• 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!

Teac VRDS-20 Review (CD Player)

Nice report, test and an even nicer device.
I also own a Teac VRDS-25x, an absolute tank among CD players. I still need a VRDS-20 for my pleasure, if only because of the look and construction.
Not even one of the Teak developers could explain to me at an event what Teak was thinking with the white keys.
 
Not even one of the Teak developers could explain to me at an event what Teak was thinking with the white keys.
It's probably a hangover from professional studio players (Teac made some). In a live broadcast studio DJs are talking and operating at the same time so buttons are often big, clear and obvious in half-light and there's often just play/stop, forward and back. Domestic players often had 20 tiny black buttons labelled in black on a black background.
 
Great review as always, thank you!
I'm with you, this unit was in m dream when i was a young boy and could only afford a humble Marantz CD40 (by the way still perfectly working with its immortal CDM-4, while i had to swap laser head to a much newer Denon DCD 1550AR...)
VRDS and Pioneer Stable Platter Mechanism were truly great pieces of engineering and workmanship.
 
Last edited:
Hello everyone,

First, thanks to all of you for your nice words. I'm happy to see that you enjoy these reviews, and so I shall continue.

@Roland68: I too have the VRDS-25x which I did not have time to measure or even open yet for inspection. I think it has adjustable MSB on the R2R DACs, it will be fun to play with that.

To come back to the VRDS-20, here are some measurements about the intersample-overs potential issue.

This is a wide band-view (up to 96kHz) of what is happening in that case (test file is 11'025 Hz 45° phase shift meaning ~+3dBFS over):

TeacVRDS20_InterSampleOvers_11025Hz.jpg


It indeed clips and we see a series of odd harmonics (square signal, because of the clipping), the first one at 33kHz being as high as -17.8dBr (that is 12% distorsion).

As you can see, there are no other negative impacts below 20kHz. And this is below the same measurement, but this time in log scale (instead of linear) and from 20Hz to 20kHz, to show that audio band is indeed preserved:

TeacVRDS20_InterSampleOvers_11025Hz_AudioBand.jpg


We see the usual low level power supply leakage, but nothing more.

On a scope perspective, here are the results, first if I extend the measurement to 96kHz:

TeacVRDS20_InterSampleOvers_11025Hz_96kHz.jpg


But if I restrict the measurement to audio band (max 22kHz), we have the below:

TeacVRDS20_InterSampleOvers_11025Hz_22.05kHz.jpg


Is this what you wanted/expected to see @restorer-john @pkane @AnalogSteph?

Big thanks to you @AnalogSteph for having me pushed so far. It's really cool to learn.

Enjoy your Sunday.

--------
Flo
 
For comparison you can refer to the Nielsen / Lund AES paper, who in an 80 kHz measurement bandwidth found distortion values ranging from -17 (sound familiar?) to -29 dBr, with the best value being generated by an early Discman of all things (a non-oversampling deal using Sony's CX20133 DAC... I think it's one of those jobs with one actual DAC running at 88.2 kHz which is then switched between L and R outputs periodically with external sample & hold). You should try this on the CD-73, the SAA7030 digital filter is known to have a good bit of headroom. By contrast, a bunch of otherwise good late-'80s/early-'90s parts apparently don't (e.g. Sony CXD1144B, CXD1244 or NPC SM5813AP).

(Note, they had a Marantz CD4000 in there which uses the SAA7378, a chip notorious for not being able to deliver bit-perfect digital output as it has about 0.5 dB of - crude - digital attenuation and hence correspondingly more headroom.)
 
Last edited:
Great review! I'm going to have to study a bit to understand some of the measurements.

I have a gold VRDS-20 that I bought used in Japan in the late 90s for $800. I had to replace the little belt that drives the mech that lifts the disc up to the platter, and I jumpered the power supply for 120V but otherwise it is original and still sounds great to my aging ears. One of these days I may recap it.

I just bought a SMSL SU-1 DAC ($80 via amazon) to use with a computer and decided to try it out with the VRDS-20. I tried it using the toslink and coax inputs to the DAC. I tested it with my marginal speakers and my Koss ESP950 headphones. DACs have probably improved a bit, but my hearing has diminished to where I am unable to hear any difference between them. One of the "benefits" of getting old- you can afford the good stuff but you can't hear the difference any more...
 
For comparison you can refer to the Nielsen / Lund AES paper, who in an 80 kHz measurement bandwidth found distortion values ranging from -17 (sound familiar?) to -29 dBr, with the best value being generated by an early Discman of all things (a non-oversampling deal using Sony's CX20133 DAC... I think it's one of those jobs with one actual DAC running at 88.2 kHz which is then switched between L and R outputs periodically with external sample & hold). You should try this on the CD-73, the SAA7030 digital filter is known to have a good bit of headroom. By contrast, a bunch of otherwise good late-'80s/early-'90s parts apparently don't (e.g. Sony CXD1144B, CXD1244 or NPC SM5813AP).

(Note, they had a Marantz CD4000 in there which uses the SAA7378, a chip notorious for not being able to deliver bit-perfect digital output as it has about 0.5 dB of - crude - digital attenuation and hence correspondingly more headroom.)
Thank you for the interesting source. From this paper, I see that all CD players suffered, except the one running without oversampling filter.

I already did the measurements of THD+N up to 96kHz, but since the distortion are high, it won't change much the end results.

Here is the table comparing the Teac VRDS-20 and the only non-oversampling CD player that I have at hand, and that is the Yamaha CD-1 from 1982, which I shall review soon.

Here are the results for the two players:

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
-79.6dB​
-35.3dB​
-78.1dB​

What we see from these results is that there's no headroom in the oversampling filter of the Teac VRDS-20.

Now, since this is due to clipping, they are the odd harmonics generating this distortion. Therefore, only one of them is is the audio band, and that's the one of 5512.5Hz which is found at around 16.5kHz, as we can see below (from the Teac):

TeacVRDS20_551.52Hz_0.69dBFS_L.jpg


Let's have a look at the same view from the Yamaha:

Yamaha-CD1_IntersampleOvers_5512.5_0.69_L.jpg


There's no large clipping here because of no-oversampling and because that player is solely relying on its analog filter. And so even if the signal would clip, it would be filtered.

There's one exception though, with the test tone of 7350 Hz sine (Peak = +1.25dBFS) as its third harmonic is at 22'050Hz. It means it will fall into the pass-band of many filters, analog or digital, especially if they are of the type "slow".

Let's have a look at what happens with the Teac VRDS-20 (red trace) compared to the Yamaha CD-1 (blue trace):

Yamaha-CD1_IntersampleOvers_7350_1.25_L.jpg


There you go, the Yamaha has too some difficulty to reduce the first odd harmonic and that's because we are in the pass-band of its analog filter. It explains the calculated THD+N of -35.3dB.

Let's have a final look at the 11'025 Hz sine (Peak = +3.0dBFS), Teac vs Yamaha:

Yamaha-CD1_IntersampleOvers_11025_3_L.jpg


Again we see that the analog filter of the Yamaha CD-1 does a good job, but at the same time, it does not suffer from the intersample-overs issue since it does not use oversampling. And by the way, still about the Yamaha, we see the H2 dominating in that view, meaning its distortion comes from "standard" harmonic distorsion and is not due to signal clipping as it the case with the Teac.

I will keep this message as a reference and update, if need be. All of that to say that in my future reviews, I will include measurements of the three values, and this is the reason why.

EDIT: I might continue to update the table, when I test additional Players. I don't know where to put it yet, so I'm editing here in the meantime (and also not to disrupt the text above):

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
-79.6dB​
-35.3dB​
-78.1dB​
Denon DCD-900NE
-34.2dB​
-30.4dB​
-19.1dB​
Denon DCD-SA1
-33.6dB​
-27.6dB​
-18.3dB​
Marantz CD-73
-49.7dB​
-31.2dB​
-27.7dB​
Onkyo C-733
-79.8dB​
-29.4dB​
-21.2dB​
Denon DCD-3560
-30.1dB​
-26.2dB​
-17.4dB​
Teac VRDS-25X
-29.4dB*​
-27.2dB*​
-24.8dB*​

*The clipping is worse than it looks and generates a lot of non-harmonic distortion.

--------
Flo
 
Last edited:
Looks like 11025 Hz is not a good choice for testing NOS players then, for the reason you outlined. 5512.5 and 7350 will have to do for them.
*The clipping of the filter also clips the subsequent analog stage, so the situation is worse than it looks.
Hmm. You sure that's an analog problem? This player sports an AD1893 ASRC, and ASRCs are kind of notorious for hard-clipping at 0 dBFS. They kind of were the latest fad around the turn of the millennium (as DACs supporting higher sample rates became available). The analog stage runs on +/-12 V and should have plenty of headroom.

Analog stages running on +/-5 V exist but are mostly a thing in DVD players. I've seen +/-7 V on a late-'90s Sony CDP-XB630.

If you can find a Sony player sporting either the CXD2560M or CXD2567M filters, those should be exemplary. Well, those that aren't running their output stage on +/-5 V anyway, which is a whole bunch of 'em. CDP-X339ES and up would be OK (shame about the - admittedly nice - unobtainium pickups); CDP-897 may be moddable if you insist (there is an unregulated but RC filtered +/-10 V available).
 
Last edited:
Nice report, test and an even nicer device.
I also own a Teac VRDS-25x, an absolute tank among CD players. I still need a VRDS-20 for my pleasure, if only because of the look and construction.
Not even one of the Teak developers could explain to me at an event what Teak was thinking with the white keys.
The essential keys were consistently backlit across their product lines (stuff like the awesome 8030S cassette player). Any other color does not work great with back lighting.
 
Looks like 11025 Hz is not a good choice for testing NOS players then, for the reason you outlined. 5512.5 and 7350 will have to do for them.

Hmm. You sure that's an analog problem? This player sports an AD1893 ASRC, and ASRCs are kind of notorious for hard-clipping at 0 dBFS. They kind of were the latest fad around the turn of the millennium (as DACs supporting higher sample rates became available). The analog stage runs on +/-12 V and should have plenty of headroom.
No I’m not sure but it looked like it (wrong assumption obviously :) ). When I first tested the 11’025Hz, the output made my interface clip because it was over 0dB at the input. So I reduced the gain and then it was ok, except for the VRDS-25X which continued to show the same type of clipping. I assumed it was its analog output, but yes if there’s an ASRC then it could explain. I did not investigate yet the VRDS-25X, I should have! I’ll update the text. Thank you ;)
 
Last edited:
This player sports an AD1893 ASRC, and ASRCs are kind of notorious for hard-clipping at 0 dBFS.

You're right, it’s mentioned in the datasheet :)

"Clipping
Under certain rare input conditions, it is possible for the AD1893 to produce a clipped output sample. This situation is best comprehended by employing the interpolation/decimation model. If two consecutive samples happened to have full-scale amplitudes (representing the peak of a full-scale sine wave, for example), the interpolated sample (or samples) between these two samples might have an amplitude greater than full scale."


And it is that ugly:

1727198918591.png


:eek:
 
Last edited:
Beautiful, thanks for putting in the effort. It is so interesting to see these measurements in so much more detail than when the products actually came out. I hope you make more of these.
 
Thank you @René - Acculution.com.
Yes, there will be more. I plan the next reviews to focus on CD Players with R2R DACs, as my initial objective was to understand how ancient tech performed. I've seen, so many times, people claiming that the best sounding DAC was the Philips TDA1541A, its "crowned" version being a further delight. So I had to tangibly know, and not only listen of course ;)
 
Hello Everyone,

This is a review and detailed measurements of the Teac VRDS-20 stereo CD player.

View attachment 393567

I already mentioned multiple times that I’m into older CD players (like the Marantz CD-73), this is one more proof.


Teac VRDS-20 - Presentation

This CD player made me dream when I was much younger. It was way out of my league at the time.
It was released in 1993 at a very high price, featuring a unique look in the Teac range, with the 4 adjustable pillars at each corner. It was available in black and gold.

It plays only the initial silver disc, no SACD, no USB input, no, no Bluetooth either, sorry :p No digital inputs, too bad, Teac could have made an effort here.

The elements of interest are:
  • The VRDS (Vibration-free Rigid Disc-clamping System) mechanism of course! This is the version CMK-3.2 with resin (or polymer, I’m not sure) molded “bridge” over the (aluminum) clamper that is supposed to reduce vibrations of the disc when rotating. Below that is the Sony KSS-151A magnetic head, one of the most reliable, fastest head. And 30 years later, it runs flawlessly.
  • D/A architecture (thanks to @AnalogSteph for the deep dive): A double mono conversion via two stereo Philips TDA1547 converter which was the first and high-end 1bit conversion of Philips, at the time. It is seconded by the Philips noise shaping (SAA7350) which is necessary for rejection of the noise generated by the decimation to 1bit. Preceding the SAA7350 is the oversampling and filtering function provided by an NPC SM5840 that runs at 8 times. If the SAA7350 could be used to convert directly, as it includes a 1bit DAC too, the couple SAA7350 + TDA1547 was the high-end conversion offer of Philips, and it was called DAC7.
  • Incredible mechanical construction: it does not stop at the VRDS as the entire device is built like crazy, with attention to every detail (I’ll come back to that later). As such it looks like an “Esoteric” before the time, and before it became a brand as such, outside of Japan.
Back panel has RCA and XLR analog outputs. We also find standard digital coax and Toslink, with a less standard St optical out:

View attachment 393568

There’s a window on the top of the player, the VRDS mech had to be seen.

View attachment 393569

As with many of these old players, and despite the complex and heavy VRDS, it is super-fast to skip one or multiple tracks back and forth (thanks to the Sony KSS-151A). With the big buttons on the front, very easy to read, it’s only a pleasure to use.

Before we go to measurements, let me guide you to the inside, because it’s worth a detour.
Below the aluminum top, we find … a second steel cover, and after removal of side panels with the 4 pillars, we find … a third copper cover. Really? :)

View attachment 393570

When finally inside, we are faced with something NASA-class built, except that it’s too heavy to go into a satellite :eek:

View attachment 393571

The player is divided into 4 sections:
  • VRDS mech in the middle
  • Power supply behind the drive
  • Servo card and digital out on the right
  • D/A conversion (with oversampling and noise shaping ICs) and analog out on the left
This is very neat, and attention to details goes as far as adding a copper pad on critical components as you can see above on the two TDA1547. I guess this is again NASA-habit to protect 0s and 1s from spatial radiations which could change their status? :cool:

I will spare you more details as I guess you want to know how this nuclear plant measures. But note that below the beautiful VRDS bridge (not decorum in this player, as it was with the T1 transport for instance), the aluminum clamp is so heavy that it requires a small disc brake with a caliper and respective pad. Why make it simple?


Teac VRDS-20 - Measurements (Analog out)

From now on, I will be consistent with my measurements as I described them on the Onkyo C-733 review. So over time, this will help comparing the devices I reviewed.

The Teac VRDS-20 outputs 1.9Vrsm from its RCA outputs and 3dB more from XLR (2.62Vrms). RCA and XLR showed the same performances, only a little more power supply noise from RCA.

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

View attachment 393573

Both channels are represented but only one gets evaluated in that window. Left channel is a little less performing (THD loses 4dB). That kind of difference happens especially when two different DACs running in mono mode are used.
THD sits at -104dB and so will clear CD Audio content. SINAD and ENOB are limited by the dithered noise of the test CD (roughly -93dB).

XLR and RCA are very close in performance, except that RCA has more power supply related noise.

It is to note that distortion stays very low even when digital signal goes down, which was not always the case with old DACs. As seen below, THD is still at -105dB at -6dBFS:

View attachment 393574

This is an overlay of left and right channels, from RCA and XLR out, so you get the best (XLR right) and the worse (RCA Left) at once. THD is well below -100dB which is very good for the time. Overall, the trace shows some “grass” of harmonic and non-harmonic low level distortions, but that’s not a surprise for a 1bit DAC and is well contained considering this was the first iteration of Philips (more than 30 years ago).

The datasheet of the TDA1547 specifies a best case THD of -88dB at -20dBFS (worst case is -84dB). And -88dB THD is very precisely what I get from the Teac. In other words, best possible implementation of that DAC here, thanks to mono mode too.

You probably already noticed, there is a little power supply leakage (50Hz and harmonics) which, even if without negative consequences at this very low level, I would have preferred not to see (1kHz @0dBFS, XLR out, left and right channels):

View attachment 393575

That said, the player being 30 years old could explain this.

Now, let's have a look at the bandwidth:

View attachment 393576

There’s a 0.1dB channel mismatch, and the service guide says it’s a pass up to 0.5dB, so I shall be very happy :)
We can see the oversampling filter ringing at the end of the frequency range. This is typical of oversampling filters lacking some power. And so, staying with the oversampling filter effect, let's have a look at wide band (up to 48kHz):

View attachment 393578

This an overlay of periodic white noise (red trace) and dual tones (18kHz and 20kHz). It shows the ringing of the filter. Out of band attenuation is only 60db (see the blue spikes of images from the test tones at 26.1kHz and 24.1kHz respectively). To be honest, I was disappointed to see that, even if better performing, it is similar to previous 4x oversampling filters of the era, especially the one of Philips. But I’ve seen better 8x filters of the same time both on ripple and attenuation.

So, was it intentional from Teac?

If one was to attribute a sound characteristic to such oversampling filters, and if it was believed to be the reason why the previous Philips converter TDA1541 and its associated filter SAA7220 were praised, then it could have been intentional. It is important to remember that the DAC is only the final part of the digital journey before reaching our amplifiers and speakers. DAC filters work well with matching ADC filters, and are ideal in that case.

At the time, the CD catalogue was already significant, most of it being built from Analog maters. The ADC digital filtering in use, to reject out of audio band frequencies, lacked power too and were likely to match what you see above (my guess). That’s the reason why I (like to) think this was intentional from Teac. Let me dream :) It could very well be a (less poetic) technical partnership to satisfy too, and so many other reasons.

Let's continue with the multitone test (1/20 decade):

View attachment 393579

I am sorry I forgot to adjust the FFT length which explains the increased noise floor of the trace at low frequencies. I’ll redo it when time allows. This is the “worst” left channel shown here (XLR and RCA), and we see that resolution is enough for CDA (minimum 16bits), except that spike at 150Hz from RCA. So nothing to worry about.

The Teac VRDS-20 showed reasonable amount of Jitter:

View attachment 393581

The red trace is what is recorded on the test CD (From the digital outputs), it can't be better. The blue trace is from the XLR ouput. We have again low level noise not of real concern (power supply related though).

Let’s have a look at an undithered 1kHz sine at -90.31dBFS. With 16bits, the signal should appear (on a scope) as the 3DC levels of the sign magnitude smallest digital signal:

View attachment 393583

This is a relatively good trace, and it’s disturbed by the low-level noise that I mentioned before.

Starting with this review, and on your request + support (more information here), I am adding an "intersample-overs" test. It intends to identify if the oversampling filter has sufficient headroom to process near clipping signals. Indeed, and because of the oversampling, there might be interpolated data that go above 0dBFS and would saturate (clip) the DAC and therefore the output. This this effect is highlighted with the measurements below, and revealed through THD+N measurements 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​
Onkyo C-733
-79.8dB​
-29.4dB​
-21.2dB​
Denon DCD-900NE
-34.2dB​
-30.4dB​
-19.1dB​

I kept some references in the table and will keep the same for other reviews, so you can quickly compare. The results of the Teac VRDS-20 mean the oversampling filter generates clipped signal when presented with specific sine at 0dBFS, so it has no headroom to process them. The Yamaha CD-1 shines here because it's old enough not to have an oversampling filter.

And here are some other measurements:
  • SNR : 96.4dB (1kHz @-60dBFS without dither)
  • Crosstalk : -116dB (@1kHz)
  • IMD AES : -94.2dB (18kHz+20kHz 1:1 @-5dBFS)
These are very good results for the time.

What else? Oh yes, of course, one of my favorite measurements, the THD vs Frequency at -12dBFS (THD only, over the first 5 harmonics). I already commented that I use it especially to test older R2R architectures as it shows their difficulty to be linear below full scale. Below are measurements of the right channel, overlaid with the previously tested (and near best-in-class) Denon DCD-900NE CD Player:

View attachment 393584

Very good! This was the real benefit of going 1bit as it improves linearity (compared to R2R DACs), at the expense of noise created (but that can be pushed beyond 20kHz with noise shaping).

All of these are very good measurements for the time, but I have to mention that JVC, with its proprietary DAC, did better (see the review of the JVC XL-Z335), even in low cost players.


Teac VRDS-20 - Measurements (Digital out)

Are you surprised if I tell you that I did not find any flaws in its digital output? I’ll keep it simple, with what I believe to be the most representative measurement of the digital output quality, and that is a 1kHz sine at -90.31dBFS which shows the 3DC levels of the smallest digital signal in 16bits sign magnitude representation:

View attachment 393585

Because this signal forces a regular change of all bits (due to 2’s complement representation around 0), any deviation from the binary content would immediately show here, with such a “fragile” signal. But if you think there are better ways to check the digital output, let me know.

Other than that, this is the standard dashboard of the 1kHz sine @0dBFS (no averaging because I did this measurement way before I decided to go for 32 averages, but it does not change anything here since there's no distortion):

View attachment 393589

This is what is recorded on the CD, so that is perfect digital out, as I see it.


Conclusion

This player and a few others made the “VRDS legend”, and today Esoteric and Teac continue to leverage this legacy. I could not find a mechanical advantage to all of this, and no issue as well. This is completely over-engineered, and I guess those who worked on this player had a lot of fun (I would have had).

No doubt this player knows how to respect what's on a CD. It is also a delight to use, a pleasure even before pressing the play button (which I won’t miss even without my glasses, thanks so much Teac!).

With an external modern DAC, it would be a perfect transport, as I hoped to be able to report. And it delivers.

I hope you enjoyed the review as much as I enjoyed writing it. Let me know how to improve and if you have questions. I can run additional measurements, if you want, but since this player joined me in the main listening room, you’ll have to be patient enough for me to extract it from its current location.

Thank you.

--------
Flo
Thank you so much for the great work
 
With an external modern DAC, it would be a perfect transport, as I hoped to be able to report. And it delivers.
Yeah, I had one for a while, "as players they make a great transports", to go with any decent external dac not theirs. (maybe mechanical engineers designed them)
Teac just didn't get it together when designing their dac sections for some reason, I've had a few go through, all the same.

Cheers George
 
Hello Everyone,

This is a review and detailed measurements of the Teac VRDS-20 stereo CD player.

View attachment 393567

I already mentioned multiple times that I’m into older CD players (like the Marantz CD-73), this is one more proof.


Teac VRDS-20 - Presentation

This CD player made me dream when I was much younger. It was way out of my league at the time.
It was released in 1993 at a very high price, featuring a unique look in the Teac range, with the 4 adjustable pillars at each corner. It was available in black and gold.

It plays only the initial silver disc, no SACD, no USB input, no, no Bluetooth either, sorry :p No digital inputs, too bad, Teac could have made an effort here.

The elements of interest are:
  • The VRDS (Vibration-free Rigid Disc-clamping System) mechanism of course! This is the version CMK-3.2 with resin (or polymer, I’m not sure) molded “bridge” over the (aluminum) clamper that is supposed to reduce vibrations of the disc when rotating. Below that is the Sony KSS-151A magnetic head, one of the most reliable, fastest head. And 30 years later, it runs flawlessly.
  • D/A architecture (thanks to @AnalogSteph for the deep dive): A double mono conversion via two stereo Philips TDA1547 converter which was the first and high-end 1bit conversion of Philips, at the time. It is seconded by the Philips noise shaping (SAA7350) which is necessary for rejection of the noise generated by the decimation to 1bit. Preceding the SAA7350 is the oversampling and filtering function provided by an NPC SM5840 that runs at 8 times. If the SAA7350 could be used to convert directly, as it includes a 1bit DAC too, the couple SAA7350 + TDA1547 was the high-end conversion offer of Philips, and it was called DAC7.
  • Incredible mechanical construction: it does not stop at the VRDS as the entire device is built like crazy, with attention to every detail (I’ll come back to that later). As such it looks like an “Esoteric” before the time, and before it became a brand as such, outside of Japan.
Back panel has RCA and XLR analog outputs. We also find standard digital coax and Toslink, with a less standard St optical out:

View attachment 393568

There’s a window on the top of the player, the VRDS mech had to be seen.

View attachment 393569

As with many of these old players, and despite the complex and heavy VRDS, it is super-fast to skip one or multiple tracks back and forth (thanks to the Sony KSS-151A). With the big buttons on the front, very easy to read, it’s only a pleasure to use.

Before we go to measurements, let me guide you to the inside, because it’s worth a detour.
Below the aluminum top, we find … a second steel cover, and after removal of side panels with the 4 pillars, we find … a third copper cover. Really? :)

View attachment 393570

When finally inside, we are faced with something NASA-class built, except that it’s too heavy to go into a satellite :eek:

View attachment 393571

The player is divided into 4 sections:
  • VRDS mech in the middle
  • Power supply behind the drive
  • Servo card and digital out on the right
  • D/A conversion (with oversampling and noise shaping ICs) and analog out on the left
This is very neat, and attention to details goes as far as adding a copper pad on critical components as you can see above on the two TDA1547. I guess this is again NASA-habit to protect 0s and 1s from spatial radiations which could change their status? :cool:

I will spare you more details as I guess you want to know how this nuclear plant measures. But note that below the beautiful VRDS bridge (not decorum in this player, as it was with the T1 transport for instance), the aluminum clamp is so heavy that it requires a small disc brake with a caliper and respective pad. Why make it simple?


Teac VRDS-20 - Measurements (Analog out)

From now on, I will be consistent with my measurements as I described them on the Onkyo C-733 review. So over time, this will help comparing the devices I reviewed.

The Teac VRDS-20 outputs 1.9Vrsm from its RCA outputs and 3dB more from XLR (2.62Vrms). RCA and XLR showed the same performances, only a little more power supply noise from RCA.

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

View attachment 393573

Both channels are represented but only one gets evaluated in that window. Left channel is a little less performing (THD loses 4dB). That kind of difference happens especially when two different DACs running in mono mode are used.
THD sits at -104dB and so will clear CD Audio content. SINAD and ENOB are limited by the dithered noise of the test CD (roughly -93dB).

XLR and RCA are very close in performance, except that RCA has more power supply related noise.

It is to note that distortion stays very low even when digital signal goes down, which was not always the case with old DACs. As seen below, THD is still at -105dB at -6dBFS:

View attachment 393574

This is an overlay of left and right channels, from RCA and XLR out, so you get the best (XLR right) and the worse (RCA Left) at once. THD is well below -100dB which is very good for the time. Overall, the trace shows some “grass” of harmonic and non-harmonic low level distortions, but that’s not a surprise for a 1bit DAC and is well contained considering this was the first iteration of Philips (more than 30 years ago).

The datasheet of the TDA1547 specifies a best case THD of -88dB at -20dBFS (worst case is -84dB). And -88dB THD is very precisely what I get from the Teac. In other words, best possible implementation of that DAC here, thanks to mono mode too.

You probably already noticed, there is a little power supply leakage (50Hz and harmonics) which, even if without negative consequences at this very low level, I would have preferred not to see (1kHz @0dBFS, XLR out, left and right channels):

View attachment 393575

That said, the player being 30 years old could explain this.

Now, let's have a look at the bandwidth:

View attachment 393576

There’s a 0.1dB channel mismatch, and the service guide says it’s a pass up to 0.5dB, so I shall be very happy :)
We can see the oversampling filter ringing at the end of the frequency range. This is typical of oversampling filters lacking some power. And so, staying with the oversampling filter effect, let's have a look at wide band (up to 48kHz):

View attachment 393578

This an overlay of periodic white noise (red trace) and dual tones (18kHz and 20kHz). It shows the ringing of the filter. Out of band attenuation is only 60db (see the blue spikes of images from the test tones at 26.1kHz and 24.1kHz respectively). To be honest, I was disappointed to see that, even if better performing, it is similar to previous 4x oversampling filters of the era, especially the one of Philips. But I’ve seen better 8x filters of the same time both on ripple and attenuation.

So, was it intentional from Teac?

If one was to attribute a sound characteristic to such oversampling filters, and if it was believed to be the reason why the previous Philips converter TDA1541 and its associated filter SAA7220 were praised, then it could have been intentional. It is important to remember that the DAC is only the final part of the digital journey before reaching our amplifiers and speakers. DAC filters work well with matching ADC filters, and are ideal in that case.

At the time, the CD catalogue was already significant, most of it being built from Analog maters. The ADC digital filtering in use, to reject out of audio band frequencies, lacked power too and were likely to match what you see above (my guess). That’s the reason why I (like to) think this was intentional from Teac. Let me dream :) It could very well be a (less poetic) technical partnership to satisfy too, and so many other reasons.

Let's continue with the multitone test (1/20 decade):

View attachment 393579

I am sorry I forgot to adjust the FFT length which explains the increased noise floor of the trace at low frequencies. I’ll redo it when time allows. This is the “worst” left channel shown here (XLR and RCA), and we see that resolution is enough for CDA (minimum 16bits), except that spike at 150Hz from RCA. So nothing to worry about.

The Teac VRDS-20 showed reasonable amount of Jitter:

View attachment 393581

The red trace is what is recorded on the test CD (From the digital outputs), it can't be better. The blue trace is from the XLR ouput. We have again low level noise not of real concern (power supply related though).

Let’s have a look at an undithered 1kHz sine at -90.31dBFS. With 16bits, the signal should appear (on a scope) as the 3DC levels of the sign magnitude smallest digital signal:

View attachment 393583

This is a relatively good trace, and it’s disturbed by the low-level noise that I mentioned before.

Starting with this review, and on your request + support (more information here), I am adding an "intersample-overs" test. It intends to identify if the oversampling filter has sufficient headroom to process near clipping signals. Indeed, and because of the oversampling, there might be interpolated data that go above 0dBFS and would saturate (clip) the DAC and therefore the output. This effect is highlighted with the measurements below, and revealed through THD+N measurements 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​
Onkyo C-733
-79.8dB​
-29.4dB​
-21.2dB​
Denon DCD-900NE
-34.2dB​
-30.4dB​
-19.1dB​

I kept some references in the table and will keep the same for other reviews, so you can quickly compare. The results of the Teac VRDS-20 mean the oversampling filter generates clipped signal when presented with specific sine at 0dBFS, so it has no headroom to process them. The Yamaha CD-1 shines here because it's old enough not to have an oversampling filter.

And here are some other measurements:
  • SNR : 96.4dB (1kHz @-60dBFS without dither)
  • Crosstalk : -116dB (@1kHz)
  • IMD AES : -94.2dB (18kHz+20kHz 1:1 @-5dBFS)
These are very good results for the time.

What else? Oh yes, of course, one of my favorite measurements, the THD vs Frequency at -12dBFS (THD only, over the first 5 harmonics). I already commented that I use it especially to test older R2R architectures as it shows their difficulty to be linear below full scale. Below are measurements of the right channel, overlaid with the previously tested (and near best-in-class) Denon DCD-900NE CD Player:

View attachment 393584

Very good! This was the real benefit of going 1bit as it improves linearity (compared to R2R DACs), at the expense of noise created (but that can be pushed beyond 20kHz with noise shaping).

All of these are very good measurements for the time, but I have to mention that JVC, with its proprietary DAC, did better (see the review of the JVC XL-Z335), even in low cost players.


Teac VRDS-20 - Measurements (Digital out)

Are you surprised if I tell you that I did not find any flaws in its digital output? I’ll keep it simple, with what I believe to be the most representative measurement of the digital output quality, and that is a 1kHz sine at -90.31dBFS which shows the 3DC levels of the smallest digital signal in 16bits sign magnitude representation:

View attachment 393585

Because this signal forces a regular change of all bits (due to 2’s complement representation around 0), any deviation from the binary content would immediately show here, with such a “fragile” signal. But if you think there are better ways to check the digital output, let me know.

Other than that, this is the standard dashboard of the 1kHz sine @0dBFS (no averaging because I did this measurement way before I decided to go for 32 averages, but it does not change anything here since there's no distortion):

View attachment 393589

This is what is recorded on the CD, so that is perfect digital out, as I see it.


Conclusion

This player and a few others made the “VRDS legend”, and today Esoteric and Teac continue to leverage this legacy. I could not find a mechanical advantage to all of this, and no issue as well. This is completely over-engineered, and I guess those who worked on this player had a lot of fun (I would have had).

No doubt this player knows how to respect what's on a CD. It is also a delight to use, a pleasure even before pressing the play button (which I won’t miss even without my glasses, thanks so much Teac!).

With an external modern DAC, it would be a perfect transport, as I hoped to be able to report. And it delivers.

I hope you enjoyed the review as much as I enjoyed writing it. Let me know how to improve and if you have questions. I can run additional measurements, if you want, but since this player joined me in the main listening room, you’ll have to be patient enough for me to extract it from its current location.

Thank you.

--------
Flo
Very nice. Another great review. Good job! I also share a relative passion for CD players and I'll try and measure as many as possible myself, using your test CD and your input :)
 
Back
Top Bottom