This is an improved Amanero:
Amanero USB I2S module PCM combo384 module DSD512 Upgraded with Crystek CCHD-957 | eBay
And these have the much better XMOS XU216 core:
XMOS 384kHz DXD DSD256 high-quality USB to I2S/DSD/SPDIF PCB - DIYINHK
Isolated XMOS 768kHz DXD DSD512(DSD1024) high-quality USB to I2S/DSD PCB - DIYINHK
If you can look for some form of galvanic isolation either built in or external such as Twisted Pear Hermes.
Amanero USB I2S module PCM combo384 module DSD512 Upgraded with Crystek CCHD-957 | eBay
And these have the much better XMOS XU216 core:
XMOS 384kHz DXD DSD256 high-quality USB to I2S/DSD/SPDIF PCB - DIYINHK
Isolated XMOS 768kHz DXD DSD512(DSD1024) high-quality USB to I2S/DSD PCB - DIYINHK
If you can look for some form of galvanic isolation either built in or external such as Twisted Pear Hermes.
Last edited:
From what i've gathered, a while back the berkley alpha usb was the best implementation. Now i see people with offramp saying it reaches close....i'm still unsure if usb or ethernet is better if the whole chain has clean powered ocxo's and the best implementation of usb. There was someone who had preffered rednet over all his other gear for a long while but now he had gotten to a 6 piece usb chain that sounded better to him than rednet. HOWEVER, i think rednet can also be improved with mods, where his usb chain was very close to top of the peak or maybe. There is another option which i'm pretty sure beats both....experience will tell.
REDNET is also limited at 24bit 192khz and aes, but again, the gains may make this unimportant, as long as one doesn't need DSD.
REDNET is also limited at 24bit 192khz and aes, but again, the gains may make this unimportant, as long as one doesn't need DSD.
Last edited:
Beyond the overall implementation, there are a few more (technical) things that count, so on short:
-XMOS - its firmware can be edited and tuned for meeting the design requirements, but the Windows (full) driver is not free; if an USB DAC is used with Windows and Linux based distros for playing PCM and DSD, than XMOS is the best solution.
-AMANERO - it already has a very decent implementation, the Windows driver is free, but the firmware user control is limited; there are some issues playing DSD in Linux based environment; if an USB DAC is intended to be used on Windows,or PCM on Linux, than AMANERO is the best solution.
Speaking about the implementation, as a diy-er, you can build your own XMOS based device and you will have the full control for hardware and firmware. There also are some licensed AMANERO modules better than the original one, but the user control are still limited.
Could you say which one, in your audible experience and/or just understanding is better for pcm and separately for DSD, on windows. From what you said it appears the xmos is the better all rounder, however i am curious what you mean when you say the AMANERO is the best solution. From what i gathered i was made sure that it was better for DSD upsampled playback, but for PCM not clear. Would be great if you could share the other amanero implementations ( if there are any available outside commercial dacs).
Now what i really wonder is if someone has compared(with measurements or even group listening ) head to head clocking accuracy/jitter when using the same chain .
In short : Does anyone actually know which one has the better architecture for timing and jitter ?
@RolandPSP Actually Markw4 answered to your question, I have the same opinion, but still I have to offer you an explanation:
when I said "Amanero is the best solution" , I had in my mind the first level of DIY activity when some pre-made boards (or made by ourselves from some shared projects) are combined and used to build a fully functional device. In this case, If Amanero meets your requirements (to listen music on Windows based systems), it is unbeatable because it is unchanged from many years, it is still on top of preferences of many users, it was reviewed over time by hundreds of users, it become a standard in its field, so it past the time test.
If you want to build from scratch your own USBtoIIS module, this is another story.
I will tell you only one "audiophile" fact, as an example: the ATSAM chip from Amanero module has its core 1V power supply derived from a switching DC-DC converter included in the chip (like XMOS U8 had, do you remember it?), but for XU208 you should provide 1V core power supply externally, so you can made an "audiophile" linear snake oil 1V power supply for that purpose.
My in field experience is limited to Amanero and a custom build XMOS (XHRA and XU208) modules, and I can hear no sound difference between them. I believe in good layout, in good low noise power supply, in good decoupling, in good isolation, in good clocks, but I will never spend money on expensive oscillators since I am convicted that my ears can't hear the fifth harmonic.
Last edited:
Both of them are superior solutions for the USB Audio.
There are positive and negative to both of them, but have in mind that the XMOS needs a re-clocking stage for sure. Amanero it is free from this (it is not necessary).
Mine experiments and measurements shows that the XMOS has an internal pulse instability to periodical time (some kind of jitter) and needs a 100% good re-clocking stage to correct them.
After re-clocking the result will be fine.
There are positive and negative to both of them, but have in mind that the XMOS needs a re-clocking stage for sure. Amanero it is free from this (it is not necessary).
Mine experiments and measurements shows that the XMOS has an internal pulse instability to periodical time (some kind of jitter) and needs a 100% good re-clocking stage to correct them.
After re-clocking the result will be fine.
Recklocking is usually done in the same fashion, which is not the right solution. This is in a 99% just one flip-flop clocking with MCK. That resulting in just 1/2 of period of MCK clocking and inverting the lines. And introduce posibile "metastabile" state...
.
This should be done with serial 2 x flip flpop say 74 type, and all lines will be 1 MCK period delayed and recklocked and in the same rising way like in the input of recklocker...
.
Other not good thing is that You have to have minimum 2 x MCK Fo for recklocking higher Sample rates. 24.xxx and 22.xxx MHz are good up to 192KHz. The MCK rates should be double for SR higher than 192KHz. That is because BCK is higher as higher SR, and with 35x and 38x KHz is equal than MCKs, and simply can not "recklock"
...
cheers
.
BTW consider Savitech USB interface. I like the sound.
AND mesure MCK because I found at some of XMOS USBs with new chip that MCK is inverted. Amanero have right MCK rising edge. Savitech too.
.
This should be done with serial 2 x flip flpop say 74 type, and all lines will be 1 MCK period delayed and recklocked and in the same rising way like in the input of recklocker...
.
Other not good thing is that You have to have minimum 2 x MCK Fo for recklocking higher Sample rates. 24.xxx and 22.xxx MHz are good up to 192KHz. The MCK rates should be double for SR higher than 192KHz. That is because BCK is higher as higher SR, and with 35x and 38x KHz is equal than MCKs, and simply can not "recklock"
...
cheers
.
BTW consider Savitech USB interface. I like the sound.
AND mesure MCK because I found at some of XMOS USBs with new chip that MCK is inverted. Amanero have right MCK rising edge. Savitech too.
You can also just use I2S slave mode and derive your own BCK and WCK, so no need to reclock.
Topping D90 Windows driver should be OK. There was a strange choice made by Topping a few years ago: to use XMOS-XHRA-2HPA in their products, I think it was D30.
This XMOS chip was actually a locked version of XU208, so the user couldn't edit VID and PID since they were "burned" inside this particular chip (XMOS proprietary VID and PID). It is no more produced.
XMOS-XHRA-2HPA was my first USB to I2S first project. I have tested it with the fully functional driver that Topping finally provided for their D30 product, and it perfectly works.
D90 should have Topping proprietary VID and PID, so there should be no problem with its driver.
This XMOS chip was actually a locked version of XU208, so the user couldn't edit VID and PID since they were "burned" inside this particular chip (XMOS proprietary VID and PID). It is no more produced.
XMOS-XHRA-2HPA was my first USB to I2S first project. I have tested it with the fully functional driver that Topping finally provided for their D30 product, and it perfectly works.
D90 should have Topping proprietary VID and PID, so there should be no problem with its driver.
This is a driver licencing issue and is still present, AFAIK (as it is a business). D90 has a Thesycon licenced PID/VID and any updates to the drivers recognise D90 as a legitimate device. Topping D30 use PID/VID from a XMOS development board and Topping gives for download a hacked version of a driver (which otherwise would work in demo mode). In this case you are limited to a driver from a Topping website and Topping has never provided updates to this driver.
XMOS offers a free limited driver for users who buy at least 150 chips. The most important limitation for most users is the absence of DSD native.
A few years ago, when I bought XMOS-XHRA chip, I was able to download for free this driver, from their website.