Marantz CD67 mk2 measurements

[horias2000]

I see that testing of CD Players on ASR has gone up due to @NTTY interest in the topic. Therefore, with his help and the help of others around here, I will post some more CD player measurements. This time, of the CD67mk2. At the moment, I have two players that I listen to, the CD52mk2-SE (https://www.audiosciencereview.com/...cial-edition-measurements.57703/#post-2110235) and the CD67mk2 (non-SE).

I bought both of them a while ago and I couldn't decide which one to keep. Hopefully the results will make it easier.

The CD67mk2 was released in mid 90s and it uses the SM5872 DAC chip. This is a two channel 16bit with 8x oversampling FIR filter inside, made by NPC. The transport is a Philips CDM12.1 and I had some issues with it. I even bought a replacement but that didn't work very good either so I reverted to the original one.

Here are some pics of the device (inside and outside)

Inside it's pretty well made and it has the Marantz "HDAM" analog output stage. As far as I know, this is an amplifier build with discrete components. Besides the marketing fluff, I do not think this has and real advantages on using a high quality opamp.

Now onto the measurements with the 1kHz test tone:

THD is pretty good at alomost -98dB but the noise is average, giving us a THD+N of -90dB. Not bad but not great either.

Next is the same 1kHz test tone but with extended frequency range, up to 48Khz to see if there are traces of noise shaping:

And now the same 1kHz test tone but at -3dB:

THD improves quite a bit and THD+N does not decrease too much. This is pretty good.

Multitone:

This actually looks very good, with 10dB of distortion free range.

Now onto filtering:

Jitter:

This looks pretty standard.

And now the two tones for IMD:

Again, pretty good performance as far as I can tell. Nothing special but not too bad either.

Now the "inter-sample overs"

I've made a table with the other CD player that I tested so that we can easily compare:

CD player​	5512Hz​	7350Hz​	11025Hz​
Marantz CD52mk2 - SE	-34dB​	-27.6dB​	-16.2dB​
Marantz CD67mk2	-37.6dB​	-29.1dB​	-19.3dB​

The two players are pretty similar.

Lastly, I've also performed a measurement with a 997Hz test tone (shaped dither):

If I'm reading this correctly, there is some noise shaping that affects the audio band, but this is at around -120dB so it isn't audible.

Conclusion:

I was hoping to have a clear winner between the CD52mk2-SE and the CD67mk2 but this is not the case. Performance wise, the two units seem to be very close, with the CD52mk2-SE winning on THD+N.
Overall, I've enjoyed using the CD67mk2 and I can't say I could hear any differences between the CD52 and the CD67. Using them with an external DAC makes the differences even smaller. The only real issue is the CDM12.1 transport that doesn't seem as reliable as the CDM4/19 that is inside the CD52.

I definitely enjoyed measuring my CD players and I will continue to measure other units as well. I'll borrow all the players from my friend

 

[NTTY]

Thanks for the review!

I did not know that DAC so that’s a discovery. There are lots of low level distortion, and some seem to be correlated. I’ve never seen the 180Hz and 205Hz (or so) on 1kHz test tone.

About the intersamples over, you can remove the low pass filter in the distorsion panel of RTA. It will compute the THD+N up to half the sampling frequency (48kHz in your case).

The 997Hz with shaped noise adds this low level high frequency noise on purpose. And with this player, it lowers the noise floor a little up to 5kHz, by roughly 5dB.

Cheers

 

[AnalogSteph]

Now onto filtering:

With a vertical scale of 60 dB, virtually anything will look dead-flat. The periodic ripple that we're interested in is often going to be 0.04 to 0.1 dB peak-peak, so a zoomed-in view by a factor of about 100 would be appreciated.

 

[restorer-john]

The periodic ripple that we're interested in is often going to be 0.04 to 0.1 dB peak-peak

And the audible significance of that level of ripple is what exactly?

It's like pixel peeping.

 

[restorer-john]

@horias2000 Check the crystal for the D/A converter. Marantz use a bond glue that goes corrosive when attaching the rubber dampers. It can eat under the crystal and cause the player to fail.

Marantz's HDAM...

 

[AnalogSteph]

And the audible significance of that level of ripple is what exactly?

It's like pixel peeping.

If these were analog LC or IIR filters, that would likely be the case and you could probably get away with +/-0.2 dB ripple easily. With FIR filters (as all digital filters prior to the early 2000s would have been), it's not quite that simple...

https://www.nanophon.com/audio/antialia.pdf

The amplitude and frequency of periodic ripple determine how critical it is as these factors determine the magnitude of the pre-echo and how likely it is to be masked.

This here seemingly was enough to cause subjectively wonky treble in an IDT 92HD93 chip, as all other measurements checked out:

(It's the difference between 44.1k straight and it being upsampled to 192k first, generated by subtracting both measurements.)

The audible problem was basically gone by 96k, with 192k possibly being a hair better still. Other aspects of measured performance stayed pretty much the same regardless of sample rate.

44k upsampled to 192k then looks like this:

Smoooth.

It sounds, well, normal.

@horias2000 Check the crystal for the D/A converter. Marantz use a bond glue that goes corrosive when attaching the rubber dampers. It can eat under the crystal and cause the player to fail.

Oh no, killer glue. Switch-mode power supplies commonly use a white kind of silicone for things like this (presumably acid-free).

Marantz's HDAM...

It's a folded cascode opamp (buffer), and not a particularly well-performing one either (these things tend to lend themselves to integration but deliver disappointing results if discrete). You're literally better off bypassing the stupid thing (remove RH24, install R620).

 

[restorer-john]

You're literally better off bypassing the stupid thing (remove RH24, install R620).

Pull the whole thing out and use those rare Toshiba FETs for something useful.

 

[restorer-john]

Switch-mode power supplies commonly use a white kind of silicone for things like this (presumably acid-free).

Maybe silicone that releases extra acetic acid for good measure. Kind of a time fuse for the PCB. Built in failure point. LOL.

 

[horias2000]

Thanks for the review!

I did not know that DAC so that’s a discovery. There are lots of low level distortion, and some seem to be correlated. I’ve never seen the 180Hz and 205Hz (or so) on 1kHz test tone.

About the intersamples over, you can remove the low pass filter in the distorsion panel of RTA. It will compute the THD+N up to half the sampling frequency (48kHz in your case).

The 997Hz with shaped noise adds this low level high frequency noise on purpose. And with this player, it lowers the noise floor a little up to 5kHz, by roughly 5dB.

Cheers

I agree that the 180Hz and 205Hz spikes a bit weird but these are present on most of the measurements. These are low enough not to cause any concerns but still, interesting seeing them there. I do not have a good explanation for these.

 

[horias2000]

With a vertical scale of 60 dB, virtually anything will look dead-flat. The periodic ripple that we're interested in is often going to be 0.04 to 0.1 dB peak-peak, so a zoomed-in view by a factor of about 100 would be appreciated.

I will modify the graph and post it again a bit later, when I get home.

 

[jeroboam]

Pull the whole thing out and use those rare Toshiba FETs for something useful.

Does your and AnalogSteph's opinion apply to the current implementation of HDAM in the Marantz lineup. They are touting it as a desirable feature.

 

[NTTY]

It's a folded cascode opamp (buffer), and not a particularly well-performing one either (these things tend to lend themselves to integration but deliver disappointing results if discrete). You're literally better off bypassing the stupid thing (remove RH24, install R620).

@horias2000 ie. downgrade to CD-57, it might be better

 

[horias2000]

@horias2000 ie. downgrade to CD-57, it might be better

I could just bypass the stage and use an LM4562/LME49720 opamp. I don't have enough time to do this right now

 

[NTTY]

This here seemingly was enough to cause subjectively wonky treble in an IDT 92HD93 chip, as all other measurements checked out:
View attachment 399201
(It's the difference between 44.1k straight and it being upsampled to 192k first, generated by subtracting both measurements.)

We could agree on the graph axis min/max so we all represent the same way. I usually go for +-7dB, but here below with +-2dB, as you showed (with a Marantz CD-73 and the Onkyo C-733):

I prefer linear frequency scale, but I might be the only one:

 

[NTTY]

I could just bypass the stage and use an LM4562/LME49720 opamp. I don't have enough time to do this right now

You can do as @AnalogSteph suggested, it bypasses the circuitry and reverts it to CD-57 setup.

 

[horias2000]

You can do as @AnalogSteph suggested, it bypasses the circuitry and reverts it to CD-57 setup.

Yes, I might do this at one stage and I'll change the opamps as well as the NJM2140 is less of an opamp than an LM4562 and I have many of the later sitting around

 

[restorer-john]

Does your and AnalogSteph's opinion apply to the current implementation of HDAM in the Marantz lineup. They are touting it as a desirable feature.

I don't know. You/we would need to test in circuit and out of circuit for the so-called current (latest) design "HDAM". In those early implementations of the Marantz HDAM, the ultimate performance suffered. I have a few machines here I could test, in and out of circuit, but that is historical. If it's of interest, I'm happy to spend the time and do some testing. The HDAM as described is a unity gain bufer, so not much to be gained and a whole lot to be lost.

 

[NTTY]

Yes, I might do this at one stage and I'll change the opamps as well as the NJM2140 is less of an opamp than an LM4562 and I have many of the later sitting around

I don't recommend changing AOP randomly, have a look here.

And a friend tried the same on a Yamaha CD Player.

But you can go for it and report the changes, it might be interesting.

 

[NTTY]

I don't know. You/we would need to test in circuit and out of circuit for the so-called current (latest) design "HDAM". In those early implementations of the Marantz HDAM, the ultimate performance suffered. I have a few machines here I could test, in and out of circuit, but that is historical. If it's of interest, I'm happy to spend the time and do some testing. The HDAM as described is a unity gain bufer, so not much to be gained and a whole lot to be lost.

I'd be interested, especially because historical

 

[AnalogSteph]

I prefer linear frequency scale, but I might be the only one:

Definitely not. When looking at periodic ripple, it actually makes a ton of sense.

It would be neat if there was an easy way to do a frequency decomposition... often a few different ones will be overlaid.

BTW, SM5872 filter passband ripple as per its datasheet:

The periodic part is about +/-0.015 dB with about 4 periods per .25fs, so roughly 2.75 kHz @ 44.1.

Yes, I might do this at one stage and I'll change the opamps as well as the NJM2140 is less of an opamp than an LM4562 and I have many of the later sitting around

Yeah, but I mean NJM2114 is basically a hot-rodded 5532 to begin with and used inverting with highish impedances to boot here, so it should be quite bored as-is and is unlikely to be a limiting factor. The Asus Xonar ST used the part and posted loopback measurements like this. Also, as good as the LM4562 is, it does have its quirks, including an RF allergy that @pma pointed out.