CHANGSTAR: Audiophile Headphone Reviews and Early 90s Style BBS
Lobby => Amp and DAC Measurements => Topic started by: Hands on December 27, 2013, 02:04:08 PM
-
Test Hardware and Setup
-Acer Aspire 5560G Laptop - Locked at 1.4GHz and 0.9125V via K10Stat, dedicated GPU disabled, internal fan disconnected (low temps and stability verified), most Fidelizer tweaks applied, and other hardware, software, and services disabled. Battery powered.
-JKSPDIF MK3 - Input with generic USB cable, output with 75ohm RF attenuator and 75ohm BNC to RCA Coaxial cable to DAC. Battery powered.
OR
-DIY Gaming Desktop - i5 2500K locked at 4.5GHz, Biostar TZ77XE3 motherboard, GTX 780, Antec EarthWatts 650W Green PSU, Auzentech Bravura sound card, 4 140mm fans, goofy UV cathode lights in case, Fractal Design Define R4 case...you get the idea (not optimal for audio). Plugged into APC H10 power conditioner. I did run Fidelizer on it for the tests.
-Auzentech Bravura sound card - It's plugged into the beef rig. Digital coax out to DAC using a digital coax cable from Monoprice (should be 75 ohm).
-----------------------------------------------------------------------------
-Audio-GD NOS-1704 DAC - Fed from either Auzentech coax out of JKSPDIF. Plugged into APC H10 if hooked up to the gaming rig. Tests were done in high voltage mode (RCA out) unless specified otherwise (file names will have "L" or "LV" in them if low voltage, "H" or "HV" for high voltage). Used 6' BJC RCA cable plugged into large, recessed RCA plugs. Unpublished tests show that you generally get slightly better results using the other RCA plugs (maybe 0.1-0.2dB difference, but there).
I am also using the CS8416 "upgrade" option for the coax inputs on the NOS-1704.
-Creative X-FI HD USB (SB1240) - Takes line-out signal from DAC to line-in for recording purposes, powered by USB bus, always connected to Acer laptop for measurements (even when DAC was hooked up to desktop). Unit is limited to 48/96KHz input/output.
Software
RightMark RMAA 6.3.0, ARTA
Misc. Notes
Creative SB1240 has a slight roll-off in the bass.
I have seen as much as a 2dB variance with the same tests. What I've posted is the least you should expect from the unit.
I wasn't going to test the Auzentech coax out with the NOS-1704 originally, but I tried it one day out of laziness and decided I had to measure it and compare against the JKSPDIF. The results might surprise you.
Links
Audio-GD NOS-1704 DAC - http://www.audio-gd.com/Pro/SA/NOS1704/NOS1704EN.htm
Auzentech Bravura sound card - http://www.auzentech.com/site/products/x-fi_bravura.php
JKSPDIF MK3 - http://www.johnkenny.biz/home-1/mk3-hiface
Creative SB1240 - http://us.creative.com/p/sound-blaster/sound-blaster-digital-music-premium-hd
RightMark Audio Analyzer - http://audio.rightmark.org/index_new.shtml
ARTA: http://www.artalabs.hr/
-
Jitter tests using a 24/48 JTest file. JTest file was played back through JRiver Media Center 19 on both the laptop and desktop setups. Hardware sound device settings and software (ARTA) were set to 24/96, however. I ran tests with everything set to 24/96, and there was no difference vs. these results.
As mentioned in the first post, this is using the first RCA coax input on the DAC (technically coax input 2).
These results were surprising to me, and you'll see a difference in other measurements as well. Files attached are:
1. Laptop playing JTest in JRMC (no upsampling, kernel streaming) -> USB to JKSPDIF -> NOS-1704 -> SB1240 -> Back to laptop through USB
2. Gaming desktop -> Auzentech Bravura sound card coax out -> NOS-1704 -> SB1240 -> Laptop USB in (No bitstreaming or upsampling enabled in JRMC)
3. Same as above but with bitstreaming and without upsampling
4. Same as above but without bitstreaming and with upsampling (24/96)
-
24/96 Spectrum tests in ARTA, using the laptop -> JKSPDIF setup. Files attached are:
1. 1KHz sine wave spectrum
2. 100Hz sine wave spectrum
3. 60Hz + 7KHz dual sine wave spectrum (4:1 ratio)
4. 25Hz + 66Hz dual sine wave spectrum (1:1 ratio)
5. Noise
-
24/96 spectrum tests in ARTA, using the gaming rig -> Auzen Bravura coax out setup. I seem to have not saved the 100Hz sine wave spectrum results.
1. 1KHz sine wave spectrum
2. 60Hz + 7KHz dual sine wave spectrum (4:1 ratio)
3. 25Hz + 66Hz dual sine wave spectrum (1:1 ratio)
4. Noise
-
RMAA tests at 16/96, using laptop -> JKSPDIF setup, kernel streaming mode. KS mode with this setup tends to produce better results than other modes in RMAA. I'll attach a zip of all my tests later.
Frequency response (from 40 Hz to 15 kHz), dB | -0.00, -0.65 | Good
Noise level, dB (A) | -95.9 | Excellent
Dynamic range, dB (A) | 94.1 | Very good
THD, % | 0.0033 | Very good
THD + Noise, dB (A) | -84.4 | Good
IMD + Noise, % | 0.0075 | Excellent
Stereo crosstalk, dB | -92.2 | Excellent
IMD at 10 kHz, % | 0.027 | Good
General performance | Very good
-
RMAA tests at 24/96, using laptop -> JKSPDIF setup, DirectSound mode. See how this compares to the Auzentech Bravura coax out tests in the next post (also DirectSound). Also note the improvements KS brings (post above).
Frequency response (from 40 Hz to 15 kHz), dB | -0.01, -0.66 | Good
Noise level, dB (A) | -90.6 | Very good
Dynamic range, dB (A) | 90.5 | Very good
THD, % | 0.0039 | Very good
THD + Noise, dB (A) | -81.7 | Good
IMD + Noise, % | 0.011 | Very good
Stereo crosstalk, dB | -88.1 | Excellent
IMD at 10 kHz, % | 0.029 | Good
General performance | Very good
-
RMAA tests at 24/96, using gaming desktop -> Auzentech Bravura coax out setup, DirectSound mode. In some areas it measures better, in some worse. I might give the overall edge to the Auzen coax out. It seems the Auzen doesn't support KS through the coax out, either that or it just doesn't work with RMAA.
Frequency response (from 40 Hz to 15 kHz), dB | +0.02, -0.91 | Good
Noise level, dB (A) | -91.4 | Very good
Dynamic range, dB (A) | 91.3 | Very good
THD, % | 0.0050 | Very good
THD + Noise, dB (A) | -82.0 | Good
IMD + Noise, % | 0.0094 | Very good
Stereo crosstalk, dB | -88.3 | Excellent
IMD at 10 kHz, % | 0.013 | Very good
General performance | Very good
-
When I first tried the NOS-1704 with my laptop/JKSPDIF setup (what I had used with the Metrum previously), I thought the overall sound was, hmm, perhaps more syrupy(?) than the Metrum Quad. Less focused and less resolving. Much of this is likely that I haven't listened to the Metrum in a week or two, so it could just be me forgetting.
I was honestly surprised when I tried the NOS-1704 out of my Auzentech Bravura's coax out, because I was not expecting much. Why should I? It's a gaming rig and not optimal for audio. Measurements I've taken of sound cards in it before always came out with extra noise and distortion than expected.
The sound in general seemed to just come into focus better than before, though, with the Auzen coax out vs laptop/JKSPDIF. I think my measurements might hint at these differences, especially the jitter measurements. If so, this is one of the first times I've had a chance myself to listen and compare gear first, think I hear differences, and then take a look at measurements (that I happened to produce myself) that might actually validate what I heard. I find that somewhat exciting and cool, ha.
According to RMAA, the Auzen coax out does bring more of a roll-off in the highs and might have slightly higher THD. Not sure why the extra roll-off would be...any ideas?
I'm not sure what exactly is going on behind these findings. I'd have to test the JKSPDIF and Auzen on other gear, including trying the JKSPDIF from other computers. At this point, there's why too little info and too many variables to say the Auzen coax out is simply better. That would actually be pretty funny and sad if so, however. But in the context of my setups and these measurements, the results are certainly interesting.
Files attached:
1. 1KHz square wave response, 24/96
2. Periodic impulse response. Only really posting because it's inverted. I tried a different power cable to make sure I didn't get a reversed on from A-GD, as research showed me this was something that actually used to happen in the past.
3. FR and HD generated from a sweep impulse
4. Zip file of all RMAA tests for the NOS1704 (16/48 through 24/96, KS when I could, DS, etc.).
-
So how does it sound compared to your other DACs?
-
I'm planning on doing listening comparisons tonight and will report back with (likely not the best) impressions. Lately I either spend a lot of time with one set of gear, because I've had new stuff roll in and like to get acclimated with it, or I just take too many long breaks in between listening to make valuable comparisons.
But, as briefly mentioned above, I think the NOS1704 is a bit more syrupy sounding than the Metrum Quad, or perhaps it would be better to say the Metrum is a bit lighter sounding overall. That's not to say it lacks the detail or resolution I'm looking for. I don't remember being as impressed using JRMC w/ upsampling for playback on the Metrum as I am on the NOS1704. XXHE stuck out for me more quickly on the Metrum than JRMC, so I'll have to try that more with the NOS1704. Of course, we'll see if my memory serves me correctly when I actually compare them. It's been a while since I've listened to a S/D DAC setup critically, so I'll test out what I have tonight as well.
That said, I did spend a good bit of time listening to the NOS1704 from the Auzen coax out. I am happy so far...just not so great at putting what I hear into words. I was honestly surprised I was able to hear the differences between the JKSPDIF and Auzen out.
I also still need to get new pads for my Mad Dog, and the HE-500 isn't great for when others sleep (my awake time, hooray). I'll get the HE-500 out tonight for the comparisons, but I'll primarily be testing it with the modded TH-02.
BTW, do my jitter measurements look like I'm doing them correctly?
-
On the jitter, we can see the sidebands which seem high. The most important thing is to get the noise floor low. Like -120 or -130db. We will be more likely to see the sidebands which are close the 12kHz signal. From coax, I would expect to see more of this.
-
I don't have a BNC to BNC cable to try, but I wonder if the BNC input (on DIR9001, not CS8416) would do better with jitter. I've heard many say BNC tends to perform better than RCA when impedance and everything else is matched correctly. I could also try swapping out one of the coax jacks on the CS8416 for a BNC jack. Seems like a popular mod to do on NOS DACs. There's also the two optical ports I could try. I've heard some say NOS DACs are more susceptible to jitter...other say less...others say neither is true. Some say jitter doesn't affect a NOS DAC's performance as much as an OS DAC, some say otherwise. I'm not clear where it actually falls.
No idea if any other parts of my measurements setup or methodologies are falsely portraying jitter performance, good or bad, either. But, given that the JKSPDIF performed worse than the Auzen coax in this test, I'm suspect about the JKSPDIF's true performance and curious if a better chain behind the NOS1704 would also show noticeable improvements. Oh well, I'll have to work with what I have.
Edit: Marv, did my jitter settings look similar to yours, especially the Wnd? I can re-do them if you think I could get more informative results.
-
I took a few hours tonight to do some listening comparisons. The scope of my "tests" and comparisons were bit limited, admittedly, and I do hope to go back with different setups and gear. I might even do a bit more later tonight/this morning. Keeping that in mind, I went for something fairly simple but consistent.
Desktop -> Auzen coax out -> DAC -> Leckerton UHA-6S Mk.II headphone amp -> modded Tascam TH-02
JRiver Media Center, 24/96, upsampling enabled
I know it would have been better to get out the HE-500, but my pair, even with HM5 pads, seems to share many of the same characteristics after mods as Marv's pair show in his measurements. It's a bit on the warmer, bassier side, but it's surprisingly capable of rendering details except where it dips around 4-5KHz. I don't mind the dip there because it really helps me with listening fatigue, but you do lose a noticeable amount of information there. I can't say how closely it matches the response after that point, though. At the very least, it seems to have been enough to make DAC differences mildly apparent. I'll get the HE-500 out sometime when it won't disturb people as much.
For now, I only tried the NOS1704, the Metrum Quad, and the Leckerton's built-in DAC. I plan on trying the SB1240, but I'd have needed to drag out the laptop for that. I think this is my first time I've sat down and really compared DACs, so take this all with a grain of salt.
The first thing I'd like to say is that the NOS1704 is noticeably noisier than the Metrum Quad. You could turn the gain and volume up all the way on the Leckerton with the Metrum as the DAC input, and it would barely pick up noise. Not so with the NOS1704. However, I could not hear the NOS1704's noise until the volume was set well above normal listening levels. I doubt this really had much or any of an impact when listening.
I believe the NOS1704 does indeed have a more syrupy (not necessarily less clear), darker sound than the Metrum. The NOS1704 sounded a more intimate, more densely packed, and less expansive than the Metrum. Instruments and vocalists generally had more of a tangible presence to them than on the Metrum. The NOS1704 seemed better at conveying dynamics and the emotion behind the music.
On the other hand, the Metrum seemed to do a better job putting space between everything. It had a lighter, more expansive, airier sound than the NOS1704. It was often easier to clearly identify and differentiate between instruments, especially fast-paced bass drum kicks during heavy, complex, dense music. Generally, it was the best at allowing me to pull out individual instruments and details.
That wasn't always the case, though. Because it had a lighter, more expansive sound than the NOS1704, some background details were actually more likely to get lost in the mix and a bit hazy despite most other things still being easier to pick out. I think the denser, more up-front nature of the NOS1704 helped bring out some of these smaller details. And while the Metrum provided more space between everything in the sound, I often found the stronger, more immediate presence brought by the NOS1704 to give a slightly more life-like presentation.
At one point in my notes, I mentioned that the Metrum sounded the most sterile of the bunch, but not fatiguing. That's just speaking relatively among these three unites alone, of course. I often noted that I thought the NOS1704 sounded the most natural with cymbals. Overall I gave the nod for clarity to the Metrum. Most times it was too close to really tell between the NOS1704 and Metrum.
When using the DAC in the Leckerton, I thought it sounded good, but it was missing something. Often times it placed itself in the middle of the two DACs sound-wise, but it had this sort of mushy quality to the sound that was hard to describe. Whenever complexity picked up in the music, even in softer genres, it had a tendency to present the least coherent sound of the three DACs. It did not share the separation of space and clarity from the Metrum or the sense of presence and dynamics from the NOS1704. It was as though it tried to capture the good elements of both DACs but instead mostly pulled more of their unique bad traits and mixed them together. I thought it had the least natural presentation of cymbals of the three, but otherwise noticed no oddities among all of them. At one point while listening to orchestral music, I switched to the Metrum from the Leckerton and was more easily able to hear individual instruments that were tightly packed together.
Again, I did think it sounded good and enjoyed listening to it. And, also again, take this all with a grain of salt, especially if anything I say sounds confusing or contradictory (likely falling into the realm of placebo).
I'll try to test out the SB1240 and try some other software/hardware setups. I'd love to add the Beef Roast with Uber AnalGland Upgrade in the mix for comparison purposes, given all the good things I've heard about it lately.
-
Thx for the impressions.
I don't know if I've ever been clear on my impressions of the M7 or not so I'll clarify something. I think the M7 sounds like shit via USB or coax (well shit might be a strong word here). Basically an expensive paper weight that takes up too much desktop space. However, put a proper spec'd OR5 in front and it moves waay higher up the food chain. Not sure how the digital input compares to the NOS1704.
-
I would really like to try the OR5, but I'm not willing to spend that much money on it...yet. I would also be very curious to do jitter measurements with it. Clearly the JKSPDIF wasn't doing the NOS1704 any favors with how I had it setup, at least.
I also can't speak for how similar the M7 and NOS1704 are in terms of design and whether the coax implementation on the NOS1704 is better, the same, or worse.
-
Does the JKSPDIFXYZABC sound better than the Auzen out - from a purely subjective sense?
-
With the NOS1704, I definitely think so. It seems to work better with the Metrum, but I'm in the process of posting additional measurements for that which show it performing
no better probably worse than the Auzen out. I'll have to try the JK with my desktop to verify.
-
I wanted to post some revised measurements. I better adjusted the Windows volume settings for more accurate results, and I also changed the Wnd for jitter tests. Also re-verified the oddities I was seeing with the JKSPDIF and NOS1704.
1. Re-done jitter test with Auzen out, JRMC
2. Above, but with upsampling enabled in JRMC to 24/96
3. With JKSPDIF
4. With JKSPDIF and JRMC upsamling 24/96
5. 1KHz sine spectrum from Auzen out
6. Above but with JKSPDIF
7. 100Hz sine spectrum I forgot earlier, from Auzen out
Also, I re-did the RMAA tests from the Auzen out and got the following (graphs look approx. the same):
Frequency response (from 40 Hz to 15 kHz), dB | +0.02, -0.90 | Good
Noise level, dB (A) | -98.4 | Excellent
Dynamic range, dB (A) | 98.1 | Excellent
THD, % | 0.0027 | Excellent
THD + Noise, dB (A) | -87.6 | Good
IMD + Noise, % | 0.0052 | Excellent
Stereo crosstalk, dB | -95.2 | Excellent
IMD at 10 kHz, % | 0.012 | Very good
General performance | Very good
-
Those jitter measurements don't look "right" - as with what is typically seen.
Which jitter file are you using again? The 16-bit?
-
Those jitter measurements don't look "right" - as with what is typically seen.
Which jitter file are you using again? The 16-bit?
Yeah, I was wondering if my settings weren't what was normally used. I've found it hard to get good info on setting these up. It's the 24/48 JTest file. I used this same method for the new Metrum jitter measurements as well, so those might also need revised. Did the first batch of jitter tests for this look better? Let me know what you think.
Any Wnd setting I tried primarily changed the shape of the response directly around 12KHz...not sure which is best to use or what other settings I might have set incorrectly.
-
Welcome to the party. I hate doing jitter measurements and haven't found a good consistent way to do them. Funny stuff keeps happening with the noise floor, sidebands, etc. One day...
-
Well, if anyone knows the best settings to use, I'd be glad to re-do jitter measurements with them. Always glad to improve my techniques or try new ones. Otherwise, I think what I've done will at least give us a basic picture of a DAC's jitter performance given whatever gear and software is running behind it. Or they're no good and I'm wasting internet space...oh my.
I won't give any sort of final judgement on the JK unit, even with the NOS1704, until I've had a chance to try it with other hardware and some BNC input jacks on DACs. I plan on modding the Metrum and maybe one of the NOS1704 coax jacks that is on the CS8416. I have heard enough people claim this improves reflection jitter to give it a shot. I might also replace the Metrum's output transformer and see if that improves its sense of "presence" relative to the NOS1704.
-
I did a bit more research on window function settings. Unfortunately, I'm finding it very difficult to figure out what settings people generally use for JTest, so what I've read thus far is just general information on the different window settings.
Originally, I had used the flat-top window because it most closely matched the look of other jitter measurements I've seen (i.e. the response starts to arc up at the base of the signal being played, which is 12KHz for the 48KHZ test).
Unfortunately, flat-top isn't the only window setting that produces similar results depending on the gear you're measuring, so I'm really not sure what others tend to use. I'm assuming there is a good reason for using what they do...
But, for the revised tests, I thought it would make more sense to run with whatever window function setting gave the most detailed amount of information. Generally this was the Kaiser5 or 7 window function setting. Otherwise, it seemed that a lot of measurement information would get "buried" (see my original Metrum ARTA spectrum measurements). As such, I also decided to try it with JTest, and it made it easier to see what was going on closer to that 12KHz signal.
Here is something I read about Kaiser windows: "The Kaiser-Bessel window is useful for detecting two signals of almost the same frequency but with significantly different amplitudes." My tests do seem to better display fine details with Kaiser window settings. Got that quote from: http://zone.ni.com/reference/en-XX/help/371361H-01/lvanlsconcepts/char_smoothing_windows/
On the other hand, the SB1240 did not show that arc before and after 12KHz even with the flat-top window, so perhaps such windows are actually used for a particular reason.
As for why the NOS1704 exhibits that dip in noise before and after 12KHz where as the Metrum did not with the Kaiser window setting, I do not know. You can see from my first NOS1704 jitter measurements that the different window setting does not exhibit the same dip. (Keeping in mind I'm looking at the non-JK measurements, which already did funky things.)
In the future, I will probably just measure using a few different windows and post all of them. As far as I could tell, it only really impacted the response immediately surrounding the test tone being played (sidebands stayed the same but might get "buried" when the response arcs up).
Update: The following page shows they use Hanning for their jitter tests. I think that's the default setting in ARTA. It might not matter so much as long as the sidebands and other jitter artifacts are clearly visible. They also have a rough chart for quick approximations to turn dB into ps, but I'm not sure how correct it might be. These measurements of the NOS1704 would likely mean more than 500ps of jitter. The Metrum was able to get below 180ps according to this chart with the same setup.
http://www.lessloss.com/page.html?id=42
-
I did further tests tonight with different DAC inputs, the Auzen vs JK coax out, cables, etc. By running an RCA to BNC coax cable from my Auzen out to the NOS1704, I was able to get less "weird" and generally better results. I found out that adding the 10dB RF attenuator (came with JK) lifted the treble response back up a bit, which was an oddity I originally commented on with my fist Auzen out-based measurements. In some ways the attenuator gave better measurements, and in some ways it made things weirder (ex: lowered sidebands below 12KHz in 24/48 JTest but raised sidebands above 12KHz vs no attenuator).
I'm not sure if this is because I'm running RCA-to-BNC or because the BNC input on the NOS1704 is wired to the DIR9001, while the RCA coax inputs are wired to the CS8416. I will eventually have to replace one of the RCA coax jacks with BNC to find out.
I will also retry the JK with the NOS1704 once I get some BNC-to-BNC cables in.
I should have the new jacks and cables in by next weekend, so I'm hoping to do more tests then. I'll spare everyone a bazillion measurement pictures and stats until I expand on these tests, get more validation on my methods, and decide what would be worth posting in the end.
-
So... I bought your old NOS-1704 from the guy you sold it to. I'm hoping it comes this week, but it could be next because of the holiday.
I'll do the default RMAA tests when I get it, then again when the DI-2014 comes.
Interested to compare results for the exact same dac.
-
Hopefully your ADC isn't weird about sources that don't follow the usual 2V standard.
-
NOS on my M7 is pretty fkd up - phase shift between left and right channels dependent on audio signal frequency and sampling frequency. If you got a DSO you should check it out.
The M7 DSP board is a heap of junk. Going to rip it out and replace with an I2StoPCM board from diya.