Friday 18 September 2015

MEASUREMENTS: ASUS Xonar Essence One MKI with DSD upgrade kit. (Part I: EEPROM)


Ahhhh... The ASUS Essence One! Even though looking back, compared to others I have listened to or measured, this DAC had its various imperfections, it has been my constant companion on the computer workstation. Through it, I have listened to hundreds of hours of music, edited my share of audio, evaluated pairs of headphones, and run various "experiments". Like an old friend, you get to learn (and at times even appreciate) the quirks. To this day, it remains my "workhorse" DAC. Why, you might ask?

Because it has 2 nice big volume knobs. One for speaker output, the other for the headphones... Simple as that :-).

If you look back at my measurements and review of this DAC back in 2013, you'll see that I generally liked it. With my perspective today, though I believe much can be improved (everything from channel/volume control balance [apparently fixed with later Essence One MKII edition which also incorporates the DSD feature], J-Test could look better, noise floor cleaner, etc...), it was the hyped up "symmetric upsampling" that was perhaps the most bizarre let-down. Imagine a digital filter that acted as a brick wall at 15kHz! Certainly made the DAC sound "mellow", but in a bad, inaccurate way...

Despite rumours of a fix to the firmware, nothing happened after I reported this anomaly. Until late 2014 by the looks of it. And I wasn't aware of this until this past month...
As I had previously mentioned, one of the joys of running a blog is the opportunity to interact and virtually "meet" some of you out there. And Raul from Europe got in touch and informed me of something I was unaware of - ASUS offers an "upgrade kit" for the Essence One! Despite this kit being listed in Europe and Asia (Amazon.de from Germany for example), it seems to be MIA here in North America.

Product stock photo...

With Raul's generous assistance, I was able to get the kit contents shipped to my doorstep:

As you can see, it's all rather simple... A couple of New Japan Radio Corp MUSES 02 DIP8 bipolar opamps and behind the opamps is the DIP8 EEPROM (4Mbit/512KB, EN25F40) which contains the updated firmware. A real shame that this could not have been just a software firmware upgrade. Note that this kit costs ~97 EUR (~$110USD) plus tax/shipping. The most expensive parts are the opamps. If you are looking at just acquiring the MUSES opamps at Mouser.com, they're about $45/piece retail currently. Very expensive opamps indeed (watch out for eBay fakes)! So, that's basically $90USD for the opamps, and ASUS is charging about $20 for the EEPROM, the main feature of which being the ability to play DSD64 via DoP. Let's see if this is worth it...

For today's segment, we'll just look at the features brought on by this EEPROM upgrade... "Opamp rolling" is another topic altogether and I think deserves its own post (Part II) another time especially given the cost of these MUSES.

For those doing this upgrade, remember that it's not just a swap of the EEPROM in location (1) below:

But also an upgrade of the MCU (C-Media CM6631 microcontroller unit) firmware (go here --> Support --> Driver & Tools, drop the menu down to Windows 8.1 64-bit to see the v0.64 MCU firmware update). Follow the instructions to put the DAC in "update mode" and remember to only update the firmware with a USB2 port. Don't worry if it takes a few tries - I needed to repeat the update process four times before it "passed"! I assume this only works in Windows as I did not see support for Mac or Linux.

Remember to also update your Windows ASIO driver. I'm using the latest Windows 10 beta.
Original left, latest DSD upgraded right (ignore the MKII designation). Ironically, MCU version 0.64 is newer than 1.27.

It has been years since measuring the Essence One, and potentially the last time, let's acquire the full set of results and see if anything has changed over the years of use... For these measurements, I'll just use the single-ended RCA analogue output.

I mentioned this in my last post. My Essence One is actually not the stock unit - I had replaced all NE5532 opamps with LM4562's previously and have been using this configuration over the years.

Part I - Oscilloscope, Impulse Response, Digital Filter

As usual, let's look at the "macro" picture with the 0dBFS 1kHz sine wave on the digital oscilloscope:

Other than the fact that in 2013, I had swapped the channels (I have since standardized CH1 = left, CH2 = right as shown in the 2015 measurements), obviously there's really no difference here. One complaint with the Essence One is that there is a small channel imbalance that can be seen here (I believe this may be fixed with newer hardware versions). Max peak voltage is around 2.80V; or ~2Vrms.

Impulse response with the DSD Kit update - without and with UPSAMPLING turned on:

Well, well, well... Looks like ASUS has decided to just stick with a typical linear phase sharp filter for both standard and with "symmetric upsampling" (compare this with the impulse response with upsampling before this DSD Kit). This is actually encouraging in that it means we're not going to see that issue with 15kHz bandwidth limit like we did before upgrading the EEPROM when upsampling. However, looking at the digital filter composite ("Reis Test") which I first showed recently with the Samsung Galaxy Note 5, we see that indeed the UPSAMPLING does invoke a different type of pattern:



The UPSAMPLING algorithm is an improvement over the default DAC filter. The noise floor is better, and this is evident also when we're playing the 19 & 20kHz tones. Notice with the wideband white noise with 0dBFS peaks that when stressed, the digital filter in the original as well as with UPSAMPLING do unfortunately "overload", resulting in elevation of the noise floor.

Part II: RightMark Tests & Comparisons

Setup:
Microsoft Surface 3 Pro --> shielded USB --> ASUS Essence One --> 3' shielded RCA --> E-MU 0404USB --> shielded USB --> Measurement Windows 8 computer
Test linkup!

16/44:

Here's the big summary. To the left, you have the E1 "Original" measured right before the minor surgery and EEPROM upgrade; default setting and with the UPSAMPLING engaged. Notice the issue with significant drop in frequency response when UPSAMPLING before. The next 2 columns are with the new "DSD Kit" EEPROM. Now the UPSAMPLING is much flatter in the frequency response. Then we have comparisons with the Logitech Transporter and TEAC UD-501. Although not really audible, notice that IMD+N performance improved slightly with the DSD Kit. However the Essence One is still higher than what I measured with the Transporter and TEAC (not that I consider these small differences significant).

As usual, I think it's fair to say as I have before that 16/44 poses no problem / challenge at all to modern DACs.

ASUS fixed the bandwidth issue with UPSAMPLING in the original firmware.

Noise floor. Notice the higher 60Hz mains hum with the Essence One. (At -100dB, not generally audible even with 100% headphone volume and close headphones.)

IMD+N

24/96:
Big summary table:

For fun, in the first column, you see the Essence One measurements from 2013 when it was almost brand new! Back in those days, I was using an earlier version of RightMark and recording with MME rather than ASIO drivers so it's not as directly comparable. However you see that it's still pretty similar to the 2015 result before and after the DSD Kit upgrade. Again, Transporter and TEAC UD-501 results shown for comparison. There's quite a bit of variability in the IMD+N result for the Essence One which I believe is inter-test variability. I also don't worry much about the variation in stereo crosstalk at such low levels since different cables will have an effect.

Frequency Response. Notice that the "DSD Kit" upgrade is slightly different from the "Original" whether 2013 or 2015... I suspect ASUS did some tweaking of the digital filter parameters.\


Noise floor - Essence One only.
Notice what has happened to the 60Hz mains hum. It has gotten worse over time between 2013 to 2015 suggesting that after hundreds of hours, if one considers this as "break-in", the tendency is to actually deteriorate performance. Although the software and computer hardware is different between the two measurements, the magnitude of the difference (>10dB) suggests that this is indeed a real finding and consistent with my previous exploration of this topic with the TEAC DAC in early 2014.

THD. In comparison to Essence One, The Transporter and TEAC had lower 60Hz noise even back in 2013.
DSD64 via DoP:
The main advertised benefit for this EEPROM (and presumably why it should be worth $60USD beyond just the MUSES 02 opamps), is the fact that the Essence One can now play DSD64 via DoP.

ASUS has instructions for foobar on their website (straightforward, install ASIO support, foo_input_sacd, and ASIOProxy in the package, set DSD64 to DoP). JRiver is also easy, just make sure "Bitstreaming: Yes (DSD)" is selected in "Player --> Playback Options..." and the ASIO setting is turned on to bitstream DSD as DoP. DSD64 .dff and .dsf files should play back fine including those with DST compression:

ASUS Xonar Essence One playing DSD64. Notice the UPSAMPLING LED turns red now when operating in DSD/DoP mode. Also, I stuck silver heatsinks over the CM6631 and the 2 PCM1795 DAC chips (they run pretty warm).
So, how's the DSD64 resolution, you ask? Using a 24/192 RightMark PCM test signal converted to DSD64 using KORG AudioGate, here's what I get:


Frequency Response - note the ultrasonic noise affecting the curve with DSD playback.
Noise floor - Essence One DSD noise level very high!
IMD+N. Again, Essence One DSD noise floor high.

In white, we see the Essence One playing the 24/192 test as native PCM. In green, it's playing DSD64... And sadly as you can plainly see, it's awful. Noise level is up at -82 dB or so! This means that the dynamic range is even worse than a standard 16-bit CD!

For comparison, you see the same test done using the TEAC UD-501 in cyan. Now this is what it should look like. DSD64 should be able to provide high resolution dynamic range within the audible spectrum to just over 20kHz and thereafter the high frequency noise shaping ultrasonic noise starts to escalate.

Out of curiosity, here's what the 1kHz 0dBFS square wave looks like through the digital oscilloscope:

So it looks like the DSD64 analogue output at 100% volume is  -6.5dB compared to the direct PCM output. You can also see the noise in the square waves.

Part III: Jitter Test

Compared to others I've tried, the Essence One does not produce the cleanest Dunn J-Test spectrum:



Despite the lack of "cleanliness" to the noise floor, we can see the 16-bit jitter modulation pattern quite well in each of the cases. Not much difference noticed in the 24-bit J-Test between the "original" firmware and the updated EEPROM. But we do see that engaging UPSAMPLING in the 24-bit test does seem to induce a bit more jitter sidebands. Of course, even though I can show this, I am not particularly concerned since these sidebands are less than -110dB below the primary signal peak.

Now out of curiosity and completeness, I wondered what the J-Test would look like played back as DSD64 (again using KORG AudioGate conversion):

Not unexpectedly, that high noise floor simply obliterates any low-level effect we might see on a J-Test like the stimulation of jitter sidebands or other spurious noise...

Part IV: Subjective Impression

I don't think there's much to say here since this is the ASUS Essence One after all with just a firmware revision. For the most part, I've enjoyed the sound of the Essence One over the years and really cannot complain much about the general presentation. Although I know the measurements are not perfect and can point to areas of deficits, they do not impair my ability to really enjoy what I hear and that dual-volume-control remains an essential feature for my computer desktop listening. The drivers have worked flawlessly switching between ASIO samplerates reliably and with ease which is of course essential when you're doing sound editing!

Good to see that they did finally fix the UPSAMPLING feature. Frequency response is clearly better now that the full bandwidth has been restored. I didn't test in a blind fashion and it does take about 5+ seconds to switch between settings after pushing the front button (a few seconds longer than before). Instantaneous A/B testing is therefore not possible. I do believe there is a small audible difference nonetheless. With upsampling turned on, it seems to sound a bit smoother and "relaxed". I'll certainly consider experimenting and try using this feature now...

As for DSD64... Yikes. To be clear, 80dB (~13.5 bits) or so of dynamic range is not horrific as to become unenjoyable, but it is not "high fidelity". With modern loud masterings with DR <10dB, it's probably not a big problem. But with dynamic music where the average amplitude is lower and you need to pump up the volume a bit, sadly, the limitation is audible. In fact, in a quiet room I can hear the background hiss go down when switching between DSD64/DoP playback and PCM playback at normal headphone levels. Not impressed, ASUS...

Part V: Conclusions

So, let's summarize what this EEPROM upgrade does:

1. The ASUS "8X Symmetrical Upsampling" is worth trying/using now. It's a sharp linear phase digital filter capable of producing lower noise floor and distortion products as seen with the digital filters test. However, it does induce a little more jitter which IMO is not an issue.

2. Indeed, you can now play native DSD64 as a DoP bitstream.

3. BUT... The DSD64 playback is noisy and I would likely not be using it much. Noise level restricts the dynamic range to the low-80's dB. This is worse than 16-bit PCM plus you get that elevated DSD ultrasonic noise from noise shaping. From what I have heard with my other DACs, this is unacceptable for "audiophile" playback.

So, discounting the cost of the MUSES 02 opamps, is the EEPROM worth ~$20USD (assuming you don't mind getting those MUSES opamps of course)? A reluctant "okay" I guess... Although the upsampling feature should have been better the first time around, $20 gets this fixed and some nominal DSD64 playback. As I indicated, unless ASUS actually does something to improve that DSD playback noise level, it's really not a satisfactory quality DSD listening experience. If they find a solution, I truly hope this will be a free update of some kind and ideally not another EEPROM swap!

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

In the weeks ahead, I will evaluate the MUSES 02 swap for a taste of "opamp rolling" and especially look for objective sonic change... Stay tuned!

Thanks again Raul for getting the EEPROM upgrade to me! Enjoy the music, everyone :-).

8 comments:

  1. Looking forward to the Muses thoughts...
    I sold my my original Essence One almost 2 years ago, but now found a Muses Edition for a very good price.
    Biggest problem with the original was that the gain was way to high for sensitive headphones. The Muses edition has a gain jumper, so it should be fine now. Would be interesting to see what the difference between Muses 01 and 02 is.

    ReplyDelete
  2. Hello!
    I'm also looking forward to the Muses review, as I have the original Essence One and use it daily. I've always wondered if I should do some opamp rolling or not, but the MUSES 02 review got my interest, to see if it's worth or not.
    Regards

    ReplyDelete
  3. Hello!

    I did myself a couple of A/B tests between 2 identical E1 DACs:
    - one had NE5532 in I/V and another one had MUSES01: difference was obviously in the next few seconds with a clear advantage for MUSES01 in details and soundstage.
    - one had NE5532 in I/V and another one had MUSES8920: differences were quite subtle, just a bit more clear and detailed with MUSES8920, but soundstage was about the same as NE5532. MUSES8920 sounds very neutral, but clearly is not on the same league as MUSES01.
    - one had NE5532 in LPF and another one had MUSES02: MUSES02 is a little bit more clean sounding, but couldn't hear a real improvement in soundstage.
    - one had MUSES02 in LPF and another one MUSES8820: I wasn't able to pick up any noticible differences.

    Looks like upgrading the 4 x I/V OPAMPs to MUSES01 will make a big difference in clarity and soundstage and this is a highly recommended upgrade for Essence One. Also, if possible, go balanced for further soundstage increase, but this means to buy a balanced amplifier and also to upgrade your headphones cables.

    MUSES02 could be used in LPF stage and also in voltage-gain of internal headphone amplifier (instead of the 2 x LME49720). This upgrade will make this DAC sounding even more clean and detailed.

    So, my final recommendations for upgrading Essence One OPAMPs would be:
    - 4xMUSES01 in I/V
    - 2xMUSES8820 in LPF (or 2xMUSES02 if you really have the money for this, to be 110% sure you get the best sound quality)
    - 2xMUSES8820 in headamp's voltage-gain (or 2xMUSES02 if you really have the money for this, to be 110% sure you get the best sound quality)

    Haven't tested MUSES02 or MUSES8820 in output buffers, but I don't really intend to do this soon.

    Regards,
    Raul.

    ReplyDelete
    Replies
    1. Just tried BURSON SS V5 and SS V5i opamps in LowPassFilter, Voltage Amplification stage and also in Buffer stage. Need to tell you that it's a real improvement over any other opamps I've tested. Only downsize for the V5 would be the size, so perhaps V5i might be a better choice in case you intend to close the case.

      So now I have ASUS Essence One MUSES/BURSON MKii. :) With 2 x MUSES01 in I/V, 2 x SS V5i in LPF and 2 x V5i in VAS I love the soundstage, the voices and bass extension (especially when used the integrated amplifier). I can recommend SS V5i for LPF, VAS and BUFFER.

      Delete
  4. Hey. And can you dump the EN25F40 chip from the "Essence One DSD Upgrade Kit" kit? Pls! =)

    ReplyDelete
    Replies
    1. Alas, I believe ASUS would not like us to do that :-).

      Delete
    2. =))

      It was difficult, but I found a dump in another secret place.
      Asus is too greedy to just sell 1 chip to EN25F40, without Muses. I do not need Muses, but I want to try DSD. I myself will make a DSD update, at a price of $ 1 this is how much the EN25F40 chip costs. And $ 3 is worth the CH341A programmer for spi flash.

      p.s.: Sorry for the bad english.

      Delete
  5. Hey,
    I was wondering if you could dump the EEPROM chip, which I've seen you don't really want to distribute, but ASUS don't sell them anymore, so it's impossible to find anymore.
    Thanks!

    ReplyDelete