Simple, good quality DAC

[Elso Kwak]

Miguel,
This is what I use now:
R3= 3k65
C2= 3n3
???= GND
C4= 2n2 (barely readable)
Non-inverting input of the first opamp NOT connected to ground but to Vref of TDA1543.
I think you want to many things at once. First have IV-converter, one pole of low-pass filter possible, then have optional additional filtering. Currently first opamp= OPA604. This might change to AD8610. (with +/-12V supplies)

[jean-paul]

-3 dB point at 10 kHz

[Elso Kwak]

Originally posted by jean-paul
-3 dB point at 10 kHz

Hi Jean-Paul,
I explained that earlier to you on this forum. Already forgotten?

[jean-paul]

No, not forgotten. I am still confused that you throw such a big portion (>50% ) of the spectrum away. I wouldn't recommend this to others whatever explanation there is.

If cd has to be played this way I will buy a turntable again ( or go back to OS )...

[Elso Kwak]

Originally posted by jean-paul
No, not forgotten. I am still confused that you throw such a big portion (>50% ) of the spectrum away. I wouldn't recommend this to others whatever explanation there is.

If cd has to be played this way I will buy a turntable again ( or go back to OS )...

The filter and discussion I referred to was here:
http://www.diyaudio.com/forums/showt...846#post142846

[tschrama]

Hi,

I am still confused that you throw such a big portion (>50% ) of the spectrum away

Looking at the frequency transfer of an audio system in from a musical cq. harmonic perspective: the top 10KHz to 20KHz is just the top octave from about 10 usefull octaves, so only 10% and probbably the least important 10%. Besides there is still usefull transfer at 20Khz through a 10KHz first order filter so the effect subjectively might not be so bad as the numbers suggest.

Regards,

Thijs

PS. miguel2, don't use ground as the non-inverting input for your I/V stage, use the ref pin of the TDA1543

[ThorstenL]

Koinichiwa,

Originally posted by jean-paul
No, not forgotten. I am still confused that you throw such a big portion (>50% ) of the spectrum away. I wouldn't recommend this to others whatever explanation there is.

If cd has to be played this way I will buy a turntable again ( or go back to OS )...

Get a turntable anyway, But I will say that in this context the TDA1543 (or an even nastier chip) is an essential requirement.

All a matter of taste.

Sayonara.

[Elso Kwak]

Originally posted by jean-paul
No, not forgotten. I am still confused that you throw such a big portion (>50% ) of the spectrum away. I wouldn't recommend this to others whatever explanation there is.

If cd has to be played this way I will buy a turntable again ( or go back to OS )...

Hi Jean-Paul,
Agter a night sleep I decided to come back on the subject.
First your verdict was upgraded from telephone sound to LP-sound. That's quite a improvement and if the sound is as the LP I have reached my goal and I am quite happy.

May I ask you what crossoverpoint (-3dB) filterorder, and filtertype (Bessel, Butterworth, Chebysheff, etc. ) you suggest?. With FilterPro it is quite easy to calculate the values and modify the circuit.

I had the opportunity to compare my DAC with the most expensive JK-Acoustics DAC in 't Harde.
http://www.jkacoustics.nl/DACReference.htm
Johan Ketelaar was quite perplexed that I had such a good sound from a homebrew affair. Last Friday I had on audition in my home a Arcam FMJ CD33 and a Marantz SA-17S1. The Arcam was beter on CD's than the Marantz The Arcam is Euro 2000, the Marantz Euro 2500.
My DAC had more bass but less highs than the Arcam. The Marantz had a anomaly in the bass, just did not have it under control. I decided not to buy either of them, as I preferred the sound of my own DAC/Philips CD-650. To my ears it sounds more musical more like the real thing. A week ago I attended harpsichord and clavichord concerts here in town. I don't hear the exaggerated highs at the live concert I hear from upsamplers and Hifi-ish sets. Also the volume is not that high as most Hifienthusiasts like to play, in fact volume is quite low. I concur with Kuei. Y.W. that it is all a matter of taste but I like to gauge my ears every now and then, before I loose track on the real sound.

I concur with you if you devide 20 K by two you get 10k or 50% of the audioband. I prefer to use a logaritmic scale and then 10k-20k is a very small portion of the spectrum. The filtering I use is choosen after extensive listening tests. As Thijs points out 10k-20k is only 10% of the usable octaves...

[deandob]

I've been looking at a non-os DAC for a while now. Looking around on the net has shown:
- There must be something special about non-os given the DIY interest and the great reviews.
- >10K measurements for a non-os DAC are poor, although it does not seem to affect the sound from what people are reporting.
- The out of audio band artifacts are high and difficult to filter given the 44.1Khz sampling frequency.
- Common concensus seems to be that for os solutions its the digital filter that affects the sound quality.

What about applying a simple sample rate doubler that ups the effective sample rate to 88.2Khz and use an analog filter? The analog filter will be much easier to implement a cut off above 20Khz if the sampling rate is doubled.

Either a chip with 2x upsampling (althugh no longer a non-os solution!!) or a simple DSP to modify the i2s bitstream would work?? I'm sure someone with more experience than me has thought of this, but I'm thinking that this approach might be a good compromise to the problems outlined with non-os.

Feedback?

[Elso Kwak]

Originally posted by deandob
I've been looking at a non-os DAC for a while now. Looking around on the net has shown:
- There must be something special about non-os given the DIY interest and the great reviews.
- >10K measurements for a non-os DAC are poor, although it does not seem to affect the sound from what people are reporting.
- The out of audio band artifacts are high and difficult to filter given the 44.1Khz sampling frequency.
- Common concensus seems to be that for os solutions its the digital filter that affects the sound quality.

What about applying a simple sample rate doubler that ups the effective sample rate to 88.2Khz and use an analog filter? The analog filter will be much easier to implement a cut off above 20Khz if the sampling rate is doubled.

Either a chip with 2x upsampling (althugh no longer a non-os solution!!) or a simple DSP to modify the i2s bitstream would work?? I'm sure someone with more experience than me has thought of this, but I'm thinking that this approach might be a good compromise to the problems outlined with non-os.

Feedback?

Hi Deandob,
I have thought about the upsampler approach but gave up when I realised the upsamplers like CS8420 and AD1890 - 1893 have a digital filter "on board".