PDA

View Full Version : question on DAC roll off filter



earwaxer9
2011-04-13, 09:38
I have been playing with server 7.6. The option of selecting a roll off filter has me thinking. Personally I like the slow filter. Not a big difference.

It has me wondering what this filter setting does. Is "fast" a brick wall type vs. "slow" a non linear? Is this done when the DAC upsamples? What frequencie does it operate on? Is there less "ring" with the slow filter? Any info would be appreciated.

thanks much...

Wombat
2011-04-13, 10:29
It is all in the datasheets and even with pretty detailed data about the corner frequencys. On other DACs you often only see a tiny frequency plot with absolutely no evidence.

earwaxer9
2011-04-13, 19:51
Where do I find the data sheet?

michael123
2011-04-14, 03:46
http://www.asahi-kasei.co.jp/akm/en/product/ak4396/ak4396.html

Wombat
2011-04-14, 04:29
Where do I find the data sheet?

When you look into it and read about the delta-sigma design our Transporter uses youīll notice it has the advantage to already only need pretty gentle filters. Maybe this is why Sean Adams didnīt take the slow roll-off in from the beginning.
Afterwards you may think twice about your advice you give all the time about upsampling all cds while ripping to 24/96 with dbpoweramp...
I donīt have exact numbers but the lowpass alone dbpoweramp uses at upsampling may add more garbage as playing 44.1 natively on our Transporter cause of its delta-sigma design ;)
Please correct me if i am wrong.

earwaxer9
2011-04-14, 09:21
When you look into it and read about the delta-sigma design our Transporter uses youīll notice it has the advantage to already only need pretty gentle filters. Maybe this is why Sean Adams didnīt take the slow roll-off in from the beginning.
Afterwards you may think twice about your advice you give all the time about upsampling all cds while ripping to 24/96 with dbpoweramp...
I donīt have exact numbers but the lowpass alone dbpoweramp uses at upsampling may add more garbage as playing 44.1 natively on our Transporter cause of its delta-sigma design ;)
Please correct me if i am wrong.

Not sure why the upsample sounds better to me... My thought was that it was more due to the nature of the upsample being done beforehand vs. in realtime. Similar to the idea of why (in general) ripped redbook sounds better from a PC hard drive vs. a CD player. I dont really think the dbpoweramp algorithm is the best (SSRC). I have switched to soX using foobar. It seems to be better, and it tests a bit better. The algorithm fascination of late is more academic than anything else, although its surprising that math makes a noticeable difference.

I was trying to get some discussion generated about the upsample phenomenon in regards to the Transporter - hence the "advice you give all the time" about it. From the limited feedback I have gotten so far, it appears there is some agreement to upsampling redbook to 24/96. Thanks for the info!

I havent looked at the data sheets yet. Do that tonight over a couple of beers and some music! I know, I dont have much of a life!http://forums.slimdevices.com/images/icons/icon7.gif

link for SRC test
http://src.infinitewave.ca/

Wombat
2011-04-14, 10:30
I have switched to soX using foobar. It seems to be better, and it tests a bit better.
Pay attention to dither here. The sox plugin doesnīt dither by itself.


From the limited feedback I have gotten so far, it appears there is some agreement to upsampling redbook to 24/96. Thanks for the info!
I wonīt give you positive feedback on that one with my Transporter :)


I havent looked at the data sheets yet. Do that tonight over a couple of beers and some music!
Now that sounds like a good plan, Cheers!!

earwaxer9
2011-04-14, 16:33
Pay attention to dither here. The sox plugin doesnīt dither by itself.


Isnt dither something you only want to do if you are downsampling? Some clarification would be appreciated. I have noticed the dither option. I have not chosen it.

Wombat
2011-04-14, 16:43
Isnt dither something you only want to do if you are downsampling? Some clarification would be appreciated. I have noticed the dither option. I have not chosen it.

Since you have the internal precision of the resampler that is higher as the 24bit signal you have in the end dither makes some sense. Since you upsample to 24bit it shouldnīt be a big problem not to dither. As you already "heard" the benefits of the sox resampler you have even no need to.

earwaxer9
2011-04-14, 17:38
thanks

earwaxer9
2011-04-15, 16:35
http://www.asahi-kasei.co.jp/akm/en/product/ak4396/ak4396.html

Why does the Transporter with the AK4286 not accept 192khz sample rate?

Looking at the filter characteristics on pgs 11 on. It appears there is a significant difference in filter effects (as expected) with 44.1, 96, and 192khz. The frequencies are shifted much higher with the higher sample rates. With resultant improved frequency response and accuracy. Especially with the "slow" filter. Should this not be audible?

Wombat
2011-04-15, 17:41
Why does the Transporter with the AK4286 not accept 192khz sample rate?
The cumputing power inside the Transporter just isnīt up to it and it is for sure hard to always guarantee the datastream. Mr Sean Adams himself somewhere explained that somewhere in detail afaik.


Looking at the filter characteristics on pgs 11 on. It appears there is a significant difference in filter effects (as expected) with 44.1, 96, and 192khz. The frequencies are shifted much higher with the higher sample rates. With resultant improved frequency response and accuracy. Especially with the "slow" filter. Should this not be audible?
One thing to my understanding is for sure, using the slow roll-off with 44.1 gives the HF part that much of a hit the chances are high you hear the missing highs not some theoretical better phase responses. Of cause some will hear that for better. To me some fine sounding recordings sound dead with it. Bad sounding recordings may sound better YMMV
For the other effects of these filterings especially for 96kHz all this is at a very theoretical level brought up by some marketing in the last years. Set the filter what you think sounds best and enjoy. Some 96kHz music you buy is already hard lowpassed and you canīt hear things up there at all.
If you argue now that some bad stuff creeps in from the higher to the lower frequencys someone even should filter everything hard above his hearing capabilities and never wonder about bad sounds creeping down. I think i even have read some eastern high-end studio argueing that way. They filter their 96kHz material at 22kHz!
You see this is all at a very esotheric level and i doubt many claims in this direction.

earwaxer9
2011-04-15, 22:14
thanks wombat!

earwaxer9
2011-04-16, 15:42
If you argue now that some bad stuff creeps in from the higher to the lower frequencys someone even should filter everything hard above his hearing capabilities and never wonder about bad sounds creeping down. I think i even have read some eastern high-end studio argueing that way. They filter their 96kHz material at 22kHz!
You see this is all at a very esotheric level and i doubt many claims in this direction.

It seems to be a trade off between aliasing and destructive pre-ring, with hard filters producing more ring.

Phil Leigh
2011-04-16, 15:46
It seems to be a trade off between aliasing and destructive pre-ring, with hard filters producing more ring.

the pre-ringing is not destructive... it is inaudible...

michael123
2011-04-17, 23:59
the pre-ringing is not destructive... it is inaudible... is it individual..

Phil Leigh
2011-04-18, 01:00
is it individual..

The latest research on this I could find (2004) concluded that there is no proof of audibility or non-audibility of pre-ringing.