Hello everyone,
After several reviews of CD Players, you've been quite a number to ask me how to do the same. I'm really pleased to see that I'm not the only one interested in these old (or sometimes new) devices. So here you go with a very quick guide on how to do the same.
I will regularly update this message with your input and feedback, so feel free to come back as it will evolve.
It all starts with a Test CD
We first need a test CD to perform the measurements because the CD Player itself will be the generator.
When I started to review CD Players, I was using available test CD like those of Sony, Denon, CBS,... But they were old, difficult to procure, expensive, not respecting the latest standards (eg: no dither) and often incomplete in terms of test signals. So for the following 12+ months, I worked at creating a Test CD, with the below objectives:
Please note that the current iteration is still a "prototype", but it's already very advanced. Few highlights of what makes this test CD unique:
To get the latest iteration of the test CD and the accompanying excel file with all details, please PM me (current version is 7.1.02).
!!!Warning!!!
Thanks to @restorer-john as I copy/paste his words:
This CD contains tones at frequencies and levels that can easily destroy expensive loudspeakers/tweeters and/or amplifiers.
It is not a CD to 'play' on your HiFi system. Use only with measurement instruments unless you are 100% confident you know what you are doing and particularly, what track is coming next. For example you may be trying to 'hear' a -80dB 1kHz sine by turning up the volume and then along comes a 20kHz 0dBFS sine which instantly smokes your Revel Salon 2 tweeters...
We accept no responsibility for any damage you cause to your gear. Entirely at your own risk. Be sensible.
How to burn the CD
If you use Windows, when burning the CD, order the files by file names and you'll get the same order as the one shown in the Excel. If you use Windows Media Player to burn the CD, don't forget to remove "Apply volume levelling across tracks" else it will introduce low level distortion:
The test CD can be used with many software including, and not limited to, REW, Multitone, ARTA, etc...
Quality Check of the burnt CD
It is easy to check that your burnt CD is of good quality. After the burn, simply rip the CD to a folder and perform a WAV analysis with REW or Multitone to confirm no digital transformation. Here below is an example from REW of a WAV analysis between the original file and a rip from the burnt CD (overlay):
The two traces are the same, and that means the burnt CD is flawless when compared to the original WAV files.
Required hardware
All you need are:
Since we are talking CD players, you do not need an Audio Precision as many modern audio devices have a much better resolution than CD Audio. That and the magic of FFT analyzis will help you go really far into measurements.
I personally use a Motu Ultralite MK5 and, as I said, many others can be used.
How to use the Test CD with REW - Initial steps
I've been using REW since the beginning and what follows is how to use the Test CD with REW. Most of what I wrote is valid with other Software such a Multitone, though. I think it's best you continue using your preferred measurement software, if any. Only the first two files of the Test CD are REW specific, not the others.
The below is not a substitute to the very complete REW user guide, but only complementing it in the context of testing CD players.
You can use an ASIO driver with REW, the Java driver has a limited bit depth (16 bits only), but as @staticV3 mentioned "With 64bit REW, you can use the Java EXCL output for stable, bit-exact, multi-device 32bit I/O, without installing any ASIO driver".
First, and even if you don't use the interface in a loopback (inserting the device under test in the loop out->in of the interface), I still recommend to calibrate it, especially for 44.1kHz inputs as it will reduce non-linearities around 20kHz as the filter ADC's interface is likely to influence signal. So this step will potentially improve phase and frequency response measurements (because REW forces the sweep to be measured at 44.1kHz). It's no big deal if you don't calibrate the interface right away as you can apply a "cal" file afterwards on saved measurements.
It's anyways less important to calibrate for 96kHz or 192kHz inputs as we don't need to analyze FR and phase at those high frequencies and your interface is likely to be flat from 20Hz to 20kHz with these sample rates.
So, use the below video to calibrate at 44.1kHz at least (again it's not a mandatory step):
Create REW Calibration file
Note that as opposed to what this person says, the calibration does not reduce distorsion, and only compensates FR and phase non-linearities.
There are additional videos to better understand how to play with REW (inserting a device to test in the loopback of the audio interface), but they don't apply to CD Players testing:
Play with REW, part 2
Play with REW, part 3
Play with REW, part 4
How to use information from the Excel
The below are print-screens (from REW, but it would be similar with other Software) to help you understand how to read the instructions of the Excel file.
First, the sampling rate for the measurement (eg: 24bits/96kHz):
The 24bits is of no real use because REW will always capture at this level, or 32bits if the interface allows.
The other excel columns indicate how the FFT for measurements (from RTA in REW):
Depending on the test file, measurements and sampling rate input, I recommend 128k, 256k or 512k FFT sizes. It allows the FFT bins to be small enough to get very good precision when computing THD and lowering random noise (via the FFT gain).
BH7 refers to Blackman-Harris 7 type of window which offers a good enough precision and facilitates measurement (less windowing errors). I always leave Max Overlap at 93.75%.
Now, let's go to the distorsion settings panel:
The AES notch is recommend as per the AES standard to add a notch around the fundamental, which in fact reduces the calculated noise, especially for those players that have correlated or random noise at the bottom of the fundamental (case in my example, by the way).
I always use a high pass and low pass at 20Hz and 20'000Hz respectively. Not everyone does the same. In an AP system, the user can chose too, but most people chose 20kHz or 22kHz as low pass. I personally think 20kHz makes more sense and is the recommendation of the AES17-2020. But that needs to be modified for some measurements, especially all IMD DFD at high frequencies (such as CCIF or AES).
Similar case for the computation of Intersample-overs where I simply unselect the "Use distortion low pass" . We indeed need to let REW calculate the THD up the 96kHz in that case. The excel file provides my recommended settings for each measurement.
Other REW only related information
In the RTA control panel, I always select the "Get fundamental from generator":
That way, when you perform a measurement from a CD Player, and even if REW's generator is not the one sending the signal, you can open it and select the measurement being performed from the CD Player. For instance here below a Dual Tone (SMPTE) is selected and so RTA window will adapt to it and search for the two tones and correctly calculate the IMD distortion:
No need to press the "Play" button at the bottom, RTA will adjust as per the settings of the generator.
Last but not least, the first two files of the Test CD are sweeps to measure FR, phase and THD vs Frequency in Left and Right Channels. To use them, click "Measure", drag and drop the wav file in REW window, click start and press play on the CD player (first file is for the left channel). Repeat the same for right channel.
I hope all of this is clear enough, but feel free to request assistance in this thread.
Final note:
Pay attention to fact that many files of the Test CD include multiple levels, for instance "C_Sine_999.91Hz_0dB_-1dB_-3dB_-6dB_NoDither_L.wav" is 2 minutes long, divided into 4 sequences of 30sec each of the same test tone played at 0dBFS, -1dBFS, -3dBFS and -6dBFS. So you can click the "Save Current" button of RTA to save them in the main window after REW has performed 32 averages:
Record your measurements
REW allows you to save measurements. I recommend you do so as it's really cool to overlay the results between multiple CD Players later on. I save the files names per track number, CD Player model and channel being measured. For instance, for the Marantz CD5400, the measurement with track #02 on left channel would be named "02 - CD5400 - L". I save them under a folder that is named per the version of the Test CD, so it's also easy to remember which file is what, using in the referenced excel file.
That way, I can also share my measurements with others and it makes it easier to compare between us.
Enjoy testing your CD player(s)!
--------
Flo
After several reviews of CD Players, you've been quite a number to ask me how to do the same. I'm really pleased to see that I'm not the only one interested in these old (or sometimes new) devices. So here you go with a very quick guide on how to do the same.
I will regularly update this message with your input and feedback, so feel free to come back as it will evolve.
It all starts with a Test CD
We first need a test CD to perform the measurements because the CD Player itself will be the generator.
When I started to review CD Players, I was using available test CD like those of Sony, Denon, CBS,... But they were old, difficult to procure, expensive, not respecting the latest standards (eg: no dither) and often incomplete in terms of test signals. So for the following 12+ months, I worked at creating a Test CD, with the below objectives:
- Ensure we could replicate the quality of available test CD with a burnt CD
- Have a complete set of test signals per the latest standard(s) and knowledge
- Ensure that all relevant and specific aspects of a CD Player can be reviewed (exception being calibrated errors)
- Adapt test signals to the CD Audio format (16bits/44.1kHz)
Please note that the current iteration is still a "prototype", but it's already very advanced. Few highlights of what makes this test CD unique:
- 0dBFS test tones improved as per the request of @MC_RME in this thread to account for 2's complement asymmetry
- The "regular" 1kHz test tone is 999.91Hz and has the benefit to self-dither
- Combined test signals to replicate the CBS CD-1 test CD and provide a lot more test tones
- Test files with dither of the type "Rectangle" (from Audacity) to reduce low-level noise
- High precision low level test-tones
- All single test tones files duplicated with shaped dither to allow analysis of DACs performance up to 19bits
To get the latest iteration of the test CD and the accompanying excel file with all details, please PM me (current version is 7.1.02).
!!!Warning!!!
Thanks to @restorer-john as I copy/paste his words:
This CD contains tones at frequencies and levels that can easily destroy expensive loudspeakers/tweeters and/or amplifiers.
It is not a CD to 'play' on your HiFi system. Use only with measurement instruments unless you are 100% confident you know what you are doing and particularly, what track is coming next. For example you may be trying to 'hear' a -80dB 1kHz sine by turning up the volume and then along comes a 20kHz 0dBFS sine which instantly smokes your Revel Salon 2 tweeters...
We accept no responsibility for any damage you cause to your gear. Entirely at your own risk. Be sensible.
How to burn the CD
If you use Windows, when burning the CD, order the files by file names and you'll get the same order as the one shown in the Excel. If you use Windows Media Player to burn the CD, don't forget to remove "Apply volume levelling across tracks" else it will introduce low level distortion:
The test CD can be used with many software including, and not limited to, REW, Multitone, ARTA, etc...
Quality Check of the burnt CD
It is easy to check that your burnt CD is of good quality. After the burn, simply rip the CD to a folder and perform a WAV analysis with REW or Multitone to confirm no digital transformation. Here below is an example from REW of a WAV analysis between the original file and a rip from the burnt CD (overlay):
The two traces are the same, and that means the burnt CD is flawless when compared to the original WAV files.
Required hardware
All you need are:
- A computer
- An input device like a Focusrite, Motu, RME, Cosmo, ... ideally with an ASIO driver
- Cables to connect the CD Player to the measuring interface
Since we are talking CD players, you do not need an Audio Precision as many modern audio devices have a much better resolution than CD Audio. That and the magic of FFT analyzis will help you go really far into measurements.
I personally use a Motu Ultralite MK5 and, as I said, many others can be used.
How to use the Test CD with REW - Initial steps
I've been using REW since the beginning and what follows is how to use the Test CD with REW. Most of what I wrote is valid with other Software such a Multitone, though. I think it's best you continue using your preferred measurement software, if any. Only the first two files of the Test CD are REW specific, not the others.
The below is not a substitute to the very complete REW user guide, but only complementing it in the context of testing CD players.
You can use an ASIO driver with REW, the Java driver has a limited bit depth (16 bits only), but as @staticV3 mentioned "With 64bit REW, you can use the Java EXCL output for stable, bit-exact, multi-device 32bit I/O, without installing any ASIO driver".
First, and even if you don't use the interface in a loopback (inserting the device under test in the loop out->in of the interface), I still recommend to calibrate it, especially for 44.1kHz inputs as it will reduce non-linearities around 20kHz as the filter ADC's interface is likely to influence signal. So this step will potentially improve phase and frequency response measurements (because REW forces the sweep to be measured at 44.1kHz). It's no big deal if you don't calibrate the interface right away as you can apply a "cal" file afterwards on saved measurements.
It's anyways less important to calibrate for 96kHz or 192kHz inputs as we don't need to analyze FR and phase at those high frequencies and your interface is likely to be flat from 20Hz to 20kHz with these sample rates.
So, use the below video to calibrate at 44.1kHz at least (again it's not a mandatory step):
Create REW Calibration file
Note that as opposed to what this person says, the calibration does not reduce distorsion, and only compensates FR and phase non-linearities.
There are additional videos to better understand how to play with REW (inserting a device to test in the loopback of the audio interface), but they don't apply to CD Players testing:
Play with REW, part 2
Play with REW, part 3
Play with REW, part 4
How to use information from the Excel
The below are print-screens (from REW, but it would be similar with other Software) to help you understand how to read the instructions of the Excel file.
First, the sampling rate for the measurement (eg: 24bits/96kHz):
The 24bits is of no real use because REW will always capture at this level, or 32bits if the interface allows.
The other excel columns indicate how the FFT for measurements (from RTA in REW):
Depending on the test file, measurements and sampling rate input, I recommend 128k, 256k or 512k FFT sizes. It allows the FFT bins to be small enough to get very good precision when computing THD and lowering random noise (via the FFT gain).
BH7 refers to Blackman-Harris 7 type of window which offers a good enough precision and facilitates measurement (less windowing errors). I always leave Max Overlap at 93.75%.
Now, let's go to the distorsion settings panel:
The AES notch is recommend as per the AES standard to add a notch around the fundamental, which in fact reduces the calculated noise, especially for those players that have correlated or random noise at the bottom of the fundamental (case in my example, by the way).
I always use a high pass and low pass at 20Hz and 20'000Hz respectively. Not everyone does the same. In an AP system, the user can chose too, but most people chose 20kHz or 22kHz as low pass. I personally think 20kHz makes more sense and is the recommendation of the AES17-2020. But that needs to be modified for some measurements, especially all IMD DFD at high frequencies (such as CCIF or AES).
Similar case for the computation of Intersample-overs where I simply unselect the "Use distortion low pass" . We indeed need to let REW calculate the THD up the 96kHz in that case. The excel file provides my recommended settings for each measurement.
Other REW only related information
In the RTA control panel, I always select the "Get fundamental from generator":
That way, when you perform a measurement from a CD Player, and even if REW's generator is not the one sending the signal, you can open it and select the measurement being performed from the CD Player. For instance here below a Dual Tone (SMPTE) is selected and so RTA window will adapt to it and search for the two tones and correctly calculate the IMD distortion:
No need to press the "Play" button at the bottom, RTA will adjust as per the settings of the generator.
Last but not least, the first two files of the Test CD are sweeps to measure FR, phase and THD vs Frequency in Left and Right Channels. To use them, click "Measure", drag and drop the wav file in REW window, click start and press play on the CD player (first file is for the left channel). Repeat the same for right channel.
I hope all of this is clear enough, but feel free to request assistance in this thread.
Final note:
Pay attention to fact that many files of the Test CD include multiple levels, for instance "C_Sine_999.91Hz_0dB_-1dB_-3dB_-6dB_NoDither_L.wav" is 2 minutes long, divided into 4 sequences of 30sec each of the same test tone played at 0dBFS, -1dBFS, -3dBFS and -6dBFS. So you can click the "Save Current" button of RTA to save them in the main window after REW has performed 32 averages:
Record your measurements
REW allows you to save measurements. I recommend you do so as it's really cool to overlay the results between multiple CD Players later on. I save the files names per track number, CD Player model and channel being measured. For instance, for the Marantz CD5400, the measurement with track #02 on left channel would be named "02 - CD5400 - L". I save them under a folder that is named per the version of the Test CD, so it's also easy to remember which file is what, using in the referenced excel file.
That way, I can also share my measurements with others and it makes it easier to compare between us.
Enjoy testing your CD player(s)!
--------
Flo
Last edited: