Mode B up and running (finally). Switching between the two mods was easy. Now I guess this one needs some burn in? Initial impressions are that DF1704 sounds slightly more artificial than PMD100. I'll give it some serious time now and report back later.
Review of Audio-GD DAC-19MK3
- Thread starter slim.a
- Start date
slim.a
1000+ Head-Fier
- Joined
- Jan 8, 2009
- Posts
- 1,228
- Likes
- 24
Quote:
I remember that it took the DF1704 at least 100 hours of burn-in before it sounded decent ... so I guess a little bit of patience is required
Keep us posted on your findings!
| Originally Posted by Patu /img/forum/go_quote.gif Mode B up and running (finally). Switching between the two mods was easy. Now I guess this one needs some burn in? Initial impressions are that DF1704 sounds slightly more artificial than PMD100. I'll give it some serious time now and report back later. |
I remember that it took the DF1704 at least 100 hours of burn-in before it sounded decent ... so I guess a little bit of patience is required
Keep us posted on your findings!
PrTv
100+ Head-Fier
Me too have just tried Mode B.
The module has around 150+ Hrs and I still prefer Mode A better.
I tried this with both 702 and HD650 (see my sig for a complete system info) and to me, sibilance in Mode B is too much. I don't know if anything will be better after 200+Hr.
The module has around 150+ Hrs and I still prefer Mode A better.
I tried this with both 702 and HD650 (see my sig for a complete system info) and to me, sibilance in Mode B is too much. I don't know if anything will be better after 200+Hr.
Quote:
For now I'm not noticing sibilance with Mode B. I hope it won't show up at all. The difference is actually quite small to my ears. Well I only have 10-15 hours for Mode B so we'll see. I just like that it supports 24/96. Not essential for now but in future it might be.
| Originally Posted by PrTv /img/forum/go_quote.gif Me too have just tried Mode B. The module has around 150+ Hrs and I still prefer Mode A better. I tried this with both 702 and HD650 (see my sig for a complete system info) and to me, sibilance in Mode B is too much. I don't know if anything will be better after 200+Hr. |
For now I'm not noticing sibilance with Mode B. I hope it won't show up at all. The difference is actually quite small to my ears. Well I only have 10-15 hours for Mode B so we'll see. I just like that it supports 24/96. Not essential for now but in future it might be.
slim.a
1000+ Head-Fier
- Joined
- Jan 8, 2009
- Posts
- 1,228
- Likes
- 24
More on the DF1704 filter…
Slow roll-off filter:
In my opinion, some of the negative reviews that the DF1704 got in comparison to the PMD100 (in other DACs than audio-gd) were because the DF1704 has 2 filter options: sharp roll-off and slow roll-off. I suspect that some other brands did not configure the DF1704 in the slow roll-off mode (this is just a speculation).
So I got curious and did a measurement of both filters (PMD100 and DF1704) with the dac19mk3. Looking at the RMAA graphs as well as the DF1704 data sheet, it seems that audio-gd has used the slow roll-off filter in their dac19mk3. Up until 15 kHz, both the pmd100 and df1704 have similar measurements, but the df1704 starts rolling off earlier and is already 1db lower than the PMD100 at 20 kHz. The PMD100 has a sharp roll off at around 21 kHz.
What does it mean and how does it correlate with the sound we hear?
After doing a little bit of research, I found an interesting paper about minimum phase filters done by Ayre.
Link to the paper: http://www.ayre.com/pdf/Ayre_MP_White_Paper.pdf
Looking at the graphs, it seems that the PMD100 is using a filter similar to the one presented in page 1 that exhibits a lot of pre and post ringing in the impulse response.
The DF1704 on the other hand seems to be using the slow roll-off filter on page 2 that exhibits far less pre and post ringing in its impulse response.
(Note: this is only speculation from my part, so I am willing to consider other interpretations/views)
Anyway, I found that this correlates to what I have heard earlier when comparing both filters. The PMD100 seems to blur/slow transients in comparison with the DF1704 which seemed a lot faster and seems to have a better low level resolution.
Optimizing the DF1704
As I said before, I found that the DF1704 lowers considerably the perceived noise floor in comparison with the PMD100. While that means that you can get a lot more low level details, it also means that it will be a lot more difficult to place in a system as it will reveal more flaws about what is in place: quality of the transport, digital cable, power filtration, interconnects, power cords …
Weirdly enough, the fact that the PMD100 has less resolution, I predict that it will be a lot easier to set up and to satisfy musically its users.
I have tried the both digital filters in different set-ups/configurations and I have found out that while the PMD100 sounded between good and very good in all situations, the DF1704 could sound either relatively average or exceptional depending on the configuration.
Up-sampling:
Since I bought the dac19mk3, I gave up on using upsampling as I found while it very often provides an improvement in soundstage coherency as well as sweetening of the sounds, it always comes at the expense of transparency and micro dynamics. Even when using the very good Secret Rabbit Code, I felt that the sound was slightly veiled and kept using the upsampling only when I had to listen to delta sigma based DACs such as emu 0404 usb or the audio-gd FUN.
However, along with the upgrade from the older foobar 0.8.3 to the v0.1, I recently discovered the SoX upsampling which is more transparent than all the other upsampling plugins I tried in my system without requiring a lot of cpu time.
After trying a few different combinations, I ended up preferring the following: 96K upsampling with the minimum phase (instead of linear).
The improvement was very interesting. And while I immediately liked the sound, it took me a while to understand and then to verify what was happening.
First, the most noticeable thing is the slight displacement of the soundstage further back. It gained a little bit in size (mostly in depth) and there seemed to be a better 3D imaging. And there was still plenty of air between instruments and performers thanks to a good rendering of low level details (unlike what I have encountered with other upsampling software that kill the micro details).
The second thing that I noticed was the slight shift in the tonal balance. The highs smoothed out without loosing definition.
The last thing that improved was the way the decay of the sounds was handled. Playing the 1981 Glenn Gould Goldberg Variations, the notes of the piano seemed to linger longer in the air and it was more life like than I ever heard it through my system. With the upsampling, you could distinctly hear the impact/percussion, then the resonance of the string. When listening with the PMD100 to the same track, everything is blurred: it is tonally “pleasing” but all the little details that make the sound realistic are gone and there are fewer nuances between different notes being played.
Transport/digital cables:
I have found the DF1704 a lot more sensitive to the transport and digital cable than the PDM100. Both modules benefit from better sources but the DF1704 ruthlessly expose any fault with the transport/digital cable. I have already written 2 reviews/comparisons about transport and digital cables here: http://www.head-fi.org/forums/f7/usb...hiface-449885/ and here http://www.head-fi.org/forums/f21/re...-cable-469313/ So I won’t go into the details again in this thread.
Power filtration:
Once again, the DF1704 proved to be more sensitive to the power cables/power filters than the PMD100 module. When using a basic power cord such as the olfex and a filterless mains multiplier, the DF1704 can be relatively pretty average sounding with disjointed and artificial highs. In my current system it took a good power cord (the hifi cables & Cie Powertrans Plus) and a good power filter (Bada LB5600 in my case) to extract a great sound from the df1704. However, since the quality of the AC varies from one place to another, there is no way to draw generalizations from my case.
Other tweaks:
Sadly, vibration control devices are the least talked about subjects here on head-fi while they can make significant improvement to the sound.
After trying many configurations, I ended up with a maple platform and either yamamoto footers with the DF1704 or Herbie’s Audio Lab Tenderfoot when using the PMD100.
To wrap-up:
In a revealing system, the DF1704 module can be very difficult to set-up. In fact, it is a lot easier to get a musically satisfying sound from the PMD100. However, once the DF1704 is set-up properly, it gives a more rewarding and faithful listening experience.
In my opinion, the DF1704 is a more transparent digital filter, and in the right set-up, it is also the most tonally accurate. But at the end of the day, personal preferences and system matching will be the main factors in deciding whether to go with the pmd100 or df1704.
Slow roll-off filter:
In my opinion, some of the negative reviews that the DF1704 got in comparison to the PMD100 (in other DACs than audio-gd) were because the DF1704 has 2 filter options: sharp roll-off and slow roll-off. I suspect that some other brands did not configure the DF1704 in the slow roll-off mode (this is just a speculation).
So I got curious and did a measurement of both filters (PMD100 and DF1704) with the dac19mk3. Looking at the RMAA graphs as well as the DF1704 data sheet, it seems that audio-gd has used the slow roll-off filter in their dac19mk3. Up until 15 kHz, both the pmd100 and df1704 have similar measurements, but the df1704 starts rolling off earlier and is already 1db lower than the PMD100 at 20 kHz. The PMD100 has a sharp roll off at around 21 kHz.
What does it mean and how does it correlate with the sound we hear?
After doing a little bit of research, I found an interesting paper about minimum phase filters done by Ayre.
Link to the paper: http://www.ayre.com/pdf/Ayre_MP_White_Paper.pdf
Looking at the graphs, it seems that the PMD100 is using a filter similar to the one presented in page 1 that exhibits a lot of pre and post ringing in the impulse response.
The DF1704 on the other hand seems to be using the slow roll-off filter on page 2 that exhibits far less pre and post ringing in its impulse response.
(Note: this is only speculation from my part, so I am willing to consider other interpretations/views)
Anyway, I found that this correlates to what I have heard earlier when comparing both filters. The PMD100 seems to blur/slow transients in comparison with the DF1704 which seemed a lot faster and seems to have a better low level resolution.
Optimizing the DF1704
As I said before, I found that the DF1704 lowers considerably the perceived noise floor in comparison with the PMD100. While that means that you can get a lot more low level details, it also means that it will be a lot more difficult to place in a system as it will reveal more flaws about what is in place: quality of the transport, digital cable, power filtration, interconnects, power cords …
Weirdly enough, the fact that the PMD100 has less resolution, I predict that it will be a lot easier to set up and to satisfy musically its users.
I have tried the both digital filters in different set-ups/configurations and I have found out that while the PMD100 sounded between good and very good in all situations, the DF1704 could sound either relatively average or exceptional depending on the configuration.
Up-sampling:
Since I bought the dac19mk3, I gave up on using upsampling as I found while it very often provides an improvement in soundstage coherency as well as sweetening of the sounds, it always comes at the expense of transparency and micro dynamics. Even when using the very good Secret Rabbit Code, I felt that the sound was slightly veiled and kept using the upsampling only when I had to listen to delta sigma based DACs such as emu 0404 usb or the audio-gd FUN.
However, along with the upgrade from the older foobar 0.8.3 to the v0.1, I recently discovered the SoX upsampling which is more transparent than all the other upsampling plugins I tried in my system without requiring a lot of cpu time.
After trying a few different combinations, I ended up preferring the following: 96K upsampling with the minimum phase (instead of linear).
The improvement was very interesting. And while I immediately liked the sound, it took me a while to understand and then to verify what was happening.
First, the most noticeable thing is the slight displacement of the soundstage further back. It gained a little bit in size (mostly in depth) and there seemed to be a better 3D imaging. And there was still plenty of air between instruments and performers thanks to a good rendering of low level details (unlike what I have encountered with other upsampling software that kill the micro details).
The second thing that I noticed was the slight shift in the tonal balance. The highs smoothed out without loosing definition.
The last thing that improved was the way the decay of the sounds was handled. Playing the 1981 Glenn Gould Goldberg Variations, the notes of the piano seemed to linger longer in the air and it was more life like than I ever heard it through my system. With the upsampling, you could distinctly hear the impact/percussion, then the resonance of the string. When listening with the PMD100 to the same track, everything is blurred: it is tonally “pleasing” but all the little details that make the sound realistic are gone and there are fewer nuances between different notes being played.
Transport/digital cables:
I have found the DF1704 a lot more sensitive to the transport and digital cable than the PDM100. Both modules benefit from better sources but the DF1704 ruthlessly expose any fault with the transport/digital cable. I have already written 2 reviews/comparisons about transport and digital cables here: http://www.head-fi.org/forums/f7/usb...hiface-449885/ and here http://www.head-fi.org/forums/f21/re...-cable-469313/ So I won’t go into the details again in this thread.
Power filtration:
Once again, the DF1704 proved to be more sensitive to the power cables/power filters than the PMD100 module. When using a basic power cord such as the olfex and a filterless mains multiplier, the DF1704 can be relatively pretty average sounding with disjointed and artificial highs. In my current system it took a good power cord (the hifi cables & Cie Powertrans Plus) and a good power filter (Bada LB5600 in my case) to extract a great sound from the df1704. However, since the quality of the AC varies from one place to another, there is no way to draw generalizations from my case.
Other tweaks:
Sadly, vibration control devices are the least talked about subjects here on head-fi while they can make significant improvement to the sound.
After trying many configurations, I ended up with a maple platform and either yamamoto footers with the DF1704 or Herbie’s Audio Lab Tenderfoot when using the PMD100.
To wrap-up:
In a revealing system, the DF1704 module can be very difficult to set-up. In fact, it is a lot easier to get a musically satisfying sound from the PMD100. However, once the DF1704 is set-up properly, it gives a more rewarding and faithful listening experience.
In my opinion, the DF1704 is a more transparent digital filter, and in the right set-up, it is also the most tonally accurate. But at the end of the day, personal preferences and system matching will be the main factors in deciding whether to go with the pmd100 or df1704.
haloxt
Headphoneus Supremus
- Joined
- Sep 26, 2008
- Posts
- 3,644
- Likes
- 69
You're too tweak-happy, just get a ref 1 so you don't have to worry about digital filters anymore lol.
slim.a
1000+ Head-Fier
- Joined
- Jan 8, 2009
- Posts
- 1,228
- Likes
- 24
Quote:
Half of the fun is the tweaking part LOL
| Originally Posted by haloxt /img/forum/go_quote.gif You're too tweak-happy, just get a ref 1 so you don't have to worry about digital filters anymore lol. |
Half of the fun is the tweaking part LOL
Gotta love your thorough posts slim.a. Informative and helpful.
I'm not having any problems with too sharp high end or stuff like that with DF1704. It's only slightly more aggressive sounding than PMD100. My speaker setup is extremely revealing so I would notice if the sound would've changed to sibilant or harsh. I must say that I expected much greater differences between the two filters.
I'm not having any problems with too sharp high end or stuff like that with DF1704. It's only slightly more aggressive sounding than PMD100. My speaker setup is extremely revealing so I would notice if the sound would've changed to sibilant or harsh. I must say that I expected much greater differences between the two filters.
regal
Headphoneus Supremus
- Joined
- May 27, 2006
- Posts
- 3,645
- Likes
- 19
I'm not going to tell you that I've played with power cords or cables to get the DF1704 to sound "right" but I have compared 24/96 material played via the DF1704 vs the same matterial dithered to 16/44 played with the PMD. I still preferred the PMD. With HDCD material the PMD100 is just plain magical.
noinimod
100+ Head-Fier
- Joined
- Sep 23, 2007
- Posts
- 337
- Likes
- 10
I've went through all 44 pages of this amazing thread but didn't see this mentioned anywhere - is the dac 19mk3/dac19 balanced?
Since some people managed to get kingwa to install/replace other types of inputs, i was wondering if i could get him to do xlr inputs on a dac19
Since some people managed to get kingwa to install/replace other types of inputs, i was wondering if i could get him to do xlr inputs on a dac19
slim.a
1000+ Head-Fier
- Joined
- Jan 8, 2009
- Posts
- 1,228
- Likes
- 24
Quote:
The dac1çmk3/dac19 are single ended designs (not balandced).
XLR outputs on the dac19 : Looking at the way he designs his balanced DACs, I don't think it will be an easy mod as it will require an additional dac chip (per channel) as well as modifying the output stage. However, to make sure, you should better ask Kingwa. He usually answers pretty quickly.
| Originally Posted by noinimod /img/forum/go_quote.gif I've went through all 44 pages of this amazing thread but didn't see this mentioned anywhere - is the dac 19mk3/dac19 balanced? Since some people managed to get kingwa to install/replace other types of inputs, i was wondering if i could get him to do xlr inputs on a dac19 |
The dac1çmk3/dac19 are single ended designs (not balandced).
XLR outputs on the dac19 : Looking at the way he designs his balanced DACs, I don't think it will be an easy mod as it will require an additional dac chip (per channel) as well as modifying the output stage. However, to make sure, you should better ask Kingwa. He usually answers pretty quickly.
jimmychan
100+ Head-Fier
- Joined
- Jun 26, 2009
- Posts
- 371
- Likes
- 243
Sure DAC19 is a single ended designs DAC, DAC9MK3 is a balanced DAC.
Converting to real balanced output is not possible, due to require additional DACs and amplifiers.
Converting to real balanced output is not possible, due to require additional DACs and amplifiers.
I know this thread hasnt gotten any love recently, but today i finally got an amp, my RSA Protector, so i could finally listen to my DAC19Mk3!!! though i dont ahve my jh-13's currently, in the mail from refit, i do have my ATH-A900's and while doesnt make a huge difference, the DAC was in another world, compared to the one in my iPod. I only got to listen for like 15 minutes, but man. I really was doubtful how much of a difference it was going to make. But everything just sounds so much more real. Drums in particular had so much more attack and impact, it also provided more bass weight. Overall i am very happy with my purchase, and cant wait to get my 13's back to listen to what they can do balanced from the dac19 and protector
soberspine
Head-Fier
- Joined
- Sep 6, 2009
- Posts
- 78
- Likes
- 10
is anything that you don't like about the dac? just curious.
Not yet, but honestly, im too busy listening to this beast to think up anymore impressions
Users who are viewing this thread
