Im looking for that exact signature you are talking about having effortless highs and very revealing and yet not bright.
So NewNOS would do that?
The NewNOS better than the stock filter which comes with the DAC?
You can find the NewNOS here:
http://www.diyaudio.com/forums/digi...filter-brewing-soekris-r2r-6.html#post4245752
I don't know if I'd describe the NewNOS as having effortless highs. The description of NewNOS would be "exciting, lively, full bodied". I think it has a gritty sounding mid-range but that doesn't stop people liking it.
Most of the recent filters are quite a bit better than stock. One is like watching a concert projected onto a wall, the other is closer to sitting in a darkened auditorium watching the performers... That obviously over states the contrast but to my mind it's the difference between a large flat projection of sound, vs a more solid three-d representation of the recording.
Have a listen to the EQHQ_quasiapo filter from a couple of posts above, it's what I consider one of the best as of now.
Last edited:
My main gripe with C128_100dp is the wonky soundstage, otherwise I like it a lot. It is lacking height quite a bit compared to some other filters.
The EQHQ is really impressive I have to admit. It brings stuff to the table I had no idea was missing. It surprisingly produces more low level sounds and locates these better in room than previous filters.
Been listening to it since yesterday and it has not produced ear fatigue I get from some other good filters. Female vocals gained something special from it, a sexy silky/fluid quality to these.
Occasionally there is a slight edge in upper octave, but that could easily be my chain or some recordings.
I have been on/off from 128_100dp usually going back to it for the compromises it brings, but now it is properly been replaced by the EQHQ v1 or possibly by its up coming successors.
It could be aliasing/imaging distortion. See if the same is present on the EQHQ_apo/quasiapo. That would rule in/out that aspect of the filter.
Curious how this filter EQHQ_v1 is made.
More curious in 176k and 192k filters. Rather than expected typical X2 interpolation :
....
0,..
0,..
-0,..
-0,..
0,..
0,..
-0,..
-0,..
etc.
They are:
........
-0.000000085739708881205960488131978362858
-0.000000696357571385405655397778448834067
-0.000002653102700794685577477798749113092
-0.000005672123900910470801974179455573477
-0.000004910090543267713981761494329569828
0.00001026456346530849011555296818043459
0.000046191161849048810266165493443324408
0.000080969708595834966047277703182771802
0.0000491141210945471003504469997835713
-0.00011972443424385083266762302978136745
-0.000389646284394925689238686761228791511
-0.000516102953497703806698404527253387641
-0.000127969991194807269441549579802597236
0.000901315097812419052138166453147505308
0.002000012623037677669812772762725217035
0.001902729256557807689664851835686931736
-0.00044362068195039018516506734712834259
-0.004410374137091225416051898378100304399
-0.006987558993317792219168715917021472706
-0.004295064829637136247231232744070439367
........
Sometimes x3, x4 others, even x5.
Is this correct?
More curious in 176k and 192k filters. Rather than expected typical X2 interpolation :
....
0,..
0,..
-0,..
-0,..
0,..
0,..
-0,..
-0,..
etc.
They are:
........
-0.000000085739708881205960488131978362858
-0.000000696357571385405655397778448834067
-0.000002653102700794685577477798749113092
-0.000005672123900910470801974179455573477
-0.000004910090543267713981761494329569828
0.00001026456346530849011555296818043459
0.000046191161849048810266165493443324408
0.000080969708595834966047277703182771802
0.0000491141210945471003504469997835713
-0.00011972443424385083266762302978136745
-0.000389646284394925689238686761228791511
-0.000516102953497703806698404527253387641
-0.000127969991194807269441549579802597236
0.000901315097812419052138166453147505308
0.002000012623037677669812772762725217035
0.001902729256557807689664851835686931736
-0.00044362068195039018516506734712834259
-0.004410374137091225416051898378100304399
-0.006987558993317792219168715917021472706
-0.004295064829637136247231232744070439367
........
Sometimes x3, x4 others, even x5.
Is this correct?
And so it should. In a species that has a hearing range of up to 20kHz or in truly exceptional cases 24kHz, I find the hand wringing about filters not being flat to 192kHz completely and utterly absurd.
As per Julian Dunn/Peter Craven the main benefit from higher sampling rates is relaxation of filtering requirements, not that human species has suddenly developed ultra-sonic hearing ability.
Sorry if I murdered the tooth fairy for anyone...
It is true. I read it.
I think that's not a good idea.
I have made recordings with microphones that respond above 35 kHz. With 24/96 format. High resolution information that we have made effort and money cost and will be lost in this dac.
I've never read anything quite as idiotic as this ^^^^
Post up test results from you doctor authenticating your ability to hear to 35kHz and I'll consider removing you form my ignore list...
Last edited:
It is true. I read it.
I think that's not a good idea.
I have made recordings with microphones that respond above 35 kHz. With 24/96 format. High resolution information that we have made effort and money cost and will be lost in this dac.
Post up test results from you doctor authenticating your ability to hear to 35kHz and I'll consider removing you form my ignore list...
actually don't bother, you've spoilt my morning coffee one time too many.
Back after a short break and a day 'under the car'... What progress! A new favourite, EQHQ_quasiapo, struggling to hear any difference at 44k and 352k. This has to be a first. I'd definitely rest up Paul, or at least enjoy listening to music, fixing the car, mowing the lawn, doing all those jobs you've been putting off etc, until the firmware update comes out offering scope for significant improvement. Will give you a shout about Matlab when I’m ready to have another look. Unable to get anything (which to me at least) appears meaningful from their design GUI tool. I’ve a hunch the command line option is the way to go.
Paul I can only hear to 12KHz ( older guy) but I can easily discern between a preamp with response flat to say 40 kHz and one flat to 15 kHz, believe it or not but there's more to it simply response.
I really appreciate all the work you've done, but I wish it had gone into a dac that was more worthy of all your genius. I have about a dozen dacs, of all persuasions and the dam would be lucky to sit halfway in sound quality.
Although I'd be the first to applaud its value for money, just wish it had lived up to its r2r promise.
The next round of updates my end will likely focus on the on-board power supplies having been very successful having recently started using op-amps plus a few extra parts in place of cathode resistors. Something similar may complement the excellent work put into the filters herein. I've a hunch there is little if anything more to gain with filter mods until we see the next firmware?
I think the EQHQ and especially the quasiapo filter really shows the potential of the DAM. The 44.1 apo/quasiapo filters aren't anything particularly special in terms of how they are built. The main thing is that they target potential noise/distortion from imaging above nyquist and noise energy from the stop band. It's simply a conventional filter pushed as close to the limits allowed by the firmware as possible.
The reason I went this way is primarily because I felt we really needed to understand what the DAM sounded like with a strictly correct filter before we started to try to make "improvements". With extra taps the filter can be tweaked a lot closer to an ideal brick wall filter, and that should bring some improvements. But comparing with Audirvana Plus upsampling there doesn't seem like there is a whole lot more improvement to be gained - unless of course the DAM has more potential SQ in reserve than software upsampling suggests....
I'm going bird watching (
) for a few days over the weekend so I'm definitely going to have some down time.... Currently trying to work out if the boss has totally killed the battery in his car. The joys...With MATLAB/Octave you have to juggle settings if the design won't converge.
Try using tap lengths that are multiples of 8 or 16. You might need to hunt up and down a bit to find something that converges on a solution.
here an example:
