pupsik46 said:With help oxake, I could manage to work my dac.
Unfortunately AN207 type schematics works with SAA7220, but not with PCM2706.
The only difference between the I2S output of the SAA7220 and the PCM2706 or the SAA7210, sample rate aside, is that BCK from the '7220 is a multiple of 32Fs and from the other two BCK is 64Fs. Using AN207 and the Japanese circuit, the difference is a problem easily overcome.
Below is 16bit I2S data.
An externally hosted image should be here but it was not working when we last tested it.
Delaying WS with two flips will turn the data into fully right justified data and as the data is valid on the rising edge of SCK and the delay is in increments of one data bit period, i.e. falling edge to falling edge, SCK to the flip-flop's is inverted as in AN207
(Two flip flops and one 2-input nand gate wired as an inverter)
An externally hosted image should be here but it was not working when we last tested it.
If the data is right justified, one need only stop the clock for the left channel and delaying WS by 16 cycles will accomplish this.
(16 flip flops)
An externally hosted image should be here but it was not working when we last tested it.
As the delay has effectively inverted WS, it has to be uninverted.
(One 2-input nand gate wired as an inverter)
An externally hosted image should be here but it was not working when we last tested it.
Now the clock to the left channel has to be gated off. That is done by NANDing the inverted SCK with the delayed WS.
(Two 2-input nand gates with one wired as an inverter)
An externally hosted image should be here but it was not working when we last tested it.
Hi,
I tryed the set up with CS8414 and PCM56P-K
from the beguining of thread,
that simple one...
but with the just one PCM per ch.
*
I must say that I was quite supprised with
quality of the reproduced sound...
.
I tryed the set up with CS8414 and PCM56P-K
from the beguining of thread,
that simple one...
but with the just one PCM per ch.
*
I must say that I was quite supprised with
quality of the reproduced sound...
.
And all this should be done with CPLD.
What kind of simulating software did you used for timing diagrams and can you email source files for longer brainwork🙂
PS. You should correct data, on diagrams are more than 16bit.
pupsik46 said:
or a 74HC00 and 3x 74HC164.
Xilinx Foundation 2.1i. Been discontinued a long time ago.
It is correct for 32bit I2S L & R data frames. The SAA7220 has 16bit data frames for each channel.
Harman Kardon HD760 with PMD100 and 4 x PCM1702 + 2 x hardware os ??? ( HC164 )
Square from CD:
Filter of this player could be a little bit steeper.
Test signal -60dB shows some hf.
FFT 32000 points center freq. 352,8 kHz @ -46dB:
Zoom in, window 3,9kHz:
HD760 low level distortion 1 kHz -60db, worst harmonic -55dB, other channel not better.
Philips CD960 , Saa7220B, square from CD:
Non os with steep lpf:
Square from CD:
Filter of this player could be a little bit steeper.
Test signal -60dB shows some hf.
FFT 32000 points center freq. 352,8 kHz @ -46dB:
Zoom in, window 3,9kHz:
HD760 low level distortion 1 kHz -60db, worst harmonic -55dB, other channel not better.
Philips CD960 , Saa7220B, square from CD:
Non os with steep lpf:
Ringing
Hi Bernard, I don't want to explain this again but I do:
All those digital filters have ringing on the squarewave as your pictures amply demonstrate.
If you want no ringing go NON-OS and build a Bessel filter or a passive low-pass filter comprised of a resistor in series with the signal and a cap to ground. Of course then HF attenuation of noise is worse as a Bessel does not have a steep initial response ("soft knee") and the passive is even worse in this respect.
You cannot have your cake and eat it all in other words you cannot build a brickwall filter or any steep filter and have no ringing.😎
Bernhard said:Harman Kardon HD760 with PMD100 and 4 x PCM1702 + 2 x hardware os ??? ( HC164 )
Square from CD:
Filter of this player could be a little bit steeper.
Test signal -60dB shows some hf.
FFT 32000 points center freq. 352,8 kHz @ -46dB:
Zoom in, window 3,9kHz:
HD760 low level distortion 1 kHz -60db, worst harmonic -55dB, other channel not better.
Philips CD960 , Saa7220B, square from CD:
Non os with steep lpf:
Hi Bernard, I don't want to explain this again but I do:
All those digital filters have ringing on the squarewave as your pictures amply demonstrate.
If you want no ringing go NON-OS and build a Bessel filter or a passive low-pass filter comprised of a resistor in series with the signal and a cap to ground. Of course then HF attenuation of noise is worse as a Bessel does not have a steep initial response ("soft knee") and the passive is even worse in this respect.
You cannot have your cake and eat it all in other words you cannot build a brickwall filter or any steep filter and have no ringing.😎
Re: Ringing
Hi,
I did not say that I want "no ringing"
All I want to show is that with os filters the ringing occurs before the transient while with an analog filter it cames after the transient and perhaps that sounds better even if the ringing is a little stronger.
Which does not mean that I would not prefer a "no ringing" bessel or transformer filter if it sounds better.
And the PMD isn't any better than a SAA or YM.
QSerraTico_Tico said:
Hi,
I did not say that I want "no ringing"
All I want to show is that with os filters the ringing occurs before the transient while with an analog filter it cames after the transient and perhaps that sounds better even if the ringing is a little stronger.
Which does not mean that I would not prefer a "no ringing" bessel or transformer filter if it sounds better.
And the PMD isn't any better than a SAA or YM.
Re: Re: Ringing
It would have been a lot easier on the discussion if you said straight away what you had to say.
Personally I don't like ringing at all. I tried third order analog Butterworth filters and Bessel filters after NON-OS DAC.
The former has ringing, the latter not. I preferred the sound of the Bessel.
Texas Instruments has an application program for designing low-pass filters, called FilterPro(TM) and the article has nice examples of square waves illustrating the various filter topologies.
See last page!
Bernhard said:
It would have been a lot easier on the discussion if you said straight away what you had to say.
Personally I don't like ringing at all. I tried third order analog Butterworth filters and Bessel filters after NON-OS DAC.
The former has ringing, the latter not. I preferred the sound of the Bessel.
Texas Instruments has an application program for designing low-pass filters, called FilterPro(TM) and the article has nice examples of square waves illustrating the various filter topologies.
See last page!
Yes, I knew it already...Bernhard said:
But what's your point? What are the sonic implications?
😕
Re: Re: Re: Ringing
And how do you like digital filters ?
I made non os mod on a few players with BB, Philips and Sony DACs and sound was always better, except for TDA based players.
QSerraTico_Tico said:
And how do you like digital filters ?
I made non os mod on a few players with BB, Philips and Sony DACs and sound was always better, except for TDA based players.
Re: Re: Re: Re: Ringing
Don't concur with your findings but YMMV.
http://www.sakurasystems.com/articles/Kusunoki.html
Bernard, I hate digital filters.Bernhard said:
Don't concur with your findings but YMMV.
http://www.sakurasystems.com/articles/Kusunoki.html
Re: Re: Re: Re: Re: Ringing
The figures are, at the very least, open to question.
QSerraTico_Tico said:
The figures are, at the very least, open to question.
Re: Re: Re: Ringing
I'm willing to bed real money you wouldn't be able to hear the difference in a proper blind test.
Do you mean bandlimiting by a transformer? How about the transformer nonlinearities and hysteresis? That's sounds like a terrible idea.
What is this nonsense? The ringing waves are frequencies comprising the square wave. They're still there in a perfect square wave that does not look ringing, just hidden by all the higher (to infinity) frequencies that it takes to have a true square wave. When your ear decomposes sound into the frequency domain, it is going to hear those sine components of the ringing regardless of whether you can see them in the image, or whether they are covered up by the addition of the rest of the frequencies to infinity (the latter above 20 kHz the ear will filter away anyhow, but they will cause intermodulation of the audible stuff in the analog parts of the DAC).
Re: Re: Re: Re: Ringing
Referring to the Texas Instruments article the purpose of low-pass filtering is to get rid of frequencies above the crossover point. In the article example that is 20kHz and the square wave is 1kHz. So with 5th order filtering there is not much above 20kHz. Yet the Bessel filtered square wave looks pretty nice.
Nixie said:
Referring to the Texas Instruments article the purpose of low-pass filtering is to get rid of frequencies above the crossover point. In the article example that is 20kHz and the square wave is 1kHz. So with 5th order filtering there is not much above 20kHz. Yet the Bessel filtered square wave looks pretty nice.
Re: Re: Re: Re: Re: Ringing
How it looks is irrelevant. A word such as 'nice' is meaningless in this context.QSerraTico_Tico said:
Re: Re: Re: Re: Re: Re: Ringing
🙄
Thanks for the good news! I feel much better now.Nixie said:
🙄
This thread is becoming hollow.
Off topic: Managed to repair the angry 304mk2, three foults on earthing of servo pcb. The primitive done through-metalisations where bad and others.
Tomorrow i get listeners (and they are used to good gear), have to set up one and another. Am curious what they have to comment about the pcm's
Off topic: Managed to repair the angry 304mk2, three foults on earthing of servo pcb. The primitive done through-metalisations where bad and others.
Tomorrow i get listeners (and they are used to good gear), have to set up one and another. Am curious what they have to comment about the pcm's
- Status
- Not open for further replies.