, so I am only showing the 1 kHz test):
With the coaxial digital input, distortion at 1 kHz (-5 dB) was 0.0003%.
With the Toslink optical digital input, distortion was the same as with coaxial.