Happy Christmas for you John and your familiy, and happy Christmas for all DIYers and digital aficionados especially.
M.
Many thanks on behalf of the DIYers, same to you !!
The presence of filter at the output directly on the dac out, introducing the reactive circuit as a load and phase degradation. If the filter is somewhere else in output section the phase is also significantly damaged. Try to simulate any passive filter and You can see that about the phase. Also, Zanden style multiplied notch sections are introducing ringing in the signal in audio band. I measure it and it could be seen in the simulations. So it is always some compromise in phase and upper Fo. With higher Fo filter, the phase is more accurate, but filter not doing the job...
cheers and happy new Year
cheers and happy new Year
Hi Erin
I have been experimenting with digital audio for over 8 years and there were times I arrived at similar conclusions as you describe. I tested about every imaginable filter topology including the Zanden filter:
Patent US6721427 - Analog filter for digital audio system and audio amplifier for using the same - Google Patents
But when feedback from a group of audiophiles was included one has to come up with something that sounds good to a group of experienced listeners, and that’s a whole different ballgame.
Then it is quite common that many demonstration DACs are rejected and a nerve wrecking interactive design process is started that can last for years before acceptable results can be achieved.
About what you see on the scope and what you hear through the speakers.
Have a look at this Youtube Video:
https://www.youtube.com/watch?v=uIuJTWS2uvY&spfreload=10
Interesting reading on the subject:
http://www.earthworksaudio.com/wp-content/uploads/2012/07/The-world-beyond-20kHz.pdf
http://www.stopthecrime.net/Human Perception FINAL.pdf
ultrasonic
It's Alive! Ultrasonic Spectra Isn't So Ultra Anymore
I have been experimenting with digital audio for over 8 years and there were times I arrived at similar conclusions as you describe. I tested about every imaginable filter topology including the Zanden filter:
Patent US6721427 - Analog filter for digital audio system and audio amplifier for using the same - Google Patents
But when feedback from a group of audiophiles was included one has to come up with something that sounds good to a group of experienced listeners, and that’s a whole different ballgame.
Then it is quite common that many demonstration DACs are rejected and a nerve wrecking interactive design process is started that can last for years before acceptable results can be achieved.
About what you see on the scope and what you hear through the speakers.
Have a look at this Youtube Video:
https://www.youtube.com/watch?v=uIuJTWS2uvY&spfreload=10
Interesting reading on the subject:
http://www.earthworksaudio.com/wp-content/uploads/2012/07/The-world-beyond-20kHz.pdf
http://www.stopthecrime.net/Human Perception FINAL.pdf
ultrasonic
It's Alive! Ultrasonic Spectra Isn't So Ultra Anymore
Hi John,
Happy new year....
I haven't been following but i have reading for the past few weeks to catch up from mk7 era.....
Alot has changed/improved upon but I am confused especially on the i2s/Dem reclocking using masterclock, ecl translation, etc.... Vs the above dem only reclocking.
Can I confirm if there is no more i2s reclocking, only simple cap/resistor dem clocking? I am referring to nos application.
Thanks
Lee
SAA7220 filter characteristics
Hi,
See attached SAA7220 filter characteristics. I was aware of the peak inserted from forum poster comments on the Analogue Devices application note AN207. http://www.analog.com/static/imported-files/application_notes/383160163AN207.pdf
I'm still looking at the charts for the overall characteristics but leaning towards 'worse than I thought'.
It looks like Phillips took the SAA7220 innards and put them into the early stages of the SAA7320 (DAC3). I don't know if any of those actually got to market. Strangely the SAA7320 has the option of 4XI2S (instead of the 1 bit dac) for 'high-performance'.
Hi,
See attached SAA7220 filter characteristics. I was aware of the peak inserted from forum poster comments on the Analogue Devices application note AN207. http://www.analog.com/static/imported-files/application_notes/383160163AN207.pdf
I'm still looking at the charts for the overall characteristics but leaning towards 'worse than I thought'.
It looks like Phillips took the SAA7220 innards and put them into the early stages of the SAA7320 (DAC3). I don't know if any of those actually got to market. Strangely the SAA7320 has the option of 4XI2S (instead of the 1 bit dac) for 'high-performance'.
Attachments
Isn't this SAA7220 curve the slight "lift" that many search for in their non-Oversampling filters?
I have attached the note from the SAA7320 datasheet which I believe includes the Saa7220 filter in it's internals.
I think you mean the sin (x)/x Roll Off. I'm not sure on that one. The SAA7220 datasheet also refers to a +5% DC offset.
http://www.analog.com/static/imported-files/tutorials/MT-017.pdf
Attachments
Hi John,
You've tested more filters than me, thats for sure.
Recently I've done quite a bit of listening to some of my vintage CD players which use different oversampling filters, some use "brickwall filters" and others use more gentle filters.
In my opinion, the oversampling filter sounds cleaner in the high frequncies. When I then swap to my NOS TDA1541A with gentle LC filters, there is clearly more "grunge" in the top end.
But, with the OS filter using brickwall filters, there is an obvious audible phase shift.
Using NOS with gentle filters, the phase shift is not nearly as obvious, I'd go so far to say that its inaudible. But, in comparison to OS, the aliasing in the passband is audible as a loss of detail and focus in the upper mid and high frequncies.
When I swapped to my NOS DAC without filter there is very audible grunge and aliasing in the mids and top end.
I personally cant listen to a filterless NOS DAC anymore, because the aliasing is so bad that I can hear it without even doing back to back comparisons with other players.
Digital is a flawed medium. Its main disadvantage is unnatural sounding high frequncies, no matter which DAC chip is used, and no matter what sort of filter you use (or don't use). I'm just writing about my own feelings on the matter based on my own experiences.
When I compare any DAC or CD player to my record player, the record player wins for natural and extended sounding highs. Analog cant be beaten in this area by digital. (as a side note, the price of the record player has nothing to do with if its a good one or not. If you ever heard an expensive record player / tonearm / cartridge and you didnt like it, then most likely it was a bad combination. A good record player will sound awesome, and will annihilate any digital source IMO.) The natural mids and highs of vinyl can even be heard even when using a fairly cheap cartridge. I don't think a cheap cartridge gives better sound than digital, but it can convey the natural sound that I am talking about.
- But of course I know analog is also a very flawed medium. No need to advise me
I also enjoy both Vinyl and Digital, I wasnt turning this into a this vs that debate to derail the thread.For me the "happy medium" for NOS DACs is using a gentle LC or CLC filter.
I think this audibly reduces aliasing, makes more of the passband sound clean, and the slight phase shift brings more audible benefits than disadvantages.
All what I've said above relates to 44.1Khz material. If the user wishes to upsample to 176.4Khz using their PC, then you could argue that you dont need a filter.
Last edited:
Its a lift but its much slighter, due to this being for 4X OS, not NOS. So only around 0.8dB lift is needed.
John,
In the article you linked here
It's Alive! Ultrasonic Spectra Isn't So Ultra Anymore
A quote from the above article "Recent work by Tsutomu Oohashi et al., published in June of 2000 in the Journal of Neurophysiology, shows that the brain may in fact be registering over-20 (or 22) kHz spectral energy."
This to me is even more reason to use at least a gentle filter.
The question being - do we really want our ears to be subjected to constant 44.1Khz?
Can it be bad for our hearing?
I think it is not ideal.
In the article you linked here
It's Alive! Ultrasonic Spectra Isn't So Ultra Anymore
A quote from the above article "Recent work by Tsutomu Oohashi et al., published in June of 2000 in the Journal of Neurophysiology, shows that the brain may in fact be registering over-20 (or 22) kHz spectral energy."
This to me is even more reason to use at least a gentle filter.
The question being - do we really want our ears to be subjected to constant 44.1Khz?
Can it be bad for our hearing?
I think it is not ideal.
There is 44.1 kHz spectral component only when there is change in the signal. On the other hand the amplitude of the 44.1 kHz component is proportional to the change between the consecutive samples, which is moderate for real life signals. And our speakers are less efficient (and more directional) at this frequency.
How long will that thoroughly debunked Oohashi reference still be banded around?
Last edited:
I want to thank Ceglar for sharing the schematics of his D3a tube output stage in this community. I have built a simplified version with Rod Coleman filament regulators for heating and no decoupling capacitors and it surely beats my former Tube-I-Zator SRPP stage. It sounds like it has very low distortion (I haven't done any measurements) and it is absolutely super quiet and noisefree. First I thought it was a teeny tiny bit too textureless, stiff and polite for my liking, but it has blossomed with the right output capacitors (ERO MKT1813). The 1541A is still up there with the best DAC ever - no doubt about it, and Ceglar made me really see it - and for that I'm very grateful.
Kind regards,
Vidar
Kind regards,
Vidar
You're welcome..
but the credit belongs to Thorsten for the design. Upon some anonymous advice, have removed the 56nF cap and replaced with 2n2 at 1541A analog outputs and 1k grid stopper to 24R plus ferrite bead.
Tubes will last forever at -0.5V, 10mA, 55V plate, max swing 5v6p-p and its nice and linear right down in there on the curves with horizontal load line.
Shane
You're welcome..
Upon some anonymous advice, have removed the 56nF cap and replaced with 2n2 at 1541A analog outputs and 1k grid stopper to 24R plus ferrite bead.
Tubes will last forever at -0.5V, 10mA, 55V plate, max swing 5v6p-p and its nice and linear right down in there on the curves with horizontal load line.
Shane
Hi Shane,
Can you please post a link to that schematic?
Thank you.