Accuphase DP-70 Review (CD Player)

[Bland_username42069]

If you’ve got the CD player, and the pre amp it would be foolish not to complete the trinity

 

[theREALdotnet]

At a small fraction of the cost, the SMSL PL200 of today offers way better measurements and a comparable (at least on the outside) build quality. One reason to consider here and now the Gold Age of Digital. Thank you for the excellent review, beautiful pics, and thoughtful comments, @NTTY !!

You could buy a couple dozen PL200s for the price of one DP-70. And to cover 40 years of use, you'd probably have to.

 

[NTTY]

If you’ve got the CD player, and the pre amp it would be foolish not to complete the trinity

Indeed, and I got introduced to Accuphase via an equally old big boy from them, the P-600. And so it required friends Few pics here.

 

[Bland_username42069]

Indeed, and I got introduced to Accuphase via an equally old big boy from them, the P-600. And so it required friends Few pics here.

We see lots of incredible Chinese chip amps and DACs on ASR but it’s always good to see some classic Japanese stereo and a proper matching stack. I’m looking forward to your review of the pre.

 

[SigmaR1]

6'37'' in the video, top left corner: a box (full of?) Sony KHM-230AAA laser optical pick-up mechanisms.

The optical pick-up used in Sony SCD-555ES, SCD-XA777ES and SCD-XA9000ES amongst others and unfortunately known to loose its ability to read the SACD layer of hybrid discs over time (although this very mechanism in one of my disc players still works flawlessly after 25 years of operation).

Scytales and his eagle eye!

 

[Nascimento]

Thank you very much for the excellent analysis of the equipment! If I lived in Switzerland, I would certainly send the DP-770 for your review!

 

[NTTY]

And I'd have accepted

 

[Snoopy]

You could buy a couple dozen PL200s for the price of one DP-70. And to cover 40 years of use, you'd probably have to.

With some of that type of smsl / topping/ ayama and the various „clone“ stuff sold on aliexpress etc I would be to worried to leave them plugged in at night or leave the house without unplugging the equipment.

I can’t be the only one that feels like that.. despite measurements and audible soundquality. There is a reason some of this stuff can be sold that cheap.

And who tells me 1000 smsl units will perform the same or even have all the same parts inside ?

No thanks! Made in china and manufactured in china by reputable brands is fine.. but I’m not gonna buy a power amplifier or anything that handles more than 5 volt from these cheap brands.

 

[martin900]

Thanks to this review I got myself a DP80L/DC81L CD transport + DAC combination. Haven't listened to it yet but it's built to last for a lifetime, for sure.
The KSS-190A is a work of art but so is the device as a whole.
The DC81L uses similar technology as the DP70 player but ups the resolution to 20Bit and has a fully separated, discrete output stage.

Just for reference, both units are multivoltage and accept 100, 120, 220 or 240V. There is a sticker inside telling how to connect the wires in correct order for certain mains voltage.
For 220V: BRN / WHT / YEL+GRY / RED / ORG
For 240V: BRN / WHT / YEL+GRY / ORG / RED
Same story for both the transport and the DAC.

 

[sheeple]

As opposed to what I see usually with very old conversion, the distortion does not increase. There's less noise relative to the carrier, meaning this is a quiet conversion (the SNR increased to 97.6dB).

I read 91.6dB?

 

[martin900]

@NTTY ,
Do you still have the DP-70? Can you check the following? How does the performance change after 1, 8 and 24h after power on? The resistors shouldn't drift much if at all with a 10-20*C temperature difference over idle but I'm just curious.

 

[NTTY]

There are small variations before it reaches temperature and then no changes indeed. I had it on for 24hours to check that, but I did not keep any measurements of before/during/after, so it’s all from memory. But I recall it requires a good 10min to get to stabilized performances.
PS: and yes I still have it, so I could take measurements.

 

[Scytales]

PS: and yes I still have it, so I could take measurements.

Perhaps I can suggest another test to perform, then?

 

[NTTY]

I read 91.6dB?

I meant DR, I always do that mistake, where we add the attenuation to the measured SNR.
Note that the "true" DR should be measured with a tone at -60dBr, per the AES standard, but what I meant is that the noise gets lower with lower test tone too.

 

[NTTY]

Perhaps I can suggest another test to perform, then?

Sure, shoot

I have another device on the bench to measure at the moment, but once done, I can make the Accuphase hot again.

 

[Scytales]

The test I think of has been proposed by Audio Precision in 1991. I just stumbled across the paper in which it is described. I do not think it is still relevant today (but you never know before you give it a try), but with old ladder DACs, such as the discrete ones in the Accuphase, it may still have value.

The test originally consisted of performing a 1/3 octave analogue band-pass analysis of the noise floor when playing a low frequency (41 Hz) dithered tone at decreasing levels from -50 to -100 dBFS after removal of the test tone, either by a notch filter or a high-pass filter.

The goal is to look for modulation in the noise floor in function of the levels of the test tone.

According to Audio Precision engineer R. C. Cabot, it had been established at Dolby Laboratories that modulation of the noise floor as small as 1 dB can be discernible, especially if localized in psycho-acoustically critical frequency band. Noise modulation can be observed with non-monotonic DACs.

The AES paper that describes the test: https://aes2.org/publications/elibrary-page/?id=5672

It is also described page 19 in this open access paper from the author of the test: https://www.researchgate.net/publication/241133948_Fundamentals_of_Modern_Audio_Measurement

I think that a modern variation of the test could be to accumulate on the same display the FFTs of the test tones to see if the noise floors overlap or not.

 

[NTTY]

Sure, I read the AES file and I think they used extreme examples of non linear DACs.

Before I burn a CD and torture the Accuphase, I wanted to test the concept with a poorly designed DAC that I have, something I bought on Ali for fun, and that uses 16xTDA1543 to deliver the worst performances I've seen, so far, from a DAC.

Following the AES document, this is the linearity test of the stupid thing (I went down to -110dBFS):

Linearity is lost as soon as at -75dBFS where it's already 1dB off.

And then, the below is an overlay of 3 tests, with a 41Hz sine (they recommend low frequency test signal that is not an integer sub-multiples of the sample rate) at -60dBFS, -75dBFS and -100dBFS, with triangular dither at 16bit:

As they show in the document, I started at 300Hz, so you don't see the fundamental. I also smoothed the noise curves as I see that's what they've done too. I finally zoomed (-120dB to -80dB) similar again to what they show. I used a lengthier FFT than them, though. Plot is at 10kHz, so you can read the differences in level.

And indeed, the noise profile changes, meaning that the deviation from linearity does not tell the full story.

My graph is a little hard to read, so I went for more smoothing (1/3 instead of 1/24), as I think that might have been what they did too:

We can now get a better viewing. So, indeed, at different levels of the low frequency fundamental, we see the dithered noise floor being modulated, with significant differences in level, up to 7dB depending on the frequency.

To verify no user/protocol error, I did all the above with a modern DAC and almost got a straight line:

Noise floor is limited by the dither. And it is the dither that is responsible for the modulation seen at the lowest frequencies here, not the non-linearity of the DAC which is perfectly linear down to -100dB with dither applied at 16bit, see graph below:

Conclusion

The above demonstrates the same as the AES paper:

1. Utilization of dithered noise allows for linearity testing up to -100dBFS with 16bits PCM (see last graph above)
2. Standard Linearity test does not tell the full story
3. There is an interaction between DAC non-linearities and dithered noise
4. Noise floor modulation due to DAC non-linearities could influence the audible performance

But I think, as with many AES papers, that this one goes too far and uses extreme cases, like I did above, to try to make a point. Coming from AudioPrecision, I can think there was a good intention to reveal psychoacoustic impact from DAC typical non-linearities and utilisation of dithered noise, which was spreading at the time (and is still a standard today, together with its evolved "shaped" version briefly mentioned in the AES document).

In a nutshell, I'm ok to test with the Accuphase as the guinea pig, but I'm afraid we'll loose some readers on this super-geeky test

 

[Scytales]

Neatly done !

Perhaps it is a somewhat extreme or geeky test, but at least it gives something to measure to, perhaps, discriminate otherwise difficult to take apart DAC designs on the sole basis of THD or THD+N.