CHANGSTAR: Audiophile Headphone Reviews and Early 90s Style BBS
Lobby => Amp and DAC Measurements => Topic started by: firev1 on November 29, 2014, 03:10:27 PM
-
So I saw this for real cheaps and had to grab it. Firmware has been flashed to V1.5 with the TCM(which I typoed TRM) and FRM filters. Sound wise, I'm getting the feeling that this might be the 9023 implementation and easily competes with the Herus/Concero maybe though I need a direct comparison to be sure. Sounds fairly consistent sounding with changing sampling rates and across different computers once fully warmed up which is a HUGE plus for me.
My impressions of the device as a DAC are the same as the rest in that the TCM filter sounds full bodied and the FRM filter makes piano strikes sound weird. Maybe one mode might be better than the other depending on the music you listen to.
The bad? The driver package is pretty bad, device does not get recognised sometimes on my Win 7 rig which requires me to remove and reinstall drivers in order for device to be recognised as a soundcard.
Also the device requires warm up time to get hot before any listening or evaluation of its capabilities is done. The clocks do take a while to settle and in FRM filter mode, I found it to distort and clip with a full scale 19.1khz tones when the device is cold. Jitter also looks bad when the device is cold as 16 bit and 24 bit spectrum measurements are swamped with noise, it disappears when warmed up. When cold, the device is Sabre done wrong(or normal lol) with some harshness out of box.
Measurements wise, the Geek Out pulps the ODAC and the competition at its price point. The TCM filter measures worse in most areas but looks better in some places than the FRM filter so we can see some trade offs and compromises with filter choices.
Rig:
Acer Win8 Laptop-> ASIO -> GO100-> Short 3.5mm to RCA cable-> EMU0404 via RCA to XLR adapters-> Fujitsu Laptop
Software: Foobar with RAM buffering, WaveGene, WaveSpectra, ARTA
Note: Typoed Time Coherent Mode(TCM) as TRM.
DISCLAIMERS: Not science, YMMV. Also do take note of the rising noise with wide bandwidth measurements of the 0404.
The Geek IEM as of firmware 1.5 which you can flash from the Geek website using the tools from the driver package, has a choice of a fast rolloff minimum phase filter which is the so called TCM filter and a slow rolloff symmetrical linear phase filter which they called its FRM filter. As you can see, the TCM filter has no pre ringing but more post ringing.
(http://www.changstar.com/index.php?action=dlattach;topic=1965.0;attach=8123;image)
FRM reconstruction filter measurements, this test is carried with a 44.1khz sample rate and is based on the DAC testing methods use by John Atkinson at Stereophile . A spectrum of white noise is overlayed with a separate full scale 19.1khz tone. Due to the tone's closeness to the bandwidth of the sampling rate, it stresses out the DAC's aliasing filter to produce a 25khz aliasing product.
At 0dbFS, there seems to be problems with the FRM filter thus we see a lot of spuraie, this measurement was taken when the device is warmed out. When cold, the distortion reaches pretty damn high. Also seen is the filter's performance with a -3dbFS tone. The slow rolloff till 30khz means that the aliasing image is not rejected at all. Note the levels of the tone which means that with this filter, there is some rolloff in the highs.
(http://www.changstar.com/index.php?action=dlattach;topic=1965.0;attach=8125;image)
Same test but with the TCM filter. With a steep rolloff, much of the aliasing image at 25khz is rejected. Distortion profile looks almost like the ODAC's but better.
(http://www.changstar.com/index.php?action=dlattach;topic=1965.0;attach=8127;image)
THD, FRM performs much better, observe the distortion profiles though, also the typical Sabre spuraie.
(http://www.changstar.com/index.php?action=dlattach;topic=1965.0;attach=8135;image)
Copied from below: -90dbFS 1khz Test, this test reveals the low level resolution of the GO100, typically with Sabre devices I have seen so far, whether it be the DX90, ODAC or Herus, there would typically be some spuraie at low levels. This is where the GO pounds the competition, essentially free of any pickets. Only thing is the noise levels are higher than the rest which can be seen in the 24 bit waveform.
(http://www.changstar.com/index.php?action=dlattach;topic=1965.0;attach=8140;image)
24 bit blue, 16 bit red.
(http://www.changstar.com/index.php?action=dlattach;topic=1965.0;attach=8142;image)
-90dbFS sinewaves, 16 bit is the one with 3 modulation steps.
SMPTE and DIN Intermodulation Distortion, FRM mode aces this test, though the TCM mode does not perform too shabbily
(http://www.changstar.com/index.php?action=dlattach;topic=1965.0;attach=8131;image)
(http://www.changstar.com/index.php?action=dlattach;topic=1965.0;attach=8129;image)
CCIF high frequency twin tone tests, again FRM puts out better numbers but we can see some extra spuraie with the FRM filter, probably due to the aliasing image? EDIT: On another note, it came to me that this is FANTASTIC intermodulation distortion performance. TONs better than any device I have tested so far except the EMU0404 DAC section.
(http://www.changstar.com/index.php?action=dlattach;topic=1965.0;attach=8137;image)
Lastly, jitter. Before warm up of the GO100, there is plenty of noise and it looks kinda awful for a modern DAC. Even after full warmup though, skirting of the base can still be seen indicating some low frequency jitter. Not bad but not that good either.
(http://www.changstar.com/index.php?action=dlattach;topic=1965.0;attach=8133;image)
Dunn Jtest with 48khz sampling rate, 24 bit data blue, 16 bit red.
To wrap up, the Geek offers features well above its price point with good engineering chops to back it up as well. If that could be backed by better driver support, it would be a good product.
ADDED 18/12/2014: GEEK internal pics http://www.changstar.com/index.php/topic,1965.msg54628.html#msg54628 (http://www.changstar.com/index.php/topic,1965.msg54628.html#msg54628)
-
Great work thanks for investigating.
-
Haven't had any driver issues on my Win8 machines.
-
I have it plugged it to the Wyrd now, sometimes it does not get recognised till I did the steps above. Had not encountered issues with my Win8 machine... yet.
-
Nice work! Which of those apps are you using to get those nice looking impulse response measurements?
-
Nice work! Which of those apps are you using to get those nice looking impulse response measurements?
Haha! School's copy of Audition CS5.5 when I was still a student. I wish there was freeware to do this but Audacity does not show pre and post ringing. Tones were generated with Wavegene http://efu.jp.net/soft/wg/wg.html EDIT: Perhaps you could feed the impulse into ARTA's Spectrum analyser and use the Time record to visualise the pre and post ringing.
-
I probably have a key for Audition from my school as well.
I mean...I got rid of all my student software keys once I graduated as they asked. *shifty eyes*
-
Thanks for the measurements! Could you also measure the GO with the wyrd? Interested to see jitter figures.
-
Hmmm I did some and found pretty much no difference, guess the limitation is the GEEK's overall implementation rather than USB implementation which by itself is pretty solid.
Update: -90dbFS 1khz Test, this test reveals the low level resolution of the GO100, typically with Sabre devices I have seen so far, whether it be the DX90, ODAC or Herus, there would typically be some spuraie at low levels. This is where the GO pounds the competition, essentially free of any pickets. Only thing is the noise levels are higher than the rest which can be seen in the 24 bit waveform.
Top: 16 bit test red, 24 bit blue
Below: 3 steps 16 bit quantization error is reproduced, 24 bit sinewave looks noisy relative to my EMU unit.
-
The spurious signals look like jitter to me, not harmonic distortion. It is dangerously close to -96 dBFS in TCM case. Might be a software or hardware timing error somewhere, likely the latter, since it's dependent on temperature. Solution would be to replace the crummy oscillator with something actually passable.
Disregard this, I see you've measured it. It is not jitter, but weird case of distortion. Structure and the fact it reduces with "heat up" suggests crossover distortion, not harmonic distortion. (Harmonic distortion tends to increase when the device is warmed up.) I'd check cold start square wave response. This might still be related to the crummy oscillator though.
TCM (ugly buzzword) looks like minimum phase minimax designed lowpass with 1 dB ripple, corner at 20.5kHz, better than 120 dB rejection.
Or rather "I like reverb and ripple more than pre-echo". Not my favorite. I prefer lower passband ripple, such as achievable with least-squares designed filters.
FRM (another ugly buzzword) seems to be a FIR simulation of Bessel first order lowpass using low beta (3?) Kaiser-windowed Sinc, 24 tap. (48 samples)
It is too slow roll off to suppress aliasing. They should try making this 4x longer and use higher beta, say, 8 or even 11.
Aliasing murders both piano and organ music, giving most everything atrocious timbre.
Is there a similar measurement of what they originally had in there?
-
Is there a similar measurement of what they originally had in there?
Interesting advice, I would check the cold start square wave when I have time to. As for original firmware measurements, I haven't done it at all.
As for more subjectiveness, I dislike the FRM filter for piano music and certain cases where a denser tone is needed despite the distortion. Weirdish timbre. TCM's flaws is the loosish bass that gets in the way of other kinds of music. In my case not one filter is a one size fits all solution. LH Labs already mentioned that the filters took up too much computing power though so I wonder if they could really put in what you suggested.
-
Aliasing murders both piano and organ music, giving most everything atrocious timbre.
The Metrum Quad had excellent timbre with piano. It must be full of aliasing, given its NOS design...
-
I'm curious, when you guys talk about proper Piano timbre are you able to distinguish a Steinway from a Yamaha on devices you consider to have excellent piano timbre or do they all sound like Yamahas? I'm just wondering about distinguishing accurate tone versus euphonic color when we talk about these DACs.
-
Guilty as charged for not even knowing my Steinways from my Yamahas. More adapted to hearing diffs in guitars than pianos.
-
I'm curious, when you guys talk about proper Piano timbre are you able to distinguish a Steinway from a Yamaha on devices you consider to have excellent piano timbre or do they all sound like Yamahas? I'm just wondering about distinguishing accurate tone versus euphonic color when we talk about these DACs.
What if we can't distinguish between a Yamaha and a Steinway when they are in the same room playing live for us?
-
What if we can't distinguish between a Yamaha and a Steinway when they are in the same room playing live for us?
No worries, It's just placebo. We all know all pianos sound the same as long as they aren't broken.
-
All instruments sound the same pffff They all read the same frequency!
-
Hey guys,
Casey Hartwell from LH Labs here. This is my first post so take it easy on me ahoy
The Geek Out 100 IEM is currently discounted to $231 for a limited time only through our Indiegogo campaign. Head on over and grab yours while they last.
http://bit.ly/IEMSpecialPricing (http://bit.ly/IEMSpecialPricing)
On another note, I just had this Geek Out 1000 review from TONEAudio splash into my inbox this morning. Give er' a look and let me know your thoughts.
http://www.tonepublications.com/review/light-harmonic-geek-out-1000-dac-and-headphone-amplifier/ (http://www.tonepublications.com/review/light-harmonic-geek-out-1000-dac-and-headphone-amplifier/)
-
Hey guys,
Casey Hartwell from LH Labs here. This is my first post here so take it easy on me ahoy
The Geek Out 100 IEM is currently discounted to $231 for a limited time only through our Indiegogo campaign. Head on over and grab yours while they last.
http://bit.ly/IEMSpecialPricing (http://bit.ly/IEMSpecialPricing)
On another note, I just had this Geek Out 1000 review from TONEAudio splash into my inbox this morning. Give er' a look and let me know your thoughts.
http://www.tonepublications.com/review/light-harmonic-geek-out-1000-dac-and-headphone-amplifier/ (http://www.tonepublications.com/review/light-harmonic-geek-out-1000-dac-and-headphone-amplifier/)
Hey Casey, good to see you guys here. Welcome!
-
Thanks! Looking forward to getting more involved here and being able to answer any questions or concerns you might have.
-
Great to see you here too Casey! Looking forward to the Pulse or Wave when they land.
-
Great to see you here too Casey! Looking forward to the Pulse or Wave when they land.
Pulse has landed. We're shipping about 20+ per day and will ramp production up over the course of this week and next.
Geek Wave is coming along nicely as well. Working on getting an update with pictures and a detailed progress report ASAP
-
Insanely crowded ass board. 2 Xpresso clocks for the audio clocks and 1 XO for the XMOS chip. Outputs are OP1652 into a polyprop cap. 9010k2m chip on board.
-
In my opinion Steinways do sound distinctly different from other pianos, if we're talking grand pianos that is.
There's some piano emulation software that is pretty amazing. http://www.synthogy.com/products/ivorygrand.html
I remember liking Boesendorfer also.
https://www.youtube.com/watch?v=ZaSgLU1O4IU
-
(http://i1.kym-cdn.com/photos/images/newsfeed/000/540/667/795.gif)
-
I always love to see good in-depth objective measurements and feedbacks.
Thank you so much.
Few more information would like to share here.
FRM and TCM are not only DF difference, as you could see in FFT. Even order HD vs Odd ratio is different too. Anti-aliasing filter can not do that.
And TIM/IMD/SPTE are all different too. When I design these 'digital mode', I try to combine different techs into one mode to shape the sound I like. In Da Vinci DAC, I even skip all digital filter in NOS mode and apply Bessel analog filters and analog interpolation to make 44.1k CD music sounds at its best. The measurement result of that is not the best, but sounds right to me.
Again, very impressed for your detail measurement. Thank you!
Larry
-
FRM and TCM are not only DF difference, as you could see in FFT. Even order HD vs Odd ratio is different too. Anti-aliasing filter can not do that.
Yup. Seems something else is going on beside the digital filter. What does FRM and TCM mean?
And TIM/IMD/SPTE are all different too. When I design these 'digital mode', I try to combine different techs into one mode to shape the sound I like. In Da Vinci DAC, I even skip all digital filter in NOS mode and apply Bessel analog filters and analog interpolation to make 44.1k CD music sounds at its best. The measurement result of that is not the best, but sounds right to me.
Seems the Da Vinci DAC is a completely different animal. Besides the possible high frequency roll-off (which may not be a big deal and actually sound plenty good), how much high frequency image rejection are you getting with the Bessel filter? (would be cool to know what the S/H rate is and the order of the filter given NOS implementation)
Also, Hi! and Welcome! :)p6
-
In my opinion Steinways do sound distinctly different from other pianos, if we're talking grand pianos that is.
There's some piano emulation software that is pretty amazing. http://www.synthogy.com/products/ivorygrand.html
I remember liking Boesendorfer also.
https://www.youtube.com/watch?v=ZaSgLU1O4IU
In the US most people only know Steinway as the being the top brand, they do a LOT with marketing, that's not to knock their quality but their marketing efforts does push other worthy brands out of the public's eye. The New York Steinway sound is also a bit brighter and more aggressive than the Hamburg Steinway sound, well, in general it is more aggressive than many European brands. There are some great European brands like Bosendorfer, there is C. Bechstein, Bluthner, and one of my favs is the Fazioli which is not known that well. All are quite pricey. From japan Kawai's Shigeru brand is highly regarded, and some of Yamahas models are quite good. Sviatoslav Richter was a Yamaha artist and he is arguably the greatest pianist of the 20th century. But most Yamaha pianos are not highly regarded.
I once went to a music trade show about 8 years ago in Shanghai, it's the largest trade show of musical instruments in the world and all the big piano names were there, you go to the Bluthner booth and some guy with the last name of Bluthner is working it! The founder of Fazioli was there showing off his 10 ft grand and shorter models. Of course the acoustics were horrible, you can't really get any sense of what each piano sounded like, I'm sure it's not all that different than the audio trade shows where you can't really get an honest audition with so-and-so headphone, etc etc.
-
We are going a bit off topic here but Richter actually played for several years with Bosendorfer imperial (his Bach recordings are on B.I), he later moved to Yamaha for commercial reasons. Yamaha actually owns Bosendorfer..
Bosendorfer was Liszt's fav piano! He had a Steinway too (gift of the company) which was played by Italian pianist Michele Campanella in recent years (you can find his work on Philips LPs).
As for a great recording with Bosendorfer I would suggest Paul Badura Skoda in Valois LPs
Fazioli makes amazing instruments, I heard them on a number of occasions including a series of concerts by Angela Hewitt (who plays exclusively on the Italian pianos).
And going back to the subject, piano sound is one thing my Geek Out (1000) cannot reproduce in a convincing way, especially for the lack of harmonic vibrations coming from the gran codas. No faithful decay in long notes if you know what I mean.
-
We are going a bit off topic here but Richter actually played for several years with Bosendorfer imperial (his Bach recordings are on B.I), he later moved to Yamaha for commercial reasons. Yamaha actually owns Bosendorfer..
Bosendorfer was Liszt's fav piano! He had a Steinway too (gift of the company) which was played by Italian pianist Michele Campanella in recent years (you can find his work on Philips LPs).
As for a great recording with Bosendorfer I would suggest Paul Badura Skoda in Valois LPs
Fazioli makes amazing instruments, I heard them on a number of occasions including a series of concerts by Angela Hewitt (who plays exclusively on the Italian pianos).
And going back to the subject, piano sound is one thing my Geek Out (1000) cannot reproduce in a convincing way, especially for the lack of harmonic vibrations coming from the gran codas. No faithful decay in long notes if you know what I mean.
Oh that is interesting, I did not know that about using Yamaha for commercial reasons, do you know if there is a story behind this?
And back to the subject (slightly), what dacs do you feel perform very well specifically in the reproduction of piano?
-
@Knerian Your favorite pianist (at least judging by your Avatar, tops my preferences too with another couple of less known ones) gave an interview talking about his choice.
http://en.wikipedia.org/wiki/B%C3%B6sendorfer#B.C3.B6sendorfer_Artists
http://dvd.ciao.co.uk/Productinformation/Richter_The_Enigma_DVD__5911109
DACs that reproduce a decent piano are R2R. The ΔΣ ones usually run some algorithms that cancel harmonics in the altar of better measurements (low distortion) and tend to sound like ...
To name a couple I would say TOTALDac, dCS (both very $$$). I have a decent good analog front end and I do my serious listening with LPs. If there is interest we could open a dedicated thread for good piano recordings.
-
And going back to the subject, piano sound is one thing my Geek Out (1000) cannot reproduce in a convincing way, especially for the lack of harmonic vibrations coming from the gran codas. No faithful decay in long notes if you know what I mean.
I believe that would be from the low end bass distortion that gives the 1000 that fuller/warmer sound. You need precision and clarity as well as dynamic swing to render those harmonics.
I also agree that Steinways are a bit bright and on the analytical yet fun sound on the spectrum. Quick and playful. I talked to Tian Jiang whom I saw perform on a Yamaha awhile back and asked him what his preferred Piano was. He only said that he was Steinway contracted, lol. They do have impeccable quality.
-
That's kinda my beef with the GO but then again nothing really does it right at the price.
Also, Ahoy Larry! Glad you liked it.