SONY PCM-R300 DAC measurements

[Rja4000]

I measured my
SONY PCM-R300 DAT recorder

The PCM-R300 was released around 1998.
(A review from the time may be found here)

This was the entry model in the SONY pro DAT recorders.
Obviously, this one is based on a consumer-grade DAT platform.

I purchased it second hand to replace my defective SONY DTC-670, and since I still had 60 DAT tapes or so.
It's much more advanced than the DTC-670.

I found it back in the box where it was stored for quite a long time.
It still works fine.


Test set-up:
Digital Loopback (48kHz) : RME ADI-2/4 Pro SE Optical Out > SONY PCM-R300 Optical in > SONY PCM-R300 Optical out> RME ADI-2/4 Pro SE Optical In

DAC (44.1kHz): RME ADI-2/4 Pro SE Optical Out > SONY PCM-R300 Optical in > SONY PCM-R300 Analog Out > RME ADI-2/4 Pro SE Analog In

ADC (44.1kHz): RME ADI-2/4 Pro SE Analog Out > SONY PCM-R300 Analog in > SONY PCM-R300 Optical Out > RME ADI-2/4 Pro SE Optical In

Software is Virtins MultiInstrument 3.9.8.1



Digital Loopback measurements

I was quite surprised to find that the digital loopback works in 24 bits, actually.
RME Bit-perfect test shows 24 bits bit-perfect confirmation message after transiting through the Recorder digitally.

SINAD

Jitter test 24 bits

Text-book perfect 24 bits Jitter results.
If something goes wrong, it's neither the SPDIF optical input nor the equivalent output.

Frequency response (44.1kHz)
EDIT: Now using a full bandwicth test signal

I also measured the digital loopback with a 44.1kHz 16 bits test signal

SINAD 44.1kHz 16 bits

This sets the maximum -perfect result- we could hope in 16 bits.

Mutlitone (16 bits)

82.7dB TD+N is the exact figure of the test file itself in 16 bits.

Recording and playback measurements

SINAD

I recorded 2 minutes of 997Hz test tone from Optical input on a random tape, then played it back through optical output.

Guess what ?
We get the exact same measurement than the loopback

Frequency response
EDIT: Now using a full bandwicth test signal

Same for Frequency response

Multitone

Multitone is identical too.

DAC measurements

Of course, the DAC works in 16 bits only.
The DAT format doesn't allow more anyway.


SINAD

Distortion comes from the DAC.
It's H2 and H3, but level is around -98dB max.
Noise is reasonable too.

Overall, the DAC doesn't degrade the quality of the signal.

Noise

Except for some Mains peak and harmonics (I'm in Belgium, so it's at 50Hz multiples), this is quite clean.
(NB: 0dBFS on the vertical scale is for the measuring ADC. DAC maximum would peak at -4.7dBFS. Noise figures are vs DAC maximum.)

J-Test (16 bits)
EDIT: Re-measured. Original test signal was wrong

Here, we see some low frequency Jitter. Quite low in level though.
( To compare with state of the art values for 16 bits Jitter, you may want to look into my review of the Marantz CD6000 OS KI )

Dynamic range (-60dBFS - 16 bits dithering)

Note that when feeding the DAC with 24 bits, we can go up to 97.4dB.
But 16 bits is more relevant, since the DAT format can't record more anyway.

Frequency response (44.1kHz)
EDIT: Re-measured using a Fast Frequency Response signal from this page.

Multitone

We get a bit more than 16 bits here

Intersample overload

The DAC doesn't stand any intersample overload
At 0.2dB overload, we are already above 1% THD in this torture test.

ADC Measurements

SINAD

Input Level can be pushed up to >3.3Vrms, without improvement.


Dynamic Range

Frequency response (44.1kHz)
EDIT: Now using a full bandwicth test signal

Multitone

Pushing the input level to 1.7Vrms, we get almost 16bits-perfect result

Noise
(level set for 2V recording.)

SBM measurements

Those SONY DAT recorder were coming with a technology called SBM, for "Super Bit Mapping".
This is a noise shaping technique which, according to Sony's claim, allowed to subjectively increase SNR or Dynamic range of a 16 bits recording.
You may find more explanations and links in posts below in this thread.

This technology only works at the ADC stage, when encoding to 16 bits format
Here, there is a switch to enable or disable it on the front panel.
All measurements above were made with it disabled. I was suggested to measure with it, so here we go.

Comparing with the above ADC noise, we see immediately the result: Noise is increased above14kHz or so, while it is slightly decreased below, and especially in the 3-4kHz area, where human hearing is more sensitive.

Comparing the 2 (zoomed 2x) makes it more obvious:

By the way, this is shaping the dithering noise.
It's not, like the old-ages Dolby NR, shaping the FR of the signal itself. So it has no impact on frequency response.

Nowadays, I'm pretty sure all mastering engineers are using a similar technique to shape dithering.
This old article from 2012 puts some light to this process.



Conclusion

Not too bad for a device from previous century.

DAT recording was the only reasonable way to record CD without too much loss by then.
When losses occured, it was brutal: Dropouts, complete loss of sound, due to errors...
This one as an "Errors count" display, that warned you somehow when the tape was bad and risk of loss increased.

DAT tapes are thin and rotating head quite fragile...
And then, of course, locating a track on the tape was SLOOOOOOW...
Those are the drawbacks of the DAT format, in general.

But soundwise, this one is a very competent device, IMO.
As far as 16 bits is enough.

 

[Kwesi]

Looks it is a really well made device. I love the old Sony ES series, their CD Players were SOTA in the 90s....
Thanks for the review and data!

 

[AnalogSteph]

Apparently this model is largely the same as the DTC-790. That is using Sony's CXD8505Q DAC which is fairly ubiquitous in their equipment of the time... I don't think there's a datasheet available. (I wouldn't rule out that it can take 18-bit input, dithering to that level may be worth a shot. The "mda Dither" Foobar2000 DSP plugin can do that, for example.) ADC is AK5340, possibly even the later B model with somewhat higher dynamic range.

When confronted with a 16-bit format, I would experiment with some shaped dither to pinpoint the DAC's analog noise floor with greater accuracy.

 

[Rja4000]

All measurements updated

 

[restorer-john]

All measurements updated

Excellent and very thorough review.

Like you, I have a number of Sony DAT uits, a few DTC-55es, a 57es and one of these r300 'pro' units I picked up for parts. Ironically, it's the only one still working* and is in the storeroom gathering dust.

*even after all the rebuilding of the head amplifier RF stages with new caps. Nightmare job if you ever need to know.

 

[AnalogSteph]

Frequency response (44.1kHz)

Hmm, that's rather a lot of periodic filter ripple for an ADC that's supposed to weigh in at <±0.01 dB, not sure what's up with that...

 

[Rja4000]

Hmm, that's rather a lot of periodic filter ripple for an ADC that's supposed to weigh in at <±0.01 dB, not sure what's up with that...

The level variations are very small.
Look at the vertical scale.
Let me check.

EDIT: Now fixed

 

[jwmitchell]

UPDATE:
Updated measurements.

I measured my
SONY PCM-R300 DAT recorder

The PCM-R300 was released around 1998.
(A review from the time may be found here)

This was the entry model in the SONY pro DAT recorders.
Obviously, this one is based on a consumer-grade DAT platform.

I purchased it second hand to replace my defective SONY DTC-670, and since I still had 60 DAT tapes or so.
It's much more advanced than the DTC-670.

I found it back in the box where it was stored for quite a long time.
It still works fine.

View attachment 303567

Test set-up:
Digital Loopback (48kHz) : RME ADI-2/4 Pro SE Optical Out > SONY PCM-R300 Optical in > SONY PCM-R300 Optical out> RME ADI-2/4 Pro SE Optical In

DAC (44.1kHz): RME ADI-2/4 Pro SE Optical Out > SONY PCM-R300 Optical in > SONY PCM-R300 Analog Out > RME ADI-2/4 Pro SE Analog In

ADC (44.1kHz): RME ADI-2/4 Pro SE Analog Out > SONY PCM-R300 Analog in > SONY PCM-R300 Optical Out > RME ADI-2/4 Pro SE Optical In

Software is Virtins MultiInstrument 3.9.8.1


Digital Loopback
I was quite surprised to find that the digital loopback works in 24 bits, actually.
RME Bit-perfect test shows 24 bits bit-perfect confirmation message after transiting through the Recorder digitally.

SINAD

View attachment 304221

Jitter test 24 bits

Text-book perfect 24 bits Jitter results.
If something goes wrong, it's neither the SPDIF optical input nor the equivalent output.

View attachment 304222


Frequency response (44.1kHz)

View attachment 304226


I also measured the digital loopback with a 44.1kHz 16 bits test signal

SINAD 44.1kHz 16 bits
This sets the maximum -perfect result- we could hope in 16 bits.

View attachment 304223


And Multitone

32 tones (16 bits)

View attachment 304227


Recording and playback

SINAD
I recorded 2 minutes of 997Hz test tone from Optical input on a random tape, then played it back through optical output.

Guess what ?
We get the exact same measurement than the loopback

View attachment 304224

Frequency response

Same for Frequency response
View attachment 304228

Multitone
Multitone is almost identical too.

View attachment 304230


DAC measurements

Of course, the DAC works in 16 bits only.
The DAT format doesn't allow more anyway.
That's exactly what I wanted, as a comparison point.


SINAD
Distortion comes from the DAC.
It's H2 and H3, but level is around -98dB max.
Noise is reasonable too.

Overall, the DAC doesn't degrade the quality of the signal.

View attachment 304231


Noise

Except for some Mains peak and harmonics (I'm in Belgium, so it's at 50Hz multiples), this is quite clean.
(NB: 0dBFS on the vertical scale is for the measuring ADC. DAC maximum would peak at -4.7dBFS. Noise figures are vs DAC maximum.)
View attachment 304234

J-Test (16 bits)
Here, we see some low frequency Jitter.

View attachment 304235

Dynamic range (-60dBFS - 16 bits dithering)
Note that when feeding the DAC with 24 bits, we can go up to 97.4dB.
But 16 bits is more relevant, since the DAT format can't record more anyway.

View attachment 304237

Frequency response (44.1kHz)

View attachment 304265

Multitone

View attachment 304287

Intersample overload

The DAC doesn't stand any intersample overload
At 0.2dB overload, we are already above 1% THD in this torture test.

View attachment 303608


ADC Measurements

SINAD

View attachment 303594

Input Level can be pushed up to >3.3Vrms, without improvement.

Dynamic Range

View attachment 303596

Frequency response (44.1kHz)

View attachment 304263

Multitone

View attachment 304264

Noise

I can't measure noise.
It seems it's gated.


Conclusion

Not too bad for a device from previous century.

DAT recording was the only reasonable way to record CD without too much loss by then.
When losses occured, it was brutal: Dropouts, complete loss of sound, due to errors...
This one as an "Errors count" display, that warned you somehow when the tape was bad and risk of loss increased.

DAT tapes are thin and rotating head quite fragile...
And then, of course, locating a track on the tape was SLOOOOOOW...

But soundwise, that's a very competent device, IMO.
As far as 16 bits is enough.

Thanks! Greet to see SONY did such an excellent job of engineering and production!

 

[morillon]

on occasion I will try to make some measurements on his brother "500".. there is a good chance that it will be quite comparable..
"Super-Bit-Mapping" ...perhaps explain many good mesureaments, no?

but maybe has already been done?

 

[Rja4000]

"Super-Bit-Mapping" ...perhaps explain many good mesureaments, no?

There is a switch on front panel to enable it.
It was disabled here.

Maybe I could try to measure its effect, if you're interested?

(For the curious, the best description I could find about what SBM does is here)

 

[morillon]

There is a switch on front panel to enable it.
It was disabled here.

View attachment 305871

Maybe I could try to measure its effect, if you're interested?

Yes why not.. ;-)

 

[morillon]

Super Bit Mapping (SBM), Sony’s CD mastering technique

 

[Rja4000]

I would experiment with some shaped dither to pinpoint the DAC's analog noise floor with greater accuracy.

How would you do that ?

 

[morillon]

How would you do that ?

but if you also have the courage to do a few subjective listenings to "difficult-acoustic" pieces of music with and without.. I would also be very interested in your feeling.. sony this placing a lot on this ground in the end ;-)

 

[Rja4000]

Yes why not.. ;-)

Done

 

[morillon]

the subject seems to have been much debated, but good the means of measurement 30 years later are so democratized..
(the funny opportunity to come back to it.. ;-)
but I also react because I have always appreciated the comfort of listening over the duration of quite a bit player from the 90s "20bits" and in particular on old UltraAnalog bases.. so the "20bits" aspect appeals to me a little.. ;-) )

 

[Rja4000]

the subject seems to have been much debated, but good the means of measurement 30 years later are so democratized..
(the funny opportunity to come back to it.. ;-)
but I also react because I have always appreciated the comfort of listening over the duration of quite a bit player from the 90s "20bits" and in particular on old UltraAnalog bases.. so the "20bits" aspect appeals to me a little.. ;-) )

To be frank, I think I never recorded to any DAT with an analog source. Always from digital.
And when I purchased this one, I was mostly done with recording any new material anyway.
So SBM was never appealing to me, as it works only with the ADC.

 

[Rja4000]

I also updated the DAC Jitter measurement.

 

[morillon]

To be frank, I think I never recorded to any DAT with an analog source. Always from digital.
And when I purchased this one, I was mostly done with recording any new material anyway.
So SBM was never appealing to me, as it works only with the ADC.

me the opposite .. rather for analog.

 

[AnalogSteph]

How would you do that ?

Much like what SBM does on the ADC side - you give the DAC a signal dithered to 16 bits with noise shaping. Ideally that is heavy enough to drop quantization noise in specific frequency ranges substantially below the DAC's analog noise floor. It's not really needed here since the unit will happily accept 24-bit data as you found (which is the good thing about AES3 and S/P-DIF, lesser formats are just being transmitted with zero padding), of which I'd guess the DAC will actually interpret 18 or 20 bits - plenty to demonstrate the analog noise floor either way.

BTW, the plain AK5340 is an 18-bit ADC listed with a 96 dB dynamic range, so that would seem to line up. I would rather leave SBM enabled, as without it dither seems a bit on the insufficient side. At least the "zoo" of low-level harmonics in the dynamic range test seems suspicious - I'd like to see that again with SBM.

Both ADC and DAC performance seem pretty good for 1994-5 consumer-level technology.

(Hmm, I never quite realized that the AK5340A could do up to 96 kHz. It must not have been the most popular variant at the time, given that the B was back down to 50 kHz and improved on dynamic range instead. I literally cannot think of an application for a double speed ADC of this class in 1995ish either.)