855 digital sources from 1992 to 2024 ranked by THD

[NTTY]

Thanks for all the information, very useful

 

[Scytales]

I update the files this September 18th, 2024, with 10 new devices and 1 more test of an already tested device (see first post).

Tests in 16 bits :

Berendsen CDP-1 (2nd sample)
Exposure XXII
Lua Cantilena II
Musical Fidelity X-Ray V3
Naim CD 5i (not to be confused with CD 5i already tested)

Tests in 24 bits :

Cambridge AXN10
Cambridge CXN100
Cyrus Stream XP
Fiio R9
NAD T533
Volumio Primo

I have also identified some signal processing peculiarities previously undocumented and corrected a few remaining errors or typos.

The files have been uploaded in the first post.

Thanks for reading.

 

[Scytales]

I update the files this October 15th, 2024, with 1 new device.

Tests in 24 bits :

Rotel RSP-1098

I have also corrected a few typos.

The files have been uploaded in the first post.

Thanks for reading.

 

[Scytales]

I update the files this November 7th, 2024, with 12 new devices and 2 more tests of already tested devices (see first post).

Tests in 16 bits :

Accustic Arts DAC I
AVM Evolution CD 2
Creek CD43 MK2 (2nd sample)
JVC XLR 5000
Kenwood DPF J6030
Marantz CD 6000 OSE KI
Marantz DR 6000
Marantz DR 6050
Perpetual Technologies P-1A + P-3A
Pioneer PDR 609
Pioneer PDR W839
Rega Planet (model 2000)
Sony DVP-S9000ES (already tested in SA-CD mode, now tested in CD mode)
Theta Casablanca II

I have also given new information and corrected a few remaining errors or typos.

The files have been uploaded in the first post.

Thanks for reading.

 

[NTTY]

Hi Scytales,

Would you know what Test CD(s) was(were) used for measurements of CD players? Maybe CBS CD-1, or Sony YEDS or Denon Audio test, ...?

Thanks

 

[Scytales]

I don't remember I ever have read an article about a particular test CD employed par the German lab. Considering the fact that is was stated in some reviews that the test frequency was somewhat 315, 330 or 333 Hz, it is possible they have engraved their own test CD for whatever reasons, because for my part I don't know any test CD with all those frequencies.

 

[NTTY]

Thanks, indeed I don't know one with these test tones. The CBS CD-1 has several and anyways it does not really change the end results. I just checked with the Teac VRDS 25X that I am currently reviewing and with 3 different tones from the CBS CD-1 (251Hz, 499Hz and 997Hz), and the results are the same:

Cheers

Flo

 

[Scytales]

I just had a look on the FFT graph with a test tone a bit above 300 Hz in the VRDS 25X review published in the January 1999 issue of Stereoplay out of my library and I find that your measurements at 251 and 499 Hz correlate well with the one made by the Germans with an Audio Precision. Low even order harmonics levels are bit lower in the printed review, but the overall pattern of distortion products above the low order ones are about the same to me. I have calculated THD over the first four harmonics out of your first measurement to be 0,00147 % by the rounding rules I have followed to make the survey, which is not far from the 0,00101 % I have calculated from the measurement in the printed review (with all the accuracy issues due to reading small FFT graphs by eye). I wouldn't be surprised if that particular VRDS 25X tested in Stereoplay was a "Golden Sample" but I find the difference not shocking at all, especially considering the age of your unit by now.

Are your measurements taken at the RCA output ?

Keep the good work going. I am eager to see your further reviews.

 

[NTTY]

Happy to help!

The above were from RCA outputs and I edited the post as I realized I measured with a smaller than usual FFT size.

The below are from XLR:

They are basically the same except that the XLR outputs better reject the power supply noise (no surprise).

 

[Scytales]

I update the files this November 9th, 2024, with 1 last device.

Test in 24 bits : Linn Klimax DSM/1

That made a nice round 850 number for the title.

I have also made a correction. The Linn Majik, Akurate and Klimax DS Dynamik streamers were wrongly put in their own category. After I have re-read the actual test, they have to be put in an existing category. That error is now corrected.

The files have been uploaded in the first post.

Thanks for reading.

 

[NTTY]

I wouldn't be surprised if that particular VRDS 25X tested in Stereoplay was a "Golden Sample" but I find the difference not shocking at all, especially considering the age of your unit by now.

Hi Scytales,

Maybe not golden but maybe care was given to MSB adjust of the Analog Devices DACs (4xAD1862). I played with the trimming pots as per the recommendations of the Teac service manual, and I was able to slightly improve things (RCA outs):

Funny thing is that the Teac service manual says to use a 0dBFS test tone, while the Analog Devices datasheet says:

Analog Devices recommends that this adjustment be performed with–60 dB signal amplitudes or lower. Minor performance improvement is achieved with larger signal amplitudes such as–20 dB. Almost no improvement is possible when this adjustment is performed with 0 dB signal amplitudes.

I'll give a chance to -60dBFS adjustments later. EDIT: I tried with -40dBFS, -50dBFS and -60dBFS without dither, with rectangular and Shaped dither, but all distortion values are buried into the noise floor of the test files (CD), so there's nothing to adjust I guess it is because Analog Devices were thinking of adjusting with 20bits input data, which is not possible with the Teac.

 

[NTTY]

I just had a look on the FFT graph with a test tone a bit above 300 Hz in the VRDS 25X review published in the January 1999 issue of Stereoplay.

I forgot, since the VRDS-25X has a digital input, I can also easily use a test tone of your choice. I suppose they used a prime number above 300Hz, there are many (307, 311, 313, 317, 331, 337, 347, 349, 353, 359, 367, 373, 379, 383, 389, 397), and if you can guesstimate which one it is, I can replicate the exact same test.

In the meantime, this is 331Hz @0dBFS with 16bits/44.1kHz, no dither, RCA out:

I've put the plot on H3 (-103.3dBr) to facilitate your calculations. Note that in this dashboard, I restricted the THD calculation to 5 harmonics, as per your rule. EDIT: Wrong, it was on 50 harmonics, not 5...

Cheers

 

[Scytales]

Incredibly, the first four harmonics on your latest measurement are now much closer to those I have read on the printed review : 2nd : -109 dB / 3rd : -102 dB / 4th : -117 dB / 5th: -106 dB! The first two lower even-order harmonics are now below the level of the two first odd-order harmonics, exactly as shown in the FFT published in the Stereoplay magazine. So, you probably got it right : the MSB of the DACs of the VRDS25X that has been tested by the Germans must have been tweaked!

Computation from the levels from your last graph give 0.00098 % (about -100.2 dB) as per the rounding rules I have followed. I am a bit surprised that your measurement results and the crude calculation I have made converge that much. I am glad, because this fact tends to confirm the methodology that I followed. So I didn't do all this cumbersome work in vain, after all.

After taking out my millimeter ruler and a magnifying glass, I guesstimate that the testing frequency in the Stereoplay review is closer to the 313 prime number your mentioned above.

 

[NTTY]

After taking out my millimeter ruler and a magnifying glass, I guesstimate that the testing frequency in the Stereoplay review is closer to the 313 prime number your mentioned above.

Great, here you go with 313Hz (no dither, RCA out):

And I'm sorry, I made a mistake previously, I did not restrict measurement to 5 harmonics but to 50 This one is really restricted to 5 harmonics. Your calculation is spot on, well done

Here are the measured harmonics on the above graph:

• H2: -113.9dBr
• H3: -102.9dBr
• H4: -121.6dBr
• H5: -104.7dBr

They are extremely close to what you’ve seen and reported from that previous review of Stereoplay.

The only little difference (improvement) here might be in the “fine” trimming of the MSB (there are two pots to adjust per channel, one is for the “fine adjustment”), which needs a little more care and precise “live” monitoring of the FFT to get the best out of it. So, without a software able to run near real time FFT, it’s too cumbersome.

Cheers

Flo

 

[Scytales]

I have to re-upload the files (22a version) because I misidentified one version of the Linn Akurate DS streamer (it actually is the second version Akurate DS/1) and had misplaced it in the wrong section. Those pesky Linn products having the habit to be regularly udpdated are not easy to follow.

 

[restorer-john]

Analog Devices recommends that this adjustment be performed with–60 dB signal amplitudes or lower. Minor performance improvement is achieved with larger signal amplitudes such as–20 dB. Almost no improvement is possible when this adjustment is performed with 0 dB signal amplitudes.

I'll give a chance to -60dBFS adjustments later. EDIT: I tried with -40dBFS, -50dBFS and -60dBFS without dither, with rectangular and Shaped dither, but all distortion values are buried into the noise floor of the test files (CD), so there's nothing to adjust

-60dB is the standard way to adjust MSB trimpots on multi-bit players and you view the waveform on a 'scope to eliminate zero-cross distortions. You can never eliminate it, but you can make the waveform 'look' better.

With single bit players, it will descend into system noise/'scope noise.

 

[Scytales]

-60dB is the standard way to adjust MSB trimpots on multi-bit players and you view the waveform on a 'scope to eliminate zero-cross distortions. You can never eliminate it, but you can make the waveform 'look' better.

This kind of distortion seems to be mostly asymmetric, because even order harmonics decreased (at least on this TEAC VRDS25X) whereas odd-order harmonics remains roughly at the same level. Am I correct?

 

[restorer-john]

This kind of distortion seems to be mostly asymmetric, because even order harmonics decreased (at least on this TEAC VRDS25X) whereas odd-order harmonics remains at the same level. Am I correct?

Yes, mostly. You may be able to see it on a scope. @NTTY you can see it on your new 'scope when it arrives.

Some high end multibit vintage CD players have multiple trimpots for the 4 MSBs for each channel and must be done in order.

 

[NTTY]

-60dB is the standard way to adjust MSB trimpots on multi-bit players and you view the waveform on a 'scope to eliminate zero-cross distortions. You can never eliminate it, but you can make the waveform 'look' better.

With single bit players, it will descend into system noise/'scope noise.

Thanks John, I'll try this way too.

 

[Scytales]

I update the files this January 12th, 2025, with 5 new devices, new information and some corrections (see first post).

Test in 16 bits:

Marantz CD 6007

Tests in 24 bits:

Cambridge EXN100
Eversolo DMP-A8
iFi Audio Neo iDSD 2
Wiim Ultra

I have also made some corrections:

• The Denon DCD-1500AE uses a Burr Brown DSD1791 DAC chip, not a PCM1791.
• The Harman Kardon DVD 47 uses one Cirrus Logic CS4382 multichannel DAC and not multiple CS4392 stereo DACs.
• The Marantz DV6400 has been tested on its "high definition" multichannel output which uses Burr Brown DSD1702 DACs and not on its stereo output which uses Burr Brown PCM1751.
• The correct designation of the 8 channels DAC chip the Philips DVP-9000S uses is Cirrus Logic CS4382, not CS4362. The theory of operation of the player has been corrected.

The files have been uploaded in the first post.

Thanks for reading.