Monday 4 January 2016

MEASUREMENTS: ASUS Xonar Essence One DSD Upgrade (Part II: NJR MUSES 02 "Audio Opamp Rolling")

As you may recall, a few months back, I posted on the ASUS Xonar Essence One and the DSD Upgrade Kit. That post was only Part I because the kit not only included an upgraded firmware EEPROM to allow DSD64 playback, but there were a couple of New Japan Radio MUSES 02 opamps in there to use as well. New Japan Radio seems insistent on marketing this MUSES brand of opamps for audiophile applications and certainly the price tag is consistent with the audio "high end" - we're talking US$45 per stereo "flagship" opamp (MUSES 01 and 02)! As a result, there are quite a number of fakes out there especially on eBay so make sure you get these chips from a reputable dealer if you're in the market.

On a side note, it's interesting that companies these days are using "replaceable opamps" as part of the feature set of devices like motherboards; even motherboard manufacturers like Gigabyte are in the act of selling opamp kits!

When it comes to the old ASUS Xonar Essence One, I figure why not perform a few measurements and see if replacing the LM4562NA opamps (US$2.00 a piece) I had in there with these expensive MUSES 02's made a difference; the company claims that there's a "Profound Musicality" benefit with the MUSES. As a reminder, I had put in the LM4562NA opamps a few years ago. The stock Essence One uses NE5532 (<US$1.00) so my results may not be the same as someone going from stock configuration to these MUSES 02's. Remember that "opamp rolling" is not uncommonly discussed on message forums. And there have been some excellent write-ups in the past. Like high-end cables, there are those who swear by the improvements they hear but looking around, I have not seen anyone publish objective results from a DAC despite all kinds of testimony.

Note that there are different positions I could place these MUSES in but basically settled on the Low Pass Filter (LPF) stage which potentially could benefit all audio outputs. (In diagram below: 3A - for RCA output only, 3B - XLRs, 4 - headphones only.) I really did not have any great desire to pull out and reseat various opamps to try different configurations so just plugged them into place and closed my DAC, running measurements before and after the surgical procedure.
So here they are situated on the PCB:


I. RightMark Measurements:

Measurement set-up is as documented in Part I. Only variable of course being what opamp was plugged into the LPF sockets (everything else like cables and measurement procedure the same). Let's start with 16/44:

Frequency Response

Noise Floor - notice the 60Hz hum

IMD+N

Stereo Crosstalk - no change to RCA cables use.


Here's the data for 24/96:


Frequency Response

Noise Floor

Stereo Crosstalk
As you can see, the differences are very small. However, having done these tests for awhile now, I actually believe the slight differences in the frequency response graphs are probably real and not just inter-test variability. I actually ran the tests a couple of times and the minor difference was consistent. Of interest is that the MUSES 02 frequency response was ever so slighted more extended at the frequency extremes for both the 44 and 96kHz measurements. Small enough though that I would not expect audible differences from the minor variation!

II. Digital Filter Composite (DFC)

What I've come to call the "Digital Filter Composite" (DFC for short) based on the original "Reis Test" as discussed in Stereophile (with LM4562 opamps):

Now with MUSES 02 opamps installed:

Again, some differences which are obvious on objective testing. Of course the general shape and noise levels look about the same due to use of the same internal digital filtering algorithm. Notice however the MUSES 02 graph shows more second harmonic and intermodulation products at 35-42kHz but slightly quieter around 50kHz. Interesting to see but again hard to make a case for audibility as this is all ultrasonic and as far as I can tell, the 20-20kHz noise floor whether with the 19 & 20kHz sine waves playing or digital silence appear essentially no different.

III. J-Test

Finally here's the J-Test - first with my LM4562NA opamps:

And with MUSES 02 installed:

As expected, no difference of significance as far as I can tell. Changing opamps would have no effect on timing parameters and as we saw above, noise floor differences were not in the audible range.

IV. Subjectivity...

Subjectively, well, I did pay about US$90 for the two opamps (that's how much it costs from a place like Mouser) so I could tell you that the bass line on "My One & Only Love" on Oscar Peterson Trio's We Get Requests sounded phenomenally deep with immaculate control after I plugged in the MUSES 02. I could say that listening to Ben Webster's tenor sax on "When I Fall In Love" (from The Soul of Ben Webster) sounded remarkable smooth with excellent tonality, as fine analogue recordings and playback should :-). I could tell you that Philippe Jaroussky's voice on Caldara In Vienna (Forgotten Castrato Arias) sounded beautifully other-worldly and "present" after inserting the MUSES opamps. I could say that those chaotic bells and chimes on Pink Floyd's "Time" (Dark Side Of The Moon) was fantastic in terms of how palpably real and how wonderfully open the soundstage was. I could say the MUSES 02s were a real improvement over the LM4562NA chips they replaced and how they elevated the standard of the DAC from something that sounds like a US$500 unit to something competing at the US$1000+ level! Yes, I could say all these things because these ideas came to mind while listening here and there over the last 3 months, as I mulled over how I was going to describe what I heard.

But I will not say those things above :-). Seriously folks, without the ability to switch between two identical DACs with different opamps installed, the subtle differences I thought I heard are likely the result of selective attention and limitations of echoic memory rather than correlating with objective reality. As Nobel prize laureate Richard Feynman wisely reminded us: "the first principle is that you must not fool yourself, and you are the easiest person to fool." That's just the nature of subjectivity. Of course that's not to say subjective opinions are useless, rather, I think it's an acknowledgement that there is a "gradient of certainty". Some things are reasonably obvious and clear without needing to double check as in an A/B test, but when differences are at best subtle, it's honest to not get ahead of ourselves with proclamations of significant differences and flowery language. The music and DAC sounded great before, and certainly I'm quite happy with the sound after the MUSES 02.

V. Conclusion...

I guess, ultimately, I can say that the Essence One with MUSES 02 opamps in the low-pass filter circuitry of the machine did sound good and I would have no problem enjoying music with them... Worth US$90 for what amounts to very slight objective differences (like what looks like 0.1dB changes in the frequency response at the extremes, and measurable ultrasonic noise floor difference) compared to the US$4 opamps they replaced in the LPF circuitry of the DAC? Mmmmm... No. But I can at least say I've tried for myself. And here's some data to consider...

Time and motivation permitting, I'll see about trying these opamps further down the circuit for RCA or XLR output and see if that makes more of a difference. Remember, we're talking about a tweak here. As we have seen before, short of new speakers, improved room acoustics and better mastering, except in the words used by promoters, it's best not to expect too much beyond subtle changes only.

As usual, I would love to see if others have objective results while "opamp rolling".

For the DIY'ers out there, Raoul Trifan posted on Head-Fi forum a mod to the power supply of the Essence One to substantially improve ripple and lower noise. Check it out!

-------------------------------

Well, the Holiday Season has come and gone :-(.

I hope everyone got some rest and ready to take on 2016!

Prepare for the MQA onslaught at CES 2016 :-). It'll be interesting when comments come out with impressions of A/B listening.

BTW: I'd be nice if someone can ask if there's any DRM mechanism embedded in MQA and clarification if MQA is able to encode resolution >16-bits... Of course, any details about just how it "de-blurs" timing anomalies in a PCM stream could be enlightening! Thanks.

15 comments:

  1. Thanks for this Archimago. It confirms my hunch that there is not much to be gained from different DAC's and the like, as long as you avoid tweeky audiophile snake oil units. My latest investment has been in a handful of Chromecast Audio units to update a number of systems in the home. Installing them was easy, and they work like a treat for next to nothing, including multiroom play.
    My initial impression is that the analogue sound quality from the built-in DAC is fine, but I have no easy way to establish this rigourously. My main system should be revealing enough: Quad 33/606-2 into Quad 2805 electrostats plus B&W PV1d sub tamed by an Antimode 8033. I am playing this into the Q33 tape input, set to its lowest sensitivity. The CC is in high dynamic range mode. Any chance that you will measure a CC audio? In the meantime I have this naughty pleasure that the CC may well be as good as some streamers costing thousands of dollars, and be more convenient to use in the bargain....

    ReplyDelete
    Replies
    1. Hi Willem, yeah I'll likely pick one up to test out in the next while. These CC's are certainly cheap, accessible and should work really great.

      For the folks with DACs, I'd love to get a sense of the jitter out of the mini-TosLink. Cheers, glad you're enjoying the unit!

      Delete
  2. Hello Archimago.
    I'd also like to thank you for this. It sheds some light for me on what opamp rolling can "improve" on my E1 as I have been thinking lately of buying some OPA2134s (they are not expensive) to try it out in the I/V and LPF sections, but as of now I'm not so sure.

    But I'd like to ask you two things.
    The first one, is why did you initially changed your NE5532 opamps for the LM4562NA ones?
    And the other one, have you ever tried the MUSES or the Plus editions of the E1? Do you think those were just more like marketing gimmicks instead of having real, objective (measurable) improvements in sound quality?

    Best regards,
    Alexandre Cavaco

    ReplyDelete
    Replies
    1. Good questions Alexandre...

      I had a few LM4562's around so I thought I'd give it a try >3 years ago when I first got the E1. Back then I was curious as an "audiophile" and heard all kinds of good things about the LM sounding cleaner, hence the change. I remember agreeing with the subjectivists at the time but truth be told, I wouldn't say it was an "essential" upgrade and now that I've tried the MUSES, am not convinced about these impressions. The truth is, I still have those 5532's in my parts box somewhere if I want to revert... Again, maybe in time I'll give this a try and measure.

      As for the Plus and MUSES editions, I have not had a chance to listen to those. I know some people will swear by the opamp swaps and love their MUSES editions. I can't say for sure if it's more marketing gimmick or not but would certainly love to see objective demonstrations of change if anyone else has tried!

      Delete
  3. I am one of the many enthusiasts that spend time on OpAmp rolling. After trying 10-12 different types I have settled to the same LM4562 (or their newer TI labeling LME49720), so that's what I have in all my modded players/amps.
    One observation - for driving headphones though they are not as good as others.
    Interesting your test with the MUSES, and if you could do a THD/IMD test with a good ADC it might reveal more. I have bought for that exact reason (measurements) an E-MU 1820m that has the best ADC chip that you can find in computers sound cards.
    Also, in the filtering stage, there is only so much difference that an OpAmp can make.
    The real difference is in the I/V stage - if your DAC chip has a current output. That's where the speed and noise count the most.

    ReplyDelete
    Replies
    1. Thanks for the tip. Will see about measuring the I/V stage.

      Yeah, can always get a better ADC... The thing is, it's hard to imagine it'll make a practical difference considering how accurate the 0404 is already compared to the THD/IMD out of one's headphones and speakers!

      Delete
  4. This comment has been removed by the author.

    ReplyDelete
  5. Hi.
    Maybe you will try discrete opamps as well. I have OPA301 (from Audiofeel) in lpf section of my axeo but some people tried Burson: http://jimmyauw.com/2014/11/29/burson-supreme-sound-opamp-4th-generation-first-impression-continued/ . They change sound more spectacular but I'm wondering if that could be related to some distorsions they could create?
    Regards,
    Emil.
    ps. I also agree that I/V is most important section in axeo. I'm using lme49860 or 2xlme49990 in this section.

    ReplyDelete
  6. Very well documented article, like always, thank you! :)

    If you want to remove the 60Hz hum you might want to shield the transformer with G.O.S.S. (grain oriented silicon steel) or something similar. I've continued modding my E1 recently with http://www.head-fi.org/t/542563/asus-essence-one-headphone-amp-dac-cebit-2011/3240#post_12269635.

    Very good I/V OPAMPs are MUSES01, so I'm waiting for your 3'rd article about the DSD upgrade with MUSES01 in I/V stage. :) I'm using MUSES01 right now and I've done A/B test by using 2 different E1s and I've noticed real differences when A/B testing: much clean and detailed sound and a definitely increased sound-stage with MUSES01 vs. other FETs like OPA2132, OPA1652 or against BJT OPAMPs like LM4562 etc.

    P.S.: You can save some money and buy only 2xMUSES01 if you intend to use single-ended RCA (you may use only 1'st and 3'rd I/V sockets), but for XLR outputs you'll need to buy all the 4 OPAMPs, so...quite expensive, I know.

    ReplyDelete
  7. Could anyone confirm (measure) if that hum from rca output is also present with xlr to rca adapter (with pins: 1 and 2 conncted)? Maybe xlr output doesn't have this problem even when we take only half of signal (maybe different path of ground or sth)?

    ReplyDelete
    Replies
    1. Emil, I've read on Head-Fi that not everyone has this hum on Essence One, so not all E1s are affected by this. I used to have it myself, but I was able to hear it only with my sensitive AKG K550 and volume knob to the max at 1 AM in the morning (very quiet environment). Anyway, in case there will be a hum then this will have quite low intensity and shouldn't be heard with normal music levels.
      BTW, this hum is not caused by a ground loop, it's trafo hum and all you need to do to get rid of it is to re-position the trafo or to shield it. Hum can only be measured with case fully closed, when case is open there's no more hum (Faraday cage, perhaps?!?).

      Delete
  8. Thanks for reply. Did you try discrete opamps with your axoe? So far I've tested those:
    http://www.ebay.com/itm/1X-OPA301-DUAL-AUDIOFEEL-Hi-End-Audio-DISCRETE-OP-AMP-Board-Low-Impedance-/271838130217?hash=item3f4ad0b029:g:0gMAAOSwwTlUo1a1 very good sound and not so expensive (althoug my axeo must be opend all the time becouse I don't want to use any extension kits).

    ReplyDelete
    Replies
    1. Haven't tried discrete OPAMPs yet, but after trying a lot of BJT and FET OPAMPs I settled down to MUSES01 in I/V and MUSES8820 in LPF, headamp voltage gain and output buffers (XLR/RCA).

      To be honest, I woudn't recommend discrete OPAMPs for I/V, instead you could use fast FET OPAMPs or pasive I/V.

      Delete
  9. Hi
    would you mind providing some info about how you run the Dunn J-test

    I have looked but can't find any software to provide results like yours

    Cheers

    ReplyDelete
    Replies
    1. Hi Brad.

      The software HpW-Works does have good possibilities, to measure the Julian Dunn J-Test.

      http://www.hpw-works.com/index.php

      They have the correct signal generator signals and the correct FFT analyzer.

      It is not that, what Archimago is using, but highly recommended for that.

      Juergen

      Delete