Hi,
So does using a platform that offers only 2 audio channels via I2S.
Op-Amp's are hardly needed.
Anyway, if you want to handle a crossover in software and all that you need something different to what we are discussing here, I'd say.
Ciao T
So does using a platform that offers only 2 audio channels via I2S.
Guys we can't stand still, you want opamp crossover or what?Buying 3kg megabuck inductors?
Op-Amp's are hardly needed.
Anyway, if you want to handle a crossover in software and all that you need something different to what we are discussing here, I'd say.
Ciao T
OK thanks,
i understood correctly, i was curious if you had found a solution with these two clocks (one for usb receiver an one for spdif slaved receiver) synchronized by any means.
I was thinking about something like one Xtal osc for the slaved spdif receiver and a dds or pll synthesized clock for the musiland to mantain phase relationship between the two clocks.
Maybe this isn't feasible or anyway not so important.
Ciao
Andrea
Hi,
Sure. Secondary PLL. But strictly for DIY the other solution performs acceptable.
Ciao T
Sure. Secondary PLL. But strictly for DIY the other solution performs acceptable.
Ciao T
Alright, thanks. That was my initial thoughts also. I was going for Musiland as a closest match back in GB thread, but having some doubts till now. Biggest question still remains, can this 1 clock concept (as attached or with similar mods) come close to 2 clocks, that we all would prefer ?
Money wise musiland is still a very good price/feature performer. Manufacturer was not interested in 150psc. deal, but some reliable ebay'er sure will. So ~45 USD + shipping is achievable imho.
Attachments
2osc vs 1osc ,IMO: jitter wont really matter,
1.unless you mess up spdif
2.unless you have a 1bit dac
3.also nosdac have demonstrated that bits beyond 12 are subjective at best, music sounds good without the rest too
So we expect we trade our good jitter numbers for better bits , due async usb but I just dont expect much difference. You wont have a digital crossover either. Thats no gain over a 100 usd soundcard, just a cheap transport without handy features...
I think this one http://www.amazon.com/Creative-Professional-0202-USB-2-0/dp/B000IXLH6K/ref=cm_cr_pr_product_top
has precisely 2 clocks , plus ...
Thorsten: I have dice2 for a while now i have as many i2s as i want to
1.unless you mess up spdif
2.unless you have a 1bit dac
3.also nosdac have demonstrated that bits beyond 12 are subjective at best, music sounds good without the rest too
So we expect we trade our good jitter numbers for better bits , due async usb but I just dont expect much difference. You wont have a digital crossover either. Thats no gain over a 100 usd soundcard, just a cheap transport without handy features...
I think this one http://www.amazon.com/Creative-Professional-0202-USB-2-0/dp/B000IXLH6K/ref=cm_cr_pr_product_top
has precisely 2 clocks , plus ...
Thorsten: I have dice2 for a while now i have as many i2s as i want to
Last edited:
I think having a 44.1khz multiple clock is a fundamental requirement of any audiophile grade transport, otherwise it is just a cheap DVD player.
More data: I have an EMu 0404 PCI sound card (which has 2 clocks) and a Hiface MTech (which has 2 clocks), guess which one sounds significantly better? The EMU. There is a reason studios use PCI cards, USB has a long way to go.
More data: I have an EMu 0404 PCI sound card (which has 2 clocks) and a Hiface MTech (which has 2 clocks), guess which one sounds significantly better? The EMU. There is a reason studios use PCI cards, USB has a long way to go.
Practically, I still don't understand the need and the purpose of a USB to I2S interface.
Why not hooking on S/PDIF using a transformer for galvanic isolation, using a isolated DC/DC converter, and feeding a CS42518 containing : a S/PDIF input with advanced PLL clock recovery, 2 x 24-bit differential ADCs and 8 x 24-bit differential DACs with Digital Output Volume control. 192 kHz compatible. 114 ADC dynamic range. 110 dB DAC dynamic range. -100 dB THD + N. The clock source can be the recovered S/PDIF clock, or can be a Quartz clock.
The CS42518 datasheet is here : CS42518/28 - 110/114 dB, 192 kHz, 2-in, 8-out Multi-Channel CODECs with S/PDIF Receiver
You will reply : "that's crazy, completely overkill, because you only use 1/8th of the CS42518 functions".
Now, let me explain :
1.
The CS42518 exhibits a better performance than most oldschool separate ADCs and DACs.
2.
The CS42518 has a Digital Audio Processor Interface (pins SAI-SDOUT, SAI_LRCK, SAI_SCLK, CX-LRCK, CX_SCLK, CX-SDOUT, CX_SDIN1, CX_SDIN2, CX_SDIN3, CX_SDIN4).
3.
The CS42518 has a Microcontroller Interface (pins INT, RST, and a SPI or I2C bus)
4.
It means that once you have this CS42518 installed as galvanically isolated, high quality S/PDIF stereo DAC, all you need for stepping into a galvanically isolated digital crossover, is to hook a DSP on the Digital Audio Processor Interface, as daughterboard.
5.
The big advantage of such (overkill ?) approach is to enable a honest comparison between "simple stereo" and "4-way active crossovers" configurations, using the exact same S/PDIF, clock management and DACs.
6. If you are in search of a neat 24-bit fixed-point with 56-bit accumulators DSP to be hooked on the CS42518 Digital Audio Processor Interface, as daughterboard, please have a look to : http://www.diyaudio.com/forums/digital-line-level/167328-dsp-xover-project.html
Regards,
Steph
Why not hooking on S/PDIF using a transformer for galvanic isolation, using a isolated DC/DC converter, and feeding a CS42518 containing : a S/PDIF input with advanced PLL clock recovery, 2 x 24-bit differential ADCs and 8 x 24-bit differential DACs with Digital Output Volume control. 192 kHz compatible. 114 ADC dynamic range. 110 dB DAC dynamic range. -100 dB THD + N. The clock source can be the recovered S/PDIF clock, or can be a Quartz clock.
The CS42518 datasheet is here : CS42518/28 - 110/114 dB, 192 kHz, 2-in, 8-out Multi-Channel CODECs with S/PDIF Receiver
You will reply : "that's crazy, completely overkill, because you only use 1/8th of the CS42518 functions".
Now, let me explain :
1.
The CS42518 exhibits a better performance than most oldschool separate ADCs and DACs.
2.
The CS42518 has a Digital Audio Processor Interface (pins SAI-SDOUT, SAI_LRCK, SAI_SCLK, CX-LRCK, CX_SCLK, CX-SDOUT, CX_SDIN1, CX_SDIN2, CX_SDIN3, CX_SDIN4).
3.
The CS42518 has a Microcontroller Interface (pins INT, RST, and a SPI or I2C bus)
4.
It means that once you have this CS42518 installed as galvanically isolated, high quality S/PDIF stereo DAC, all you need for stepping into a galvanically isolated digital crossover, is to hook a DSP on the Digital Audio Processor Interface, as daughterboard.
5.
The big advantage of such (overkill ?) approach is to enable a honest comparison between "simple stereo" and "4-way active crossovers" configurations, using the exact same S/PDIF, clock management and DACs.
6. If you are in search of a neat 24-bit fixed-point with 56-bit accumulators DSP to be hooked on the CS42518 Digital Audio Processor Interface, as daughterboard, please have a look to : http://www.diyaudio.com/forums/digital-line-level/167328-dsp-xover-project.html
Regards,
Steph
Hi,
Why?
If I derive my clock via a PLL from (say) 27MHz but get lower jitter than a 44.1KHz clock, why does it make it "just a cheap DVD-Player"?
Surely, what matters is the result, not how it is attained?
May I enquire why you conclude from comparing two specific pieces of hardware (of which is is known that one has a output not designed in accordance with the SPDIF specification) that any conclusion may be made about the fundamental capabilities of a given interface?
Ciao T
Why?
If I derive my clock via a PLL from (say) 27MHz but get lower jitter than a 44.1KHz clock, why does it make it "just a cheap DVD-Player"?
Surely, what matters is the result, not how it is attained?
May I enquire why you conclude from comparing two specific pieces of hardware (of which is is known that one has a output not designed in accordance with the SPDIF specification) that any conclusion may be made about the fundamental capabilities of a given interface?
Ciao T
Agreed but show me the measurements, theoretically I don't see this as possible.
We are at the cusp of USB audio being satisfactory , mearly showing that we aren't quite there yet at least with <$200 commercial products. People are assuming that a $180 USB product is superior to PCI in general and this is just not the case at all. USB audio has had issues from ground zero compared to PCI.
Remember when people first got their new DAC with the USB input, how we were all in awe, only to discover that the SPDIF input sounded much much better? We haven't crossed over this yet except possibly for some very expensive implimentations.
Fill me in on the product that doesn't meet SPDIF specification, very interested.
Thanks
Hi,
Maybe stick to theory then? In practice noisy PSU's for dedicated single frequency oscillators can make for much worse jitter than a well implemented oscillator and PLL.
I'll repeat, it is all in the implementation.
Assuming the PCI card is powered from the PC and the USB device is bus powered both have horrific issues and are next to useless. BTW, I have tried the 0404PCI. Very bad SPDIF output, worse than that of my motherboard (which is arguably modified)!
Not sure what you are talking about. SPDIF inputs build around the CS841X series from SCirrus Logic are universally disaster areas. If well implemented even the BB PCM2904 has lower jitter and that thing has ton's.
Not neccesarily. The big problem is that > 90% of all USB products use USB bus power to supply the circuitry. To remedy this is not expensive.
The fact that they pretty much all use cheap Pierce oscillators makes the problem worse.
Decent DPLL Systems can now keep the added jitter well below 100pS if they get powered with a clean supply and the reference clocks are low jitter.
One of the two has very high levels, way higher than SPDIF Spec. If you do not know which how can you even bother testing? Did you not do a basic hardware test of each device before putting them in your system?
I always put a 'scope on the output (correctly terminated) and generally do a digital loopback into one of my test systems (EMU 1616m based) to verify that levels and waveshapes are okay and that the whole system is bitperfect/transparent.
From experience with many devices and eval PCB's I can assure that such tests are mandatory. Most fail on some or several account. The problem I suspect is that these devices are designed by computer people, who do not understand the nature of SPDIF and Audio, as many of the design practices I see are fine on a PC Motherboard or a Video card, but inappropriate for audio.
I would not bother to draw any inferences from devices that "defective by design", until these problems have been fixed By suitable and often simple) modifications.
Ciao T
Maybe stick to theory then? In practice noisy PSU's for dedicated single frequency oscillators can make for much worse jitter than a well implemented oscillator and PLL.
I'll repeat, it is all in the implementation.
Assuming the PCI card is powered from the PC and the USB device is bus powered both have horrific issues and are next to useless. BTW, I have tried the 0404PCI. Very bad SPDIF output, worse than that of my motherboard (which is arguably modified)!
Not sure what you are talking about. SPDIF inputs build around the CS841X series from SCirrus Logic are universally disaster areas. If well implemented even the BB PCM2904 has lower jitter and that thing has ton's.
Not neccesarily. The big problem is that > 90% of all USB products use USB bus power to supply the circuitry. To remedy this is not expensive.
The fact that they pretty much all use cheap Pierce oscillators makes the problem worse.
Decent DPLL Systems can now keep the added jitter well below 100pS if they get powered with a clean supply and the reference clocks are low jitter.
One of the two has very high levels, way higher than SPDIF Spec. If you do not know which how can you even bother testing? Did you not do a basic hardware test of each device before putting them in your system?
I always put a 'scope on the output (correctly terminated) and generally do a digital loopback into one of my test systems (EMU 1616m based) to verify that levels and waveshapes are okay and that the whole system is bitperfect/transparent.
From experience with many devices and eval PCB's I can assure that such tests are mandatory. Most fail on some or several account. The problem I suspect is that these devices are designed by computer people, who do not understand the nature of SPDIF and Audio, as many of the design practices I see are fine on a PC Motherboard or a Video card, but inappropriate for audio.
I would not bother to draw any inferences from devices that "defective by design", until these problems have been fixed By suitable and often simple) modifications.
Ciao T
Hi Thorsten
What measured jitter values are we talking ? Must be a lousy 4.1 clock.
best
Hi Guido,
Well, the kind commonly fitted to digital gear. You know, Pierce oscillator with cheap Xtal and noisy supply. These easily show a few 100pS...
BTW, PLL's with much better performance than I suggested exist as well.
Ciao T
Well, the kind commonly fitted to digital gear. You know, Pierce oscillator with cheap Xtal and noisy supply. These easily show a few 100pS...
BTW, PLL's with much better performance than I suggested exist as well.
Ciao T
Hi Thorsten,
A synthesized clock will at maximum be as good as the crystal clock it is based upon, what I tried to say is that synthesizers are no good for serious audio.
I am thinking about a module with a variety of audio clock frequencies, a synthesizer and 27.000 MHz clock. Such module will effectively upgrade the clocks of DV or other multi format players.
best
Guido,
I'd disagree. As long as the synthesizer is good enough not to compromise the performance of the system into which it is designed (a 1pS Jitter Clock does not make a 1pS jitter CD-Player or DAC) I think it is eminently suited for serious audio.
Ciao T
I'd disagree. As long as the synthesizer is good enough not to compromise the performance of the system into which it is designed (a 1pS Jitter Clock does not make a 1pS jitter CD-Player or DAC) I think it is eminently suited for serious audio.
Ciao T
Thorsten,
That is not what I wrote, my statement is that synthesizers add jitter, I am not discussing system requirements.
Have you actually measured the jitter on the clock output of a synthesizer and compared that with the jitter of the oscillator feeding the synthesizer?
best
Measuring jitter at the transport output is pretty meaninless, I mean its how the jitter gets spread across the frequency spectrum after the DAC that determines its impact on sound?
The lowest documented transport jitter I believe was just under 1nS (some $20k Boulder transport.) So I can see how your argument that 100 ps of jitter is insignificant, the issue is when converted to analog does that extra 100ps of jitter all center at 1khz? That would be a problem.
Hi,
I was not talking about transports, but integrated players.
A low jitter clock is shure nice too have, but it needs a squeaky clean supply to with really low noise. And then it needs to be injected into the system in ways that overcomes on-chip ground-bounce in the lead-frame and bondwire and other sources of what Ed called Logic Induced Modulation), all of which conspire mightily against having the 1pS Jitter from your clock arrive at the output of the DAC.
And common receivers from cirrus logic do not attenuate this jitter at all until we reach two digit KHz frequencies.
Well, it is worth looking at synthesized clocks jitter spectrum, they usually display a double-wing shape. Further, the PLL Synthsizers tend to be quite good in cleaning up LF jitter from the VCO within their bandwidth, if designed right.
Nowadays the best syntesised clocks manage < 50pS Jitter in the baseband up to 20KHz for "budget" devices. Mil Spec stuff can be better. And usually the nature of jitter in such system is noise, meaning it broadens the central peak of measured spectrum of (for example) a J-test spectrum, instead of producing distinct deterministic jitter, something which I think is less of a subjective impairment than deterministic jitter.
Ciao T
I was not talking about transports, but integrated players.
A low jitter clock is shure nice too have, but it needs a squeaky clean supply to with really low noise. And then it needs to be injected into the system in ways that overcomes on-chip ground-bounce in the lead-frame and bondwire and other sources of what Ed called Logic Induced Modulation), all of which conspire mightily against having the 1pS Jitter from your clock arrive at the output of the DAC.
And common receivers from cirrus logic do not attenuate this jitter at all until we reach two digit KHz frequencies.
Well, it is worth looking at synthesized clocks jitter spectrum, they usually display a double-wing shape. Further, the PLL Synthsizers tend to be quite good in cleaning up LF jitter from the VCO within their bandwidth, if designed right.
Nowadays the best syntesised clocks manage < 50pS Jitter in the baseband up to 20KHz for "budget" devices. Mil Spec stuff can be better. And usually the nature of jitter in such system is noise, meaning it broadens the central peak of measured spectrum of (for example) a J-test spectrum, instead of producing distinct deterministic jitter, something which I think is less of a subjective impairment than deterministic jitter.
Ciao T
Hi,
No, not tried these? What noise levels do you measure?
With my active supplies I get a few microvolt wideband and < 1uV above a few Hz. Are these batteries better?
Ciao T
No, not tried these? What noise levels do you measure?
With my active supplies I get a few microvolt wideband and < 1uV above a few Hz. Are these batteries better?
Ciao T
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.