MDAC First Listen (Part 00101001)

[cstanpfm]

Hi John, sorry if this had been asked and answered.

Can the detox USB 1.1 and 2.0 ports be used at the same time?

Eg USB 1.1 connected to MDAC while the USB 2.0 connected to MDAC2 (while playing a WAV file for example)?

What is the input? USB port type B?

Thanks.

 

[Ciunas Audio]

With the MDAC / FDAC the USB Microframes will be locked to the Audio Master clock so will not generate non synchronise noise.

I haven't seen this done before. How is this achieved as the USB clock is usually a multiple of 12MHz but the audio clocks aren't?

I've only seen the 8KHz frames on USB devices with poor clock recovery (Non ASync) such as Tenor, TI PCM290X / 270x and Cmedia.

Here's Archimago's measurement of this 8KHz noise spike using a Teac UD 501 which uses the TMS320 DSP chip for USB receiver duties done in software. Admittedly the 8KH noise spike was found when using this configuration HTPC --> 10' generic USB cable --> TEAC UD501 DAC --> 6' balanced analogue cables --> Emotiva XSP-1 preamp --> 6' balanced analogue cables --> E-MU USB0404 ADC --> measurement computer
It seems like the Emotiva preamp caused this noise to be measureable on the analogue outs of the preamp but it does show that the 8KHz noise is internally generated in the DAC from the PHY logic gate noise pattern when dealing with the USB micrframes

 

[JohnW]

I haven't seen this done before. How is this achieved as the USB clock is usually a multiple of 12MHz but the audio clocks aren't?

On the FDAC we use a PLL circuit to derive the USB clock form the Audio master clock, while the MDAC has a clever Clock circuit.

Here's Archimago's measurement of this 8KHz noise spike using a Teac UD 501 which uses the TMS320 DSP chip for USB receiver duties done in software. Admittedly the 8KH noise spike was found when using this configuration HTPC --> 10' generic USB cable --> TEAC UD501 DAC --> 6' balanced analogue cables --> Emotiva XSP-1 preamp --> 6' balanced analogue cables --> E-MU USB0404 ADC --> measurement computer
It seems like the Emotiva preamp caused this noise to be measureable on the analogue outs of the preamp but it does show that the 8KHz noise is internally generated in the DAC from the PHY logic gate noise pattern when dealing with the USB micrframes

This should not be taken as a representative design!!! - as I designer I'm horrified that you can see the 8KHz USB frame packets on the audio output!

 

[JohnW]

Hi John, sorry if this had been asked and answered.

Can the detox USB 1.1 and 2.0 ports be used at the same time?

Eg USB 1.1 connected to MDAC while the USB 2.0 connected to MDAC2 (while playing a WAV file for example)?

What is the input? USB port type B?

Thanks.

No, only a single output should be used at any one time - but if you really wanted then I guess its possible, but I don't like the idea of RF currents flowing between the two DAC's.

Yes, the Detox USB input is a standard USB type B connector.

 

[music fiend]

On the FDAC we use a PLL circuit to derive the USB clock form the Audio master clock, while the MDAC has a clever Clock circuit.

Any chance of this working with a 8200CD too?

 

[JohnW]

The 8200Cd does not have a clock output port but you will still gain the other benefits - especially galvanic isolation / RF attenuation and the spread spectrum clock feature.

 

[music fiend]

The 8200Cd does not have a clock output port but you will still gain the other benefits - especially galvanic isolation / RF attenuation and the spread spectrum clock feature.

Right. Thanks John!

 

[Rosewind]

I use an externally powered Paul Pang usb card in my music pc, so I would think that my usb out has no need for further detox work. For that reason I have not ordered a unit. Will I regret this?

 

[adamdea]

Great, OK so perhaps I haven't fully understood, but is 96kHz 24 bit enough bandwidth to cope with DSD, or is that limitation only for PCM?

I don't think you can fit dsd 64 (2.8mb/s) into 24/96 (2.3 mb/s)
see
http://dsd-guide.com/dop-open-standard

 

[wintoid]

I don't think you can fit dsd 64 (2.8mb/s) into 24/96 (2.3 mb/s)
see
http://dsd-guide.com/dop-open-standard

I think John is saying it's limited on the USB1 port, but not on the USB2 port, unless I've misunderstood again

 

[JohnW]

I think John is saying it's limited on the USB1 port, but not on the USB2 port, unless I've misunderstood again

Yes correct

 

[Ciunas Audio]

On the FDAC we use a PLL circuit to derive the USB clock form the Audio master clock, while the MDAC has a clever Clock circuit.

Ok, sounds interesting, thanks. So the PLL derived USB clock comes out as a better option Vs a separate standalone USB clock - interesting

This should not be taken as a representative design!!! - as I designer I'm horrified that you can see the 8KHz USB frame packets on the audio output!

Yep, I agree on both counts but I do still pause for thought - if we are to follow the logic of the premise that mistiming or jitter of the USB datapacket timing makes it's way into the DAC then the idea of the packet timing itself making it's presence felt in some way within the DAC is not that outrageous. Not necessarily as a direct translation of the microframe timing into a 8KHz spike noise on the analogue output, however!

 

[adamdea]

On the FDAC we use a PLL circuit to derive the USB clock form the Audio master clock, while the MDAC has a clever Clock circuit.




This should not be taken as a representative design!!! - as I designer I'm horrified that you can see the 8KHz USB frame packets on the audio output!

Don't worry, you can't see it on the outputs of either the emotiva amp (ie the ordinary stereo output) or Archimago's teac dac. There doesn't seem to be any reason to blame the dac. It only seems to arise from the av bypass output of the emotiva. Given that it's effect on the av is dramatically reduced with the corning we may wonder where exactly it is produced, from where it is radiated etc. Are we hearing noise direct from the computer or noise radiated from the usb receiver in the dac? Could it be radiated by the usb cable and picked up by the emotiva's analog circuits?

 

[Ciunas Audio]

Don't worry, you can't see it on the outputs of either the emotiva amp (ie the ordinary stereo output) or Archimago's teac dac. There doesn't seem to be any reason to blame the dac. It only seems to arise from the av bypass output of the emotiva. Given that it's effect on the av is dramatically reduced with the corning we may wonder where exactly it is produced, from where it is radiated etc. Are we hearing noise direct from the computer or noise radiated from the usb receiver in the dac? Could it be radiated by the usb cable and picked up by the emotiva's analog circuits?

Yes, good points. Pertinent questions are:
- what does the AV bypass switch do which causes this to appear on the analogue out
- as you say, where is this noise arising from - the USB transmitter or the USB receiver & what is it's pathway into the DAC
- the noise is obviously on the DAC's analogue out as this is the only connection with the AV receiver (via balanced cables which should reduce any CM noise) - so I'm not sure I would agree that it's not on the DAC's output

 

[iansr]

This all sounds very good guys but I'm a bit concerned that nobody has yet mentioned the dilithium crystals . . . . . .

 

[JohnW]

Ok, sounds interesting, thanks. So the PLL derived USB clock comes out as a better option Vs a separate standalone USB clock - interesting

No not at all, but the PLL derived clock (also consider its been transmitted via optical cable and the phase noise entails) is only used to "injection lock" the Dotox's discrete clock circuit.

So while the Detox's Clock circuit (even when injection locked) is very low phase noise - the USB HUB ic's internal Clock tree will be anything but. USB Hubs are not designed with ultra low Phase noise as a prime consideration.

With this in mind I'm designing the spread spectrum option mode.

Yep, I agree on both counts but I do still pause for thought - if we are to follow the logic of the premise that mistiming or jitter of the USB datapacket timing makes it's way into the DAC then the idea of the packet timing itself making it's presence felt in some way within the DAC is not that outrageous. Not necessarily as a direct translation of the microframe timing into a 8KHz spike noise on the analogue output, however!

Agreed, just screams bad design - there is some argument that its present during AV bypass - but I fail to see how this is any less relevant - it should not be present at all!

 

[JohnW]

This all sounds very good guys but I'm a bit concerned that nobody has yet mentioned the dilithium crystals . . . . . .

No, they sound bad.... too many resonance nodes which leads to sibilance.

 

[DANOFDANGER]

Thanks john for the reply, you answered most of my curiosity so appreciate as always your time.

The incoming USB Data is varying in time around its expected "time slot" the frequency and amplitude of this time variance is the encoded phase noise. When the USB device in the DAC processes this data it radiates this noise into the system.

Very interesting, so even though jitter is not improved upon past the usb, its the effects from async processing that radiates upon itself. This seems very similar to the effects from the likes of the mdacs firmware differences. It would seem that discrete dacs would be the ideal way to prevent such issues as you could isolate such effects.

The Detox prime reason for exsistances is to attenuate the "unwanted" RF and Jitter pattern of the incoming USB Data, provide a low noise PSU and also offer a galvanically isolated USB 1.1 output.

Just to be clear for myself and others, the jitter reduction from the detox is not to improve on jitter itself for the converted signal but to prevent radiated noise modulation from the async usb process. Infact wouldnt the output signal have identical jitter pattern once processed by the async usb.

Yes an optically isolated "Clock Locked" SPDIF source is the the perfect solution, but only for single sample rate applications as you are left with the problem of how to command the sample rate change / Master clock rate change. Worked well in the days of 44.1KHz CD only systems but not in our modern multi samplerate world.

I suppose thats why people clock lock their cd players with dacs. Makes sense somewhat.

 

[JohnW]

Thanks john for the reply, you answered most of my curiosity so appreciate as always your time.

Very interesting, so even though jitter is not improved upon past the usb, its the effects from async processing that radiates upon itself. This seems very similar to the effects from the likes of the mdacs firmware differences. It would seem that discrete dacs would be the ideal way to prevent such issues as you could isolate such effects.

Due to the larger Loop area, the discrete DAC solution will be more likely to be effected by external radiated fields.

The second order effects of the USB interface are not a result of the DAC being based upon an integrated silicon design, but RF and other electrical noise disturbances entering the DAC.

Just to be clear for myself and others, the jitter reduction from the detox is not to improve on jitter itself for the converted signal but to prevent radiated noise modulation from the async usb process. Infact wouldnt the output signal have identical jitter pattern once processed by the async usb.

The Detox purpose is to attenuate the second order effects of RF and USB packet Jitter etc of the USB interface. It should not directly effect the phase noise of the conversion clock, but any undesirable "RF" noise is by its nature is prone to effect other local circuit nodes.

I suppose thats why people clock lock their cd players with dacs. Makes sense somewhat.

Yes, I designed a clock-lock transport for Pink Triangle back in 1994 for this very reason - it was easy back then as we only had to care about 44.1KHz sampling rate.

 

[DANOFDANGER]

Yep im with ya, thanks. The results of the detox/regen will be of interest indeed.