congratulation for your mosaic dac! This is a very revelatory review. What I adored about the sd-player sd-1 was how analog it sounded and natural. Cant wait to hear your new mosaic!
Hi Jaffrie,
The reviewer reported it sounded -different- and the differences were only subtle:
“I have listened to both versions and even though there are subtle differences, they cannot be seen as differences in resolution. I see them more as differences in timbre or tonality.”
The Mosaic 16 T that was being reviewed was modified in order to fix reported minor source dependency. The modifications included synchronous data reclocking.
The Mosaic 24 T used in the side by side comparison was not modified at that time. This would explain the reported difference.
The Mosaic D/A converter core always generates one extra bit, so the Mosaic 16 T offers 17 bit resolution when playing hi-res files.
Both, Mosaic 16 T and Mosaic 24 T will play Hi-res files up to 192 KHz.
The reviewer reported it sounded -different- and the differences were only subtle:
“I have listened to both versions and even though there are subtle differences, they cannot be seen as differences in resolution. I see them more as differences in timbre or tonality.”
The Mosaic 16 T that was being reviewed was modified in order to fix reported minor source dependency. The modifications included synchronous data reclocking.
The Mosaic 24 T used in the side by side comparison was not modified at that time. This would explain the reported difference.
The Mosaic D/A converter core always generates one extra bit, so the Mosaic 16 T offers 17 bit resolution when playing hi-res files.
Both, Mosaic 16 T and Mosaic 24 T will play Hi-res files up to 192 KHz.
Hi jusbe,
Allow a few days to a week for getting used to transparent analogue sound. Remember we have been listening and have gotten used to digital audio and related “digital” distortion for decades.
Amplifiers and speakers now become the weakest link. Make sure to use high resolution amplifiers and speakers to take full advantage of the Mosaic T DAC.
There is nothing to break in or warm up:
Logic circuits -> resistor matrix -> output socket.
Allow a few days to a week for getting used to transparent analogue sound. Remember we have been listening and have gotten used to digital audio and related “digital” distortion for decades.
Amplifiers and speakers now become the weakest link. Make sure to use high resolution amplifiers and speakers to take full advantage of the Mosaic T DAC.
There is nothing to break in or warm up:
Logic circuits -> resistor matrix -> output socket.
Hi QserraTico_Tico,
It’s not a (segmented) R2R ladder DAC like the PCM1704 or the Soekris DAC for example.
The Mosaic D/A converter core can best be described as a discrete common-rail matrix DAC and it runs on a multi-stream interface, not I2S.
Decoder is required for translating I2S into multi-stream. This is a discrete state machine decoder that runs on a +/- 0.8V balanced power supply. The decoder is specifically designed to offer very high speed and very low switching noise.
The multi stream interface contains multiple serial data signals, serial rail control signals and timing signals.
The Mosaic D/A converter only contains logic circuits and a resistor matrix. The balanced power supply offers DC-coupled output (4Vpp, 200 Ohm output impedance).
It’s not a (segmented) R2R ladder DAC like the PCM1704 or the Soekris DAC for example.
The Mosaic D/A converter core can best be described as a discrete common-rail matrix DAC and it runs on a multi-stream interface, not I2S.
Decoder is required for translating I2S into multi-stream. This is a discrete state machine decoder that runs on a +/- 0.8V balanced power supply. The decoder is specifically designed to offer very high speed and very low switching noise.
The multi stream interface contains multiple serial data signals, serial rail control signals and timing signals.
The Mosaic D/A converter only contains logic circuits and a resistor matrix. The balanced power supply offers DC-coupled output (4Vpp, 200 Ohm output impedance).
I think we need cheaper shipping!
Congratulations -EC-
You seem to have achieved revindication of the digital format and of its proponents.
Best wishes,
M.
One couldn't have said it better...I think we need cheaper shipping!
Congratulations -EC-
You seem to have achieved revindication of the digital format and of its proponents.
Best wishes,
M.
Would love a pricelist, including the XTOS and Beagle. Maybe you do package deals too?
Obviously, he would not reveal the guts of his invention
From what he stated before the Mosaic, I think he is de-composing de signal train into sub packages (for example, first 8 MSB; then 8LSB) and then running them in parallel into discrete logic (not bi-state) chains to decode them separately (multi-stream) to make good use of the "electric silence" he talked about, for the bigger part of the time-frame, for each subsection, then re-composing the analog signal from the sub-sections.
I would do this with my humble TDA1543, if I knew how!
I bet it would sound good...
Cheers,
M.
PS: I too mailed the site with no answer...I think he has the problem of low production rate and cannot withstand higher demand...pity, just when I ate my pride and was willing to BUY something instead of building it...
Because I heard the Ecdesigns dac at a audio club meeting, I decided to visit Ecdesigns and take the opportunity of a demo set to take home.
After two weeks of listening with the mosaic t dac on different locations with differents set up's ist very clear to me that this dac is in many ways performing better than other dac's.
More detail, more focus , a bigger sound picture and most of all a very "analog" sound.
Highly recommended, please go on with your outstanding work!
After two weeks of listening with the mosaic t dac on different locations with differents set up's ist very clear to me that this dac is in many ways performing better than other dac's.
More detail, more focus , a bigger sound picture and most of all a very "analog" sound.
Highly recommended, please go on with your outstanding work!
It was a good review until the reviewer said the bass improved significantly when he changed the toslink cable from plastic to glass..... UH....
I have no doubt the DAC is great though especially at those prices
Hi aive,
In order to get data into a DAC an interface with non-zero bandwidth is required. This is the same for USB, Toslink, coax, I2S, built-in transport and so on. For 192 KHz Toslink a bandwidth of approx. 13 MHz is required so bandwidth of the source interference spectrum will be limited. With USB interface the bandwidth is much larger, the HF interference can now easily crosstalk by means of very low stray capacitances.
Measurements show that source interference can be found on the ground plane, power supplies, data and timing signals and in the form of EMI.
Non-zero bandwidth allows source interference to reach the DAC, there is nothing that can be done about this. As soon as the (optical) connection is made, source interference will flow into the DAC electronics.
Clean masterclock alone won’t fix this, one has to obtain both, clean timing signals and lowest possible source interference throughout the DAC electronics.
So every DAC will be source dependent to some dregree. In other words, when swapping between various bit-perfect sources and / or interlinks the sound will change and these changes can also be verified by spectrum analysis of the DAC output signal.
One can easily try this with one’s own DAC by experimenting with different interlinks and digital (USB) audio sources. One will even notice a change in sound when a different USB socket on the same computer is used.
When little or no changes can be heard, the distortion of the audio set is too high and the marginal differences are simply swamped by interference.
Similar, CPU load and related interference spectrum will also reach the DAC electronics and has some effect on sound. Then it even makes a difference if a processor calculates with integers or floats, the result (data) is the same but the CPU interference spectrum will be different and the related sound will be different.
Optical and electrical interlink properties will change the spectrum of the passing digital signal and thus act like filters. Strictly speaking there are no digital signals, a so called digital signal is an analogue signal that varies between two voltage zones that are identified as “0” and “1”. The exact moment of switching depends on the input voltage level and the logic element switch thresholds. The logic element threshold may change during operation as a result of thermal drift and ground-boune for example.
The slower the signal changes from the “0” zone to the “1” zone, the more sensitive it will be to fluctuations in input voltage and treshold level (trigger uncertainty). The faster a signal switches, the more ground-bounce it will produce and this also has impact on the moment of triggering. So a compromise has to be found.
Real glass fiber interlinks consist of hundred’s of extremely thin glass fibers, plastic optical interlinks consist of a single plastic fiber. It is obvious that this has consequences for the passing optical signal (reflections, bandwidth, light output).
So when swapping between glass and plastic optical interlinks there will be a change in spectrum and related sound:
HF spectrum -> demodulation (P-N junction) -> interference within and outside the audio spectrum.
The Mosaic DAC tackles source interference in multiple ways:
- Ground loop interference is completely eliminated by using Toslink interface.
- I2S interference spectrum energy is reduced by 87.5% by means of a multi-stream interface that is only active during 8 of the available 64 bit clock pulses in a frame.
- Passive masterclock that completely eliminates power supply related phase noise.
- Synchronous reclocking of all data and clock signals used throughout the Mosaic DAC circuits.
In order to get data into a DAC an interface with non-zero bandwidth is required. This is the same for USB, Toslink, coax, I2S, built-in transport and so on. For 192 KHz Toslink a bandwidth of approx. 13 MHz is required so bandwidth of the source interference spectrum will be limited. With USB interface the bandwidth is much larger, the HF interference can now easily crosstalk by means of very low stray capacitances.
Measurements show that source interference can be found on the ground plane, power supplies, data and timing signals and in the form of EMI.
Non-zero bandwidth allows source interference to reach the DAC, there is nothing that can be done about this. As soon as the (optical) connection is made, source interference will flow into the DAC electronics.
Clean masterclock alone won’t fix this, one has to obtain both, clean timing signals and lowest possible source interference throughout the DAC electronics.
So every DAC will be source dependent to some dregree. In other words, when swapping between various bit-perfect sources and / or interlinks the sound will change and these changes can also be verified by spectrum analysis of the DAC output signal.
One can easily try this with one’s own DAC by experimenting with different interlinks and digital (USB) audio sources. One will even notice a change in sound when a different USB socket on the same computer is used.
When little or no changes can be heard, the distortion of the audio set is too high and the marginal differences are simply swamped by interference.
Similar, CPU load and related interference spectrum will also reach the DAC electronics and has some effect on sound. Then it even makes a difference if a processor calculates with integers or floats, the result (data) is the same but the CPU interference spectrum will be different and the related sound will be different.
Optical and electrical interlink properties will change the spectrum of the passing digital signal and thus act like filters. Strictly speaking there are no digital signals, a so called digital signal is an analogue signal that varies between two voltage zones that are identified as “0” and “1”. The exact moment of switching depends on the input voltage level and the logic element switch thresholds. The logic element threshold may change during operation as a result of thermal drift and ground-boune for example.
The slower the signal changes from the “0” zone to the “1” zone, the more sensitive it will be to fluctuations in input voltage and treshold level (trigger uncertainty). The faster a signal switches, the more ground-bounce it will produce and this also has impact on the moment of triggering. So a compromise has to be found.
Real glass fiber interlinks consist of hundred’s of extremely thin glass fibers, plastic optical interlinks consist of a single plastic fiber. It is obvious that this has consequences for the passing optical signal (reflections, bandwidth, light output).
So when swapping between glass and plastic optical interlinks there will be a change in spectrum and related sound:
HF spectrum -> demodulation (P-N junction) -> interference within and outside the audio spectrum.
The Mosaic DAC tackles source interference in multiple ways:
- Ground loop interference is completely eliminated by using Toslink interface.
- I2S interference spectrum energy is reduced by 87.5% by means of a multi-stream interface that is only active during 8 of the available 64 bit clock pulses in a frame.
- Passive masterclock that completely eliminates power supply related phase noise.
- Synchronous reclocking of all data and clock signals used throughout the Mosaic DAC circuits.
i 'm interested by this dac but can't see any price delay or ordering info anywhere on your website. i can agree with the use of optical fiber for a dac . 20 years ago my it was the better connection for my WADIA dac.
but it was the glass ST optical not the toslink ( perhaps a better mechanical coupling with the optical receiver) .
so my question is what did you choose the toslink over the ST?
second question is for the mini-jack output , what is the benefit of this connection if 90% of the buyer will use a mini/RCA converter to use their cables ?
but it was the glass ST optical not the toslink ( perhaps a better mechanical coupling with the optical receiver) .
so my question is what did you choose the toslink over the ST?
second question is for the mini-jack output , what is the benefit of this connection if 90% of the buyer will use a mini/RCA converter to use their cables ?
Hi juanitox,
We are planning to have a webshop online soon.
These interlinks are compatible with the Toslink connectors on the Mosaic T DAC.
Lifatec Glass Toslink Cables
So one can simply choose what type of fiber one wishes to use.
This might not be very clear but the Mosaic T DAC can be converted into a “standard” USB DAC by adding the XTOS. One then has a 192/24 USB DAC while ground loops issues are completely eliminated. The XTOS contains an XMOS USB receiver with two low jitter reference clocks for the 44.1 and 48 KHz groups and synchronously reclocks the SPDIF output signal.
The reviewer didn’t test the Mosaic VC yet, the Mosaic VC (volume control & analogue / digital input selector)is placed between the Toslink source and the Mosaic T DAC. It synchronously reclocks all incoming Toslink signals and feeds a low jitter Toslink signal to the Mosaic DAC. It adds yet another jitter barrier between source and DAC. When using the XTOS and the Mosaic VC one creates 3 jitter barriers:
1) Synchronous reclocker in the XTOS
2) Synchronous reclocker in the Mosaic VC
3) Synchronous reclocker in the Mosaic T DAC.
Here is a youtube vid that explains why
https://www.youtube.com/watch?v=VkYqfiOVOVY
Small (low mass) 3.5mm jack connectors introduce lower Eddy current losses and related lower distortion. Similar it is best to use litz wire interlinks with multiple thin insulated strands as these offer lower Eddy current losses too.
It is recommended to use these interlinks throughout the audio set. I use such configuration and obtain cleanest sound with this.
When adapters have to be used, then use adapter -cables- with the correct connectors on each side. Then at least one has eliminated Eddy current losses of one RCA connector pair.
The small adapter blocks are better avoided as these introduce even higher Eddy current losses as one (unnecessary) pair of RCA connections and related Eddy current losses are added to the signal path.
We are planning to have a webshop online soon.
These interlinks are compatible with the Toslink connectors on the Mosaic T DAC.
Lifatec Glass Toslink Cables
So one can simply choose what type of fiber one wishes to use.
This might not be very clear but the Mosaic T DAC can be converted into a “standard” USB DAC by adding the XTOS. One then has a 192/24 USB DAC while ground loops issues are completely eliminated. The XTOS contains an XMOS USB receiver with two low jitter reference clocks for the 44.1 and 48 KHz groups and synchronously reclocks the SPDIF output signal.
The reviewer didn’t test the Mosaic VC yet, the Mosaic VC (volume control & analogue / digital input selector)is placed between the Toslink source and the Mosaic T DAC. It synchronously reclocks all incoming Toslink signals and feeds a low jitter Toslink signal to the Mosaic DAC. It adds yet another jitter barrier between source and DAC. When using the XTOS and the Mosaic VC one creates 3 jitter barriers:
1) Synchronous reclocker in the XTOS
2) Synchronous reclocker in the Mosaic VC
3) Synchronous reclocker in the Mosaic T DAC.
Here is a youtube vid that explains why
https://www.youtube.com/watch?v=VkYqfiOVOVY
Small (low mass) 3.5mm jack connectors introduce lower Eddy current losses and related lower distortion. Similar it is best to use litz wire interlinks with multiple thin insulated strands as these offer lower Eddy current losses too.
It is recommended to use these interlinks throughout the audio set. I use such configuration and obtain cleanest sound with this.
When adapters have to be used, then use adapter -cables- with the correct connectors on each side. Then at least one has eliminated Eddy current losses of one RCA connector pair.
The small adapter blocks are better avoided as these introduce even higher Eddy current losses as one (unnecessary) pair of RCA connections and related Eddy current losses are added to the signal path.
hi ecdesigns, Do you feel that different usb cable, that connect the computer to the XTOS, can make a difference to SQ or any usb cable work exactly the same?
About the XTOS, I see you use the Xmos chip. do you use the Cyrstek CCHD-957 clocks? what is the regulation used: double power redulation?
can we have more information about the XTOS design and maybe pictures?
About the XTOS, I see you use the Xmos chip. do you use the Cyrstek CCHD-957 clocks? what is the regulation used: double power redulation?
can we have more information about the XTOS design and maybe pictures?
Last edited: