the mosaic T is not at all 8000 Eur, its around 1500/2000 Eur for the 16 bit. dont quote me as Im not sure.
You have to contact John Brown for pricing as I dont know the exact price but its around 2K Eur.
Its quite amazing that the Mosaic actually seem to beat a 11 000 Euro DAC!
Last edited:
Pardon me, but is this not a stunning claim? What was the test procedure?
Hi analog_sa,
The bit-perfect test is described here:
EC designs - XTOS Info
Click on in depth-technical information, Bitperfect test explained.
It basically consists of a test audio file that is played on the source to be tested. The XTOS built-in bit-perfect test algorithm recognizes the bit-perfect test file and automatically initiates the test. If all 65,536 values (16 bit) or all 16,777,216 values (24 bit) are received correctly the test passes and a white indicator LED comes on. If the test fails, the wite LED stays off.
We developed this bit-perfect test in order to rule out sound degrading caused by non-bitperfect playback using USB audio. The bit-perfect test function was integrated in the Mosaic USB DAC and is now available as separate XTOS module.
The test passes with Mac OSX, Linux, Windows XP with iTunes for windows and Windows 7, 8, and 10 with JRiver (UAC1).
We tried to get bit-perfect playback with Thesycon driver but failed. Maybe we overlooked something but this isn’t very likely as we tested thoroughly.
With Mac OSX and Linux we achieved bit-perfect playback without problems. Both support UAC1 and UAC2 so no drivers are required for playback up to 384/24.
The bit-perfect test is described here:
EC designs - XTOS Info
Click on in depth-technical information, Bitperfect test explained.
It basically consists of a test audio file that is played on the source to be tested. The XTOS built-in bit-perfect test algorithm recognizes the bit-perfect test file and automatically initiates the test. If all 65,536 values (16 bit) or all 16,777,216 values (24 bit) are received correctly the test passes and a white indicator LED comes on. If the test fails, the wite LED stays off.
We developed this bit-perfect test in order to rule out sound degrading caused by non-bitperfect playback using USB audio. The bit-perfect test function was integrated in the Mosaic USB DAC and is now available as separate XTOS module.
The test passes with Mac OSX, Linux, Windows XP with iTunes for windows and Windows 7, 8, and 10 with JRiver (UAC1).
We tried to get bit-perfect playback with Thesycon driver but failed. Maybe we overlooked something but this isn’t very likely as we tested thoroughly.
With Mac OSX and Linux we achieved bit-perfect playback without problems. Both support UAC1 and UAC2 so no drivers are required for playback up to 384/24.
Now we have to study toslink, which was previously neglected as interlink.
John, what do you recommend as good toslink "cable"?
A quick search revealed that DH Labs make a promising and not so expensive Toslink.
DH Labs Silversonic - Products - Interconnects - Toslink Optical Cable
John, what do you recommend as good toslink "cable"?
A quick search revealed that DH Labs make a promising and not so expensive Toslink.
DH Labs Silversonic - Products - Interconnects - Toslink Optical Cable
Does a Toslink is a so great difference with the isolator chips present in many designs today ?
Could you believe there is a difference between many input sources : USB, asymetric, symetric, etc, as far as this sort of isolator chips + buffer (fifo) are present on the input side of the DAC ?
Some say after listening ABX there are despite a fifo, some others sa there are not ?!
Could you believe there is a difference between many input sources : USB, asymetric, symetric, etc, as far as this sort of isolator chips + buffer (fifo) are present on the input side of the DAC ?
Some say after listening ABX there are despite a fifo, some others sa there are not ?!
Hi youknowyou,
The Mosaic T DAC has fully passive DC-coupled output stage (resistor matrix) with 200 Ohm output impedance.
So you get highest power in the headphones with an impedance that is closest to 200 Ohms.
It would produce approx. 2.4mW in 250 Ohm, 2.3mW in 300 Ohm, and 1.8mW in 600 Ohm.
I use a Beyer dynamics DT 990 pro for testing. It has 250 Ohms impedance (2.4mW on the Mosaic T DAC) and I have to attenuate the signal as it plays too loud.
This might be interesting:
Sennheiser HD800 vs Beyerdynamic T1 Review
The Mosaic T DAC has fully passive DC-coupled output stage (resistor matrix) with 200 Ohm output impedance.
So you get highest power in the headphones with an impedance that is closest to 200 Ohms.
It would produce approx. 2.4mW in 250 Ohm, 2.3mW in 300 Ohm, and 1.8mW in 600 Ohm.
I use a Beyer dynamics DT 990 pro for testing. It has 250 Ohms impedance (2.4mW on the Mosaic T DAC) and I have to attenuate the signal as it plays too loud.
This might be interesting:
Sennheiser HD800 vs Beyerdynamic T1 Review
Hi Eldam,
We need HF isolation between source and DAC with connected (pre) amplifier.
So isolating I2S signals only, using chip isolators is not sufficient as a ground loop will be created between source and DAC electronics and connected (pre) amplifier.
Chip isolators (magnetic. capacitive, optical) and isolation transformers have non-zero stray capacitance so HF interference still passes.
Toslink has zero stray capacitance (1+ meters of glass or plastic fiber) and blocks source HF interference. It does not pick-up EMI either as the optical fiber is an insulator.
So in this regard Toslink makes a lot of difference as it keeps HF source interference out of the DAC, amplifier, and audio interlinks.
You can measure and hear it, the difference is no imagination.
It is not sufficient to provide a low jitter DAC sample timing signal only.
We need HF isolation between source and DAC with connected (pre) amplifier.
So isolating I2S signals only, using chip isolators is not sufficient as a ground loop will be created between source and DAC electronics and connected (pre) amplifier.
Chip isolators (magnetic. capacitive, optical) and isolation transformers have non-zero stray capacitance so HF interference still passes.
Toslink has zero stray capacitance (1+ meters of glass or plastic fiber) and blocks source HF interference. It does not pick-up EMI either as the optical fiber is an insulator.
So in this regard Toslink makes a lot of difference as it keeps HF source interference out of the DAC, amplifier, and audio interlinks.
You can measure and hear it, the difference is no imagination.
It is not sufficient to provide a low jitter DAC sample timing signal only.
Hi maxlorenz,
Toslink was neglected because of problems with related jitter (spectrum) and difficulties of slaving the source because most sources have Toslink output only.
This problem has now been solved with the Mosaic T DAC, so now we can take advantage of the unique isolation properties of optical fiber.
So for clarity, if the DAC electronics are unable to reduce source & interlink jitter to inaudible levels, Toslink jitter issue remains. It is -not- sufficient to simply put a Toslink interface on your DAC.
If there is no data corruption at 192/24 the interlink is fine.
We use mini Toslink (3.5mm jack). We tested cheapest $2 plastic interlinks and these work without any problems all the way up to 192/24 as we are only interested in zeros and one’s, not in jitter properties.
There are possible problems with 3.5mm optical adapter plugs and optical switches as these introduce significantly reduced light output. So I advise to use a “one piece” Toslink interlink.
For -jitter sensitive- Toslink DACs it helps when using a low jitter S/PDIF source (synchronously reclocked S/PDIF transmitter).
Toslink interlinks based on multiple, thin, real glass fibers can also help to reduce Toslink interlink jitter and increase light output:
Glass Toslink Digital Optical Cable
Toslink was neglected because of problems with related jitter (spectrum) and difficulties of slaving the source because most sources have Toslink output only.
This problem has now been solved with the Mosaic T DAC, so now we can take advantage of the unique isolation properties of optical fiber.
So for clarity, if the DAC electronics are unable to reduce source & interlink jitter to inaudible levels, Toslink jitter issue remains. It is -not- sufficient to simply put a Toslink interface on your DAC.
If there is no data corruption at 192/24 the interlink is fine.
We use mini Toslink (3.5mm jack). We tested cheapest $2 plastic interlinks and these work without any problems all the way up to 192/24 as we are only interested in zeros and one’s, not in jitter properties.
There are possible problems with 3.5mm optical adapter plugs and optical switches as these introduce significantly reduced light output. So I advise to use a “one piece” Toslink interlink.
For -jitter sensitive- Toslink DACs it helps when using a low jitter S/PDIF source (synchronously reclocked S/PDIF transmitter).
Toslink interlinks based on multiple, thin, real glass fibers can also help to reduce Toslink interlink jitter and increase light output:
Glass Toslink Digital Optical Cable
Hi John,
If the isolator breaks the signal and ground connection, and if the power supply is referenced accordingly, cant we consider the signal isolated for all practical purposes?
HF leakage? - ok.. why not.
If I can follow correctly - can we rely then on the use HF BW limiting [typical attenuation as per LP filter re: post I/V stage and previous to typical gain stage (lets say 1541A) to isolate this HF] as an additional measure, and consider it completely isolated?. What frequencies are we talking about here?.. I am thinking >150kHz should present little problem if appropriate LP filter is employed - but I may be incorrect about this and of course enquiring minds would like to know.
Thank you,
LH/S
Last edited:
I believe that what John is referring to is common-mode noise, which is the basis for most of the noise interference that occurs between interconnected audio devices. Transformers are commonly utilized to break common-mode noise coupling, and work well at lower noise frequencies. However, problems can begin at higher frequencies even with transformer coupled signal interfaces due to the stray capacitance which exists between a transformer's primary and secondary windings. This stray capacitance will couple high frequency common-mode noise between transport source and DAC through the isolation transformer itself, reducing the the isolation that the transformer does effectively provide on lower frequency noise. As John states, fiber optic interconnection does not provide a mechanism for common-mode noise coupling regardless of frequency. Neither is fiber susceptible to electromagnetic noise induction from external sources. John is free to correct my presumption about exactly what concern he was expressing, or simply to add his always interesting technical perspective.
John, I would be very interested in reading anything you feel comfortable in sharing about the functioning of your fuzzy logic based clock generation technique. For instance, how do you keep the incoming data stream synchronized with the local clock signal? I presume that you do not use a PLL technique. Do you utilize a FIFO data buffer interface?
Last edited:
Hi Ken,
if fiber provides 192 kHz, it's certainly good enough for most of us !
A question please : does the aops which convert the light in signal at the output of the fiber not a glitch generator by itself as it's an active powered device ?
In any case does it need an isolator at the end before the dac chip or before any reclocker?
if fiber provides 192 kHz, it's certainly good enough for most of us !
A question please : does the aops which convert the light in signal at the output of the fiber not a glitch generator by itself as it's an active powered device ?
In any case does it need an isolator at the end before the dac chip or before any reclocker?
Hi, Eldam,
Any rapidly switching electrical component will inject noise in to the surrounding circuitry. This is true of digital logic gates in general, as well as Toslink optical receivers. However, this noise will not couple across the fiber interconnect like ground loop noise can across wired interfaces, even across those that are transformer coupled. So, no, there is no need for some additional isolator at the Toslink receiver because the fiber optic cable itself functions extremely well as that.
Last edited:
Hi,
Thanks Ken for the inputs,
My question was for after the output receiver as it is after the fiber interonnect : there will be some powered stuffs after the fiber link :
I assume as the receiver is an active powered chip : its own noise and glitch can be added to the low level of the multiplexed signal ? Then this multiplexed signal has to be itself to be converted in I2S signal by an extra powered stuffs with crystals.
So we avoided the signal PS noise from the computer source, but how to avoid the lesser jitter which is generated after and of course the noise of the powered stuffs even if on the same side than the dac chips (and sharing the same close ground plane?
Could it be a good idea to puut just a mini isolator chip before the dac just for the ground of each I2S signal to isolate the ground of the two actives devices before :toslink receiver and spidf to I2S chip convertor ?
Or does a clean reference ground just for the I2S input of the dac chip will not avoid the gltchs transmitted via the hotpoints signals of the three I2S wires ? (And so : we must live with that trade off ?)
It was more the sense of my question if I understood well your answer (that is not sure with my low technical level).
Thanks Ken for the inputs,
My question was for after the output receiver as it is after the fiber interonnect : there will be some powered stuffs after the fiber link :
I assume as the receiver is an active powered chip : its own noise and glitch can be added to the low level of the multiplexed signal ? Then this multiplexed signal has to be itself to be converted in I2S signal by an extra powered stuffs with crystals.
So we avoided the signal PS noise from the computer source, but how to avoid the lesser jitter which is generated after and of course the noise of the powered stuffs even if on the same side than the dac chips (and sharing the same close ground plane?
Could it be a good idea to puut just a mini isolator chip before the dac just for the ground of each I2S signal to isolate the ground of the two actives devices before :toslink receiver and spidf to I2S chip convertor ?
Or does a clean reference ground just for the I2S input of the dac chip will not avoid the gltchs transmitted via the hotpoints signals of the three I2S wires ? (And so : we must live with that trade off ?)
It was more the sense of my question if I understood well your answer (that is not sure with my low technical level).
Last edited:
The primary tools for minimizing noise injected by any active component are bypass capacitors, and series inductance and resistance. There are secondary tools as well, such as minimizing load current swing via limiting resistors and signal voltage amplitude attenuation.
I've before seen opto-isolation utilized to seperate analog and digital grounds, but such an approach is controversial. Many feel that opto-isolators are likely to induce more jitter than they suppress by ground isolation
Last edited:
Thanks Ken,
I assume maybe John is reclocking with Spartan stuffs before the Resistive conversion a little like Soekris did !
Many seems use such technics : CHord, etc ! Hum never understood if it's need a good external clock reference with low phase noise ! Very little stuffs not DIY friendly but for the most skilled and inteligent ones to programm such stuffs !
I assume maybe John is reclocking with Spartan stuffs before the Resistive conversion a little like Soekris did !
Many seems use such technics : CHord, etc ! Hum never understood if it's need a good external clock reference with low phase noise ! Very little stuffs not DIY friendly but for the most skilled and inteligent ones to programm such stuffs !
Dear -EC-
From your site:
* The Mosaic 16 T uses 17 bit DA conversion when in 24 bit mode.
I understand by this that your Mosaic 16 T will accept 24bit files and will neglect the last seven bits.
That means I won't have to "resample" those to 16bit.
Am I right?
I've found optic cable with Toslink at one end and mini-Toslink on the other.
That way I could send the noisy PC to the next room!
Cheers,
M.
From your site:
* The Mosaic 16 T uses 17 bit DA conversion when in 24 bit mode.
I understand by this that your Mosaic 16 T will accept 24bit files and will neglect the last seven bits.
That means I won't have to "resample" those to 16bit.
Am I right?
I've found optic cable with Toslink at one end and mini-Toslink on the other.
That way I could send the noisy PC to the next room!
Cheers,
M.