• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Yamaha CDX-393 Review (CD Player)

NTTY

Active Member
Forum Donor
Joined
Apr 27, 2018
Messages
296
Likes
1,390
Location
Switzerland
This is a review and detailed measurements of the Yamaha CDX-393 (MKI) CD stereo player.

Yam393-001.jpg


By the looks of it, you can guess this is one more old CD player, and you're right, it's from 1999.

My first review of the Onkyo C-733 here contains information about my measurements which I align with the AES standard (to few exceptions). It means that, over time, you can compare the devices I reviewed.


Yamaha CDX-393 - Presentation

Released in 1999, it was a low cost CD player (I think below 200$) featuring the minimum :) Back panel shows only RCA outputs:

Yam393-002.jpg



The MKII version of the same player offered a coax output, at least. So it's a very basic player and I only have to mention the below:
  • Conversion is from Panasonic, the MN66271RA, and I did test this one yet, reason for my interest.
  • Laser head is from Sony, the KSS-213C.
In reality, I was previously impressed by a DAC from JVC/Panasonic when I tested the JVC XL-Z335. So I wanted to check if this one would replicate the same good performances.

Let's have a quick look at the inside before going to measurements:

Yam393-003.jpg


The Panasonic DAC is below the card, so we don't see it here.

As with many older players, it's really fast to skip tracks back and forth. Jumping 30 tracks on my test CD is nearly instantaneous, and I love it. It's not as fast as a KKS-151 or KSS-190, but really close.


Yamaha CDX-393 Measurements

The Yamaha CDX-393 outputs 2.111Vrsm, and there was a channel imbalance of around 0.2dB. I could not verify the polarity (see below).

Here you go with the standard 1kHz sine @0dBFS (dithered) from my test CD (RCA out):

Yamaha CDX-393_1kHz_0dBFS_LR.jpg


Left and Right channels are shown and are nearly identical. THD sits at -96dB and so will clear CD Audio content.
This player has roughly 2dB more noise than the dithered one of the test CD. This is why ENOB loses nearly half a bit compared to the Onkyo C-733 for instance.

I think you spotted the deviation from 1'000Hz, as the software sees 1'001.98Hz. This is a clock issue, I think. It gets worse with the frequency. Also, it has some instabilities when it plays multiple test tones at high frequencies. On the triple tones test (Borberly 9kHz, 19kHz and 20kHz), I've seen the 20kHz tone varying from 19'984Hz to 20'020Hz.

I think this is why I failed to verify the polarity because the phase, in my sweep, was constantly jumping from inverting to non-inverting, so I can't tell what it is.

Anyways, let's continue and move to 1kHz @-6dBFS:

Yamaha CDX-393_1kHz_-6dBFS_R.jpg


THD improves with the lower signal as well as noise floor. Obviously this DAC is not so happy playing at 0dBFS. This is a good performance.

You probably saw previously a small and negligible power supple leakage (50Hz, 100Hz and 500Hz), but let's zoom:


Yamaha CDX-393_PowerSupply.jpg


This actually better than the Teac VRDS-20, eat this big boy :cool:

Bandwidth is flat:

Yamaha CDX-393_BW_LR.jpg


We see some ripple, this is unusual and surprising for an 8x oversampling filter. Probably some cost savings had to be achieved. And let's have a deeper look at the filtering behavior with the below overlay of white noise and AES IMD test (18kHz + 20kHz):

Yamaha CDX-393_Filter.jpg


So yes, the filter is slow and has a worst case attenuation of -50dB, which is weak. In 1982, the Marantz CD-73 was already doing better.
On that view, we see a slow increase of noise floor beyond 20kHz indicating a relatively agressive noise shaping in action in this DAC.

Multitone (1/10 decade) showed a happy CD player, not having issue to achieve 16bits+ free of distortion:

Yamaha CDX-393_MT_L.jpg


Let's go with the Jitter test to see if the clock issue shows here:

Yamaha CDX-393_JTest.jpg


The red trace is what is recorded on the test CD (taken from the digital output of Onkyo C-733, since this Yamaha does not have one).
Well, I thought it'd be worse than that. We see side bands indeed, but not really of concerns. After all, there might be a deviation but it could be a fixed one on this test. I'll wait for your comments and expertise. There are two resonators in this player, one for the DAC (VJ719800 16.9344MHz) and one for the micro-controller (VJ677200 4.19MHz).

Started with the Teac VRDS-20 review, and on your request + support to get it done (more here), I'm adding now an "intersample-overs" test which intends to identify the behavior of the digital filtering and DAC when it come to process near clipping signals. Because of the oversampling, there might be interpolated data that go above 0dBFS and would saturate (clip) the DAC and therefore the output. And this effect shows through distorsion (THD+N measurement up to 96kHz):

Intersample-overs tests
Bandwidth of the THD+N measurements is 20Hz - 96kHz
5512.5 Hz sine,
Peak = +0.69dBFS
7350 Hz sine,
Peak = +1.25dBFS
11025 Hz sine,
Peak = +3.0dBFS
Teac VRDS-20
-30.7dB​
-26.6dB​
-17.6dB​
Yamaha CD-1 (Non-Oversampling CD Player)
-79.6dB​
-35.3dB​
-78.1dB​
Yamaha CDX-393
-32.8dB
-27.6dB
-18.3dB
Onkyo C-733
-79.8dB​
-29.4dB​
-21.2dB​

I kept some references and will keep the same for other reviews, so you can quickly compare. The results of the Yamaha CDX-393 mean the oversampling filter has mo headroom. The Yamaha CD-1 shines here because it's old enough not to have an oversampling filter.

Here are other standard measurements:
  • Crosstalk : -114dB (@1kHz)
  • IMD AES : -85.6dB (18kHz + 20kHz 1:1 @-5dBFS)
Lasst but not least, my favorite measurement, the THD vs Frequency @-12dBFS. This one highlights the lack of linearity from older DACs which I like to review:

Yamaha CDX-393_THDvsFreq.jpg


The results are very good, as I've put the Onkyo C-733 for reference which is near best in class on that measurement. The Yamaha CDX-393 is let down, on more time, by the higher than usual noise level, which I attribute to that "agressive" noise shaping that I've seen in wider bandwidth analysis.

And since I don't have measurements to show about the digital outputs (because there are none), let me add the below one more to show the noise shaper behavior:

1728502270501.png


We see the shape to reject quantization noise outside audio band. I do not see that with other DAC as this normally happens below the 16bits resolution. Here it comes to disrupt audio band too, and by 2dB on top of the dither noise of my test CD. I suppose this is where this DAC shows its limitations (as with the oversampling filter), but I doubt this can be heard.


Conclusions

Thank you if you made it to that point!

To be honest, outside the frequency deviations, these are good performances for a really cheap player of 25 years ago! And I find it amusing to be faced with some sort of frequency wobble, it reminds me of good old times :p

I'm tempted to replace the resonators, to see if I can fix the issue, and maybe that would improve all other measurements?

This is good to remember that old players can face issues. Maybe the resonators have aged and that could be the reason. Buying an older player comes with risks. I think it's good to be reminded of that, especially when I see older gears sold at very high price without having been serviced (and of course no measurements to validate their good health).

I hope you enjoyed this review and I recorded nearly all measurements of my latest test CD (described here), so if you want more, feel free to ask!

Cheers.
 
Last edited:
Thanks for posting another review. Inexpensive basic single disk players are getting rare, I just checked the Onkyo website and it appears they have discontinued their only single disc player. That leaves the Pioneer PD-10AE and the Yamaha CD-S303, before the prices start to rise. About five years ago I purchased a used Denon, model DCD-F100 at my local thrift store for $18.00, in perfect condition. But since then, I has seen nothing but DVD and cheap Blu-Ray players.
 
I think this is why I failed to verify the polarity because the phase, in my sweep, was constantly jumping from inverting to non-inverting, so I can't tell what it is.

You can use an impulse or a assymetric clipped sine which is 100% reliable for polarity testing.
 
I'm tempted to replace the resonators, to see if I can fix the issue, and maybe that would improve all other measurements?

I would consider replacing the DAC ceramic resonator with a proper crystal (1M res=2x caps) . That model is the poverty pack special bottom of the line plastic front panel unit- no headphone amp populated, no remote control receiver, no TOSLINK out and the crystal + caps are missing.

CDX-493 has the "upgrade":

1728520338019.png
 
Last edited:
I would consider replacing the DAC ceramic resonator with a proper crystal (1M res=2x caps) . That model is the poverty pack special bottom of the line plastic front panel unit- no headphone amp populated, no remote control receiver, no TOSLINK out and the crystal + caps are missing.

CDX-493 has the "upgrade":

View attachment 397707
Indeed!

And here we go:

IMG_9376.jpeg
 
That's one way to do a proper, rational, tweak: to take one device to the manufacturer's higher standard.
 
Or go creative (less efficient but fun) ;)
 
I don't know who your preferred parts distributor is in .ch, but e.g. Reichelt carries a 16,9344-HC49U-S for a grand total of 0.23€ a piece. And Mouser has LFXTAL003366Bulk or HC-49/U-S16934400ABJB. Then you only need some 47 pF NP0 leaded ceramic caps.

Maybe you also have a scrap CD player or soundcard to salvage parts from.

As an aside, I would think that the ceramic resonator should also respond to capacitive loading (tweak until a 10, 11.025 or 12 kHz tone is spot-on), but unless you've got an assortment of small caps or a trimmer cap around...

BTW, if you want to see a '90s DAC with a crummy filter, try a TC9268P. By the time the Onkyo DX-7222 and later DX-7333 used them, they were kind of old hat to boot. Things were kind of on a downward slope for Onkyo for a while there.

BTW #2:
On that view, we see a slow increase of noise floor beyond 20kHz indicating a relatively agressive noise shaping in action in this DAC.
Actually, quite the contrary. It looks like fairly low order, flattened out with an analog lowpass higher up. This is "aggressive" (CS4272, based on CS4392):
index.php


Speaking of noise shaping, how are the noise-shaped test tones doing?
 
Last edited:
I don't know who your preferred parts distributor is in .ch, but e.g. Reichelt carries a 16,9344-HC49U-S for a grand total of 0.23€ a piece. And Mouser has LFXTAL003366Bulk or HC-49/U-S16934400ABJB. Then you only need some 47 pF NP0 leaded ceramic caps.
I'm tempted A friend of mine saw this message and will send me necessary parts, I must be lucky :)
Maybe you also have a scrap CD player or soundcard to salvage parts from.

As an aside, I would think that the ceramic resonator should also respond to capacitive loading (tweak until a 10, 11.025 or 12 kHz tone is spot-on), but unless you've got an assortment of small caps or a trimmer cap around...

BTW, if you want to see a '90s DAC with a crummy filter, try a TC9268P. By the time the Onkyo DX-7222 and later DX-7333 used them, they were kind of old hat to boot. Things were kind of on a downward slope for Onkyo for a while there.

BTW #2:

Actually, quite the contrary. It looks like fairly low order, flattened out with an analog lowpass higher up. This is "aggressive" (CS4272, based on CS4392):
index.php
Oh wow, I need to reset my threshold indeed :eek:
Speaking of noise shaping, how are the noise-shaped test tones doing?

Oh yes, of course. Here you go (with this Yamaha),1kHz @-20dBFS (TPDF):

1728577453625.png


Shaped Dither 997Hz @-20dBFS:

1728577527185.png


And same but with calculation of THD/THD+N restricted to 20Hz - 6kHz:

1728577601192.png


I think I recorded the same measurements for the Orpheus, but is this what you wanted to see?

EDIT: from the Orpheus:

Shaped dither 997Hz @-20dFS:

1728578041339.png


Same but with calculation of THD/THD+N restricted to 20Hz - 6kHz:

1728578137423.png


Cheers
 
Last edited:
Looks like shaped dither on the Yamaha gets you about 5 dB lower noise in-band, even if it makes the typical late-'90s "converter grass" even more obvious.

For the Orpheus it's like 9-10 dB, which is arguably less than I'd have expected, but it's beautifully clean at least.

BTW, a 997 Hz tone is basically "self-dithering" since - 997 being prime - it shares no divisors with 44100. The sampling process only hits the same values after 997 seconds or over 16 minutes, while for a 1 kHz tone it's just 10 seconds. Thus you basically don't need to add any extra dither to it. Obviously this needs reasonably high levels to work well, but if you generate -60 dBFS tones (0.000999 amplitude) straight to 16 bits and amplify by 50 dB, the difference is quite audible still. 997 Hz has a bit of a mechanical rasp to it but 1 kHz has a really annoying high-frequency buzz.
 
PS:
I was slightly wrong here.
The sampling process only hits the same values after 997 seconds or over 16 minutes, while for a 1 kHz tone it's just 10 seconds.
MarcelvdG has it correct:
Not taking into account dither, when you use 1000 Hz, the greatest common denominator with 48 kHz is 1 kHz and the greatest common denominator with 44.1 kHz is 100 Hz. That means you get a pattern that repeats 1000 or 100 times per second and that tests at most 48 or 441 of the possible codes (it could even be half of that when the phase relation between sample rate and sine wave is chosen unfavourably).
And indeed, when I am notching out 1 kHz and amplifying I am seeing a nice slew of 100 Hz harmonics and hearing a corresponding buzz.

In case of 997 Hz a 1 second period sounds about right, too, but it's much better noise already.

Now, can we actually do better than that still...? Yes, yes we can - the 999.91 Hz he suggested gives basically perfect-sounding white noise. So we want a prime multiple of a small fraction of 1 Hz, eh? In this case, that's 0.01 Hz, meaning a 100 second interval.
I had an algorithm search for all primes between 999999900 and 1000000099 and got lucky, it found 9.
Code:
999999929, 999999937, 1000000007, 1000000009, 1000000021, 1000000033, 1000000087, 1000000093, 1000000097
So we could use 1000.000007 Hz, yielding 0.000001 Hz or a 1e6 second period, that's 11 days 13 hours 46 minutes and 40 seconds. I think it's safe to say that test tones generally aren't that long. ;)
But boy does that make the tone annoying to notch out exactly. I'm guessing the frequency input for the notch is rounding / truncating the value somewhere. I think I'll stick with 1000.003 Hz for the time being (1000 s), or better yet 999.9991 Hz (10000 s). (999.99989 or 1000.00007 Hz did not yield the desired noise for some reason. Are we hitting float32 mantissa limits there? Seems so, even 999.9991 is actually 999.99908447265625. 1000.003 is better represented, as 1000.00299072265625, so I think this would be my frequency of choice. Hmm, on the other hand, if I take the exact values, the GCD of those and 44100 is actually 4 vs. 36, so 999.9991 Hz would do better after all. If you need test tones that run for up to 7.9 million years, that is.)
 
Last edited:
PS:
I was slightly wrong here.

MarcelvdG has it correct:

And indeed, when I am notching out 1 kHz and amplifying I am seeing a nice slew of 100 Hz harmonics and hearing a corresponding buzz.

In case of 997 Hz a 1 second period sounds about right, too, but it's much better noise already.

Now, can we actually do better than that still...? Yes, yes we can - the 999.91 Hz he suggested gives basically perfect-sounding white noise. So we want a prime multiple of a small fraction of 1 Hz, eh? In this case, that's 0.01 Hz, meaning a 100 second interval.

CBS CD-1:


1728717442872.png
 
@restorer-john Thanks and Interesting. I don’t have it but it should be easy to replicate:

1Reference, L & R, 0 dB, 1 kHz
2Left Separation, 0 dB, 1K, 125, 4K, 10K, 16K Hz
3Right Separation, 0 dB, 1K, 125, 4K, 10K, 16K Hz
4Signal/Noise, L & R, Infinity Zero w/o Emphasis
5Dynamic Range, L & R, 1 kHz, -60dB w/o Emphasis
6Frequency Response, L & R, 0 dB, 4, 8, 17, 31 Hz
7Frequency Response, L & R, 0 dB, 61, 127, 251, 499 Hz
8Frequency Response, L & R, 0 dB, 997, 1999, 4001, 7993 Hz
9Frequency Response, L & R, 0 dB, 10007, 12503, 16001, 17989 Hz
10Frequency Response, L & R, 0 dB, 19997 Hz (Also Used for Pitch Error)
11Sweep Frequency Response, 0 dB, 5 Hz / 22.05 kHz
12De-emphasis Error, L & R, 1k, 125, 4k, 10k, 16k Hz
13Intermodulation Distortion (SMPTE Twin Tone), L & R, 60 Hz + 7 kHz, 11 kHz + 12 kHz
14Wow & Flutter, L & R, 0 dB, 3150 Hz
15Access Time, L & R, 0 dB, 317 Hz
16Square Wave, L & R, 0 dB, 1002.27 Hz
17Impulse & Polarity Test, 0 dB, L & R
18Linearity, 997 Hz, L & R, 0 dB, -1, -3, -6, -10, -20, -30, -39.99, -49.97, -59.94, -70.31, -80.77, -90.31 dB
19Linearity with Dither, 997 Hz, L & R, -70.31, -80.77, -90.31, -100 dB
20Fade to Noise with Dither, L & R, -60 dB, 500 Hz
21Monotonicity, L & R, 1102.5 Hz, 10 LSB

I might add the wow & flutter one to my test CD for that Yamaha at least :)

I had undithered linearity test files in the past, but I removed them since dither is now largely used. I kept only -60dBFS (for SNR) and -90.31dBFS (3DC) undithered, of course.

I like a lot the idea to have added multiple tests in the same track. That lowers the total number of tracks and would even speed measurements.
 
Last edited:
So we could use 1000.000007 Hz

I've been using 997Hz for a long time now. I used to use 1001Hz, but that has its own issues. It is a factor of the old 44.056kHz sampling rate used with the original Sony PCM-3200 series. But your analysis software rounds anyway?

My QA403 analyzer has a "round to eliminate leakage" check box. Key in 997Hz and it will adjust based FFT size and bin width.

32kFFT 48kFS
1728719456903.png


128kFFT 48kFS
1728719634692.png

Uncheck rounding to eliminate leakage
1728719709970.png
 
PS:
I was slightly wrong here.

MarcelvdG has it correct:

And indeed, when I am notching out 1 kHz and amplifying I am seeing a nice slew of 100 Hz harmonics and hearing a corresponding buzz.

In case of 997 Hz a 1 second period sounds about right, too, but it's much better noise already.

Now, can we actually do better than that still...? Yes, yes we can - the 999.91 Hz he suggested gives basically perfect-sounding white noise. So we want a prime multiple of a small fraction of 1 Hz, eh? In this case, that's 0.01 Hz, meaning a 100 second interval.
At least REW does not show less than 2 decimals, so I can’t go below if I want/need to check that’s what is measured.
I had an algorithm search for all primes between 999999900 and 1000000099 and got lucky, it found 9.
Code:
999999929, 999999937, 1000000007, 1000000009, 1000000021, 1000000033, 1000000087, 1000000093, 1000000097
So we could use 1000.000007 Hz, yielding 0.000001 Hz or a 1e6 second period, that's 11 days 13 hours 46 minutes and 40 seconds. I think it's safe to say that test tones generally aren't that long. ;)
I confirm :p
But boy does that make the tone annoying to notch out exactly. I'm guessing the frequency input for the notch is rounding / truncating the value somewhere. I think I'll stick with 1000.003 Hz for the time being (1000 s), or better yet 999.9991 Hz (10000 s).
I can give it a try, at least to 999.91Hz and compare with 997Hz. And without dither, right?

Now, the limit here might be the Motu itself. Again with "only" 2Vrms at the input, I can’t get to its full resolution. Or maybe buy a cosmo scaler for that purpose.
 
I might add the wow & flutter one to my test CD for that Yamaha at least

Just do a 19997Hz track and take a spot frequency reading to determine the clock error.

W&F is a non-issue on CD. My standalone W&F meter doesn't move the needle at 0.001% on any player I tested (apart from variable pitch players! - lol)
 
I've been using 997Hz for a long time now. I used to use 1001Hz, but that has its own issues. It is a factor of the old 44.056kHz sampling rate used with the original Sony PCM-3200 series. But your analysis software rounds anyway?

My QA403 analyzer has a "round to eliminate leakage" check box. Key in 997Hz and it will adjust based FFT size and bin width.

32kFFT 48kFS
View attachment 398254

128kFFT 48kFS
View attachment 398256
Uncheck rounding to eliminate leakage
View attachment 398257
It looks like the "Lock frequency to RTA FFT" in REW:
When the Lock frequency to RTA FFT box is checked the generator frequency is adjusted to the nearest FFT bin centre for the current RTA FFT length (meaning the signal is periodic within the FFT length)

I was using that feature in my first reviews here, reason why the 1kHz measurement was showing 999.95Hz. I then stopped, but I could go back to it, just from 997Hz instead of 1kHz.

Obviously there’ll be a test CD version 7, which I’ll be tempted to rename ASR test CD v1.0 :)
 
Last edited:
@NTTY Have you done the Xtal mod yet? I'm really interested to see what difference it makes. It depends of course on how accurate (tolerance in +/-ppm) the crystal is. I suggest shock mounting it in a small piece foam rubber not soldered hard down to the PCB. Vibration from the CD mech itself can resonate the PCB I have found in the past. Plenty of high range CD players physically decoupled the digital filter IC/ D/A Xtal or secured it in a foam rubber bushing.
 
Not yet, a friend of mine sent me the parts but I’m abroad for the next 2 weeks, so that will have to wait. But sure, I’ll report about the differences, and thanks for the tips ;)
I’ll need the test tone you suggested for the before/after difference.
 
Back
Top Bottom