Filter brewing for the Soekris R2R

[chfaure]

With txt file and mkrom.exe, you can build dedicated skr for A/B testing choosing which filter to expose with some manipulation.
TNT filter are very natural for red book but the txt file is not provided and A/B testing is almost impossible .

Same with 1021filt_minimum_phase.skr from @gumisb.

 

[gumisb]

I can incorporate C128DP filter or any other but it will not give one opportunity to A/B tests because in original firmware set there is no IIR filters beside DC and CD de-emphasis.
So influence will be strong and rater we achieve something new. Only one structure of IIR filters can exist, it’s not switchable.

 

[TNT]

You could allow yourself to use a 44,1 and 48 IIR bank for quick A/B comparison. There will be a small difference in Fs of course but.... maybe a try? In order to rule out if Fs is determining the SQ, you could swap the IIR banks from 48 <-> 44,1 to see if the impressions remain.

//

 

[gumisb]

You could allow yourself to use a 44,1 and 48 IIR bank for quick A/B comparison. There will be a small difference in Fs of course but.... maybe a try? In order to rule out if Fs is determining the SQ, you could swap the IIR banks from 48 <-> 44,1 to see if the impressions remain.

//

Yes this is solution for IIR switching between banks dedicated for 352,8kHz and 384kHZ or turning off IIR filters but will require feed frequency switching so source up sampling from 44,1kHz to 48kHz for example on player or eventually original material in two frequency. If there will be a will for that I can make it.

Meantime version with C128dp incorporated in current structure as F6 filter (44,1/48kHz only)

 

[TNT]

Probably better to live with the filter for a few days and then switch filters and live with new one for a while and decide. I think this is better. If one cant judge which one is preferred by this method, I personally don't think there is a lot in it for one or the other really.

//

 

[roderickvd]

@gumisb installed that latest pack but kept on listening to F7 / soft all night. All fine *until* I just encountered two seconds of music that is just full of distortion: Spotify 320 kbps, Nita Strauss "Our Most Desperate Hour" from 3:08 - 3:10.

 

[roderickvd]

Well belay that I guess, I just listened to it on my iMac where I heard the same artefacts. So must be a problem with the song itself.

 

[gumisb]

Well belay that I guess, I just listened to it on my iMac where I heard the same artefacts. So must be a problem with the song itself.

Yes. Exactly. I checked this song on Tidal and two different DACs same distortions occur.
Have You checked C128dp combined in to current set?

 

[roderickvd]

Not yet. Like TNT said I'm going with one filter for a couple of evenings, then change to another. I will switch to C128dp in a short while.

Meanwhile listening impressions from any others?

 

[TNT]

You have to try and tell me 😊

Paul’s NewNOS and C128dp they were inspiration. I trying to preserve all good things they have, tonal balance, dynamic, engagement and remove aliasing completely.
There is still a lot to improve especially with upsampling FIR filters and all pass filter. That’s why there are 4 filters to test and explore and share comments.
This kind of filtering it’s bit different from standard FIR Minimum phase filter.

The attenuation of >22k is only 55dB - are you satisfied with this?

The anti-alias IIR filters you deploy, what type, order (15?) and Fo (freq) does it have?

You talked about "external IIR filters" - are these now substituted by these DAM internal IIR? Or do you use both external and internal IIR filets?

//

 

[gumisb]

The attenuation of >22k is only 55dB - are you satisfied with this?

The anti-alias IIR filters you deploy, what type, order (15?) and Fo (freq) does it have?

You talked about "external IIR filters" - are these now substituted by these DAM internal IIR? Or do you use both external and internal IIR filets?

//

If You took that number from picture it's noise floor generated by tone sweep test. Single tone on 21kHz or signal capture show full attenuation.
I don’t have fancy equipment to show more 😊
It’s IIR 15th order filter with cut of frequency just after 20kHz and full attenuation at 22,05kHz constructed from biquads – nothing spectacular.

I’m not using any external filters with this filter set now. This filters set it’s standalone complete solution.

 

[TNT]

OK - thanks!

So 15th order = 90/oct => signal is down 90 dB at 40 000 Hz (one octave above 20k). I see attenuation at 22,05 is approx 50 dB. The theorem says no signals shall be allowed above Fs/2. Just as a comparison, my filters published here are >180 dB down at 22,04 to mitigate aliases. On the other hand its down several dB at 20k but only -0,04 at 17k... trading trading

//

 

[gumisb]

However final results seems to be similar in that aspect.

 

[TNT]

Is that a measure using my filter?

//

 

[boxerfan88]

However final results seems to be similar in that aspect.

Is that measured using REW?

Or was the chart generated by REW from the filter coefficients?

 

[TNT]

Good Q!

//

 

[gumisb]

Yes it’s standard REW measurement form 0 to 24kHz at 48kHz sampling frequency done on Focusrite Scarlett comparing TNT filter with IIR filter.

 

[TNT]

Aha - I'm not sure its my filter. Can you do a command filters while playing 48 - whats the printout?

//

 

[Mayday]

I don't know why, but every time one of these boards comes up for sale at a reasonable price, I don't have the funds.
I'd love to try it and compare it to the TDA1541A, 16xTDA1543 and PCM1794 (don't have a pair of genuine PCM1704 unfortunately).

 

[deerhunt]

I’ve been running my dam1121 a few days now.
Comparing with an AD1865. I like both😊
They are good at different things and as always it a matter of system synergy and personal taste.

I also have a borrowed Denafrips Terminator to benchmark against. The difference is not big but the Terminator has an edge!