Schiit Urd Has Launched!

[Miska]

1 hour ago, Currawong said:

All other solutions use off-the-shelf USB chips from XMOS, VIA, etc. which have a whole lot of other code in there.

 

"All other solutions" is XMOS controller or STMicro controller. It is as much "completely custom" as what the Schiit's PIC controller based solution is. There's really no practical difference between these. Just MCUs from different manufacturers.

 

[Miska]

5 minutes ago, botrytis said:

TEAC used a specialized USB controller in the UD-501 design.

 

It's similar to the others, they use NXP's MCU. Couple of other Japanese vendors also use the same implementation.

 

[mgbpuff]

So, USB is devolving into a bunch of brand specific customizations. much as is the case with I2S,  What a disservice to the industry!  

[The Computer Audiophile]

35 minutes ago, mgbpuff said:

So, USB is devolving into a bunch of brand specific customizations. much as is the case with I2S,  What a disservice to the industry!  

Better engineers are taking the USB protocol standard and designing better hardware for it. Sounds good to me. 

[Currawong]

8 hours ago, Miska said:

"All other solutions" is XMOS controller or STMicro controller. It is as much "completely custom" as what the Schiit's PIC controller based solution is. There's really no practical difference between these. Just MCUs from different manufacturers.

 

I probably didn't put it well. The Schiit implementation is entirely their own programmed code, which only what is necessary for up to 24/192k audio. That is what I meant by "completely custom", as a PIC controller is like an FPGA in that it only does whatever you program it to.

 

When a manufacturer uses the via USB receiver from VIA or XMOS, it contains code for a whole bunch of stuff, including video, in the case of the VIA chips. You can ask VIA to remove that code, and they'll do it (and it supposedly improves the sound quality removing everything but the necessary audio code). But you're still using the hardware and code from that company.

[Miska]

15 minutes ago, Currawong said:

as a PIC controller is like an FPGA in that it only does whatever you program it to.

 

In that sense it is no different from XMOS, STMicro or NXP. Just different vendor - Atmel (who purchased Microchip which was original vendor of PIC).

 

When I was still designing DACs based on COTS DAC chips, I was using PIC controllers to manage the DAC chip registers etc. So I'm very familiar with programming those - in their native assembly language.

 

15 minutes ago, Currawong said:

which only what is necessary for up to 24/192k audio

 

Probably the PIC doesn't have horsepower to do any better.

 

15 minutes ago, Currawong said:

When a manufacturer uses the via USB receiver from VIA or XMOS, it contains code for a whole bunch of stuff, including video, in the case of the VIA chips. You can ask VIA to remove that code, and they'll do it (and it supposedly improves the sound quality removing everything but the necessary audio code). But you're still using the hardware and code from that company.

 

Wrong! No certainly not! XMOS is much more like FPGA than the other ones. Which also makes it harder to program. FPGA is ultimate PITA to deal with in terms of programming. But any bigger FPGA has lot of different IP blocks such as number of PLL's and similar hardware blocks that you may or may not use. That are intended as helper to make life a bit easier and consume less logic blocks for often used basic functionality. So FPGAs commonly include plenty of unnecessary hardware blocks. In addition, FPGAs have lot of other issues, such as problems to deal with pin-to-pin timings and burning lot of electricity for unused logic blocks.

 

I have not heard of anybody using VIA and I'm not aware of any USB-to-I2S controllers by VIA. There is one fixed hardware vendor making such chips, but those certainly don't have anything else than what is needed for USB Audio Class.

 

[Currawong]

25 minutes ago, Miska said:

I have not heard of anybody using VIA and I'm not aware of any USB-to-I2S controllers by VIA. There is one fixed hardware vendor making such chips, but those certainly don't have anything else than what is needed for USB Audio Class.

Thanks for the clarification about XMOS. I had thought that all the XMOS firmware was from XMOS alone. Audio-gd, many years ago, used VIA for USB, but switched to XMOS like everyone else. 

[Miska]

1 minute ago, Currawong said:

Thanks for the clarification about XMOS. I had thought that all the XMOS firmware was from XMOS alone. Audio-gd, many years ago, used VIA for USB, but switched to XMOS like everyone else. 

 

XMOS offers some (buggy) USB Audio Class code as reference source code for their controllers. You can modify this code as you wish. But you can run as much your own code on XMOS controllers as you like.

 

Not so different with PIC:

https://www.microchip.com/en-us/solutions/consumer/audio-and-speech/pic32-digital-audio/pic32-usb-audio#

 

[Superdad]

1 hour ago, Miska said:

Just different vendor - Atmel (who purchased Microchip which was original vendor of PIC).

???  

Microchip bought bought Atmel (and probably 30 other chip companies over the years). Atmel's equivalent to PIC processors was AVR.  

[This all leaving aside all of ARM processor world--much of it also with Microchip, as with ATSAM series (which we use a couple of on EtherREGEN PCBs).]

 

Of course everything else you wrote is quite correct.  And ask any DAC maker who has spent time with XMOS and their programming environment and you will get a half hour rant about the evil and stupidity of a lot of what they have done. No surprise that Schiit and others (e.g. Amanero using Atmel and Xilinx) have steered clear of XMOS. 

Just because everything and the kitchen sink can be thrown into a chip series in an attempt to please all of the market--does not mean that designers wanting to do very specific things will like it.

While I am sure it was not a trivial exercise for Schitt to write their own USB MAC for a very small PIC processor (though perhaps they licensed a USB core; lots of those available from core code firms in India), at least now they have control over their destiny. And given Schiit having a wide range of cost-conscious DACs with USB input, it sure makes sense. A $3 PIC is a big savings versus a $13 XMOS when you are talking thousands of units! 

[Miska]

59 minutes ago, Superdad said:

???  

Microchip bought bought Atmel (and probably 30 other chip companies over the years). Atmel's equivalent to PIC processors was AVR.

 

OK, my memory may be failing which way it went, but I do remember the two companies merged.

 

59 minutes ago, Superdad said:

[This all leaving aside all of ARM processor world--much of it also with Microchip, as with ATSAM series (which we use a couple of on EtherREGEN PCBs).]

 

I have bunch of Atmel ARM based systems. networkaudiod for "armel" architecture is still being built on on AT91SAM9G25 running Debian Linux (good for up to DSD256 by the way, through the single 400 MHz ARM9 core).

 

59 minutes ago, Superdad said:

No surprise that Schiit and others (e.g. Amanero using Atmel and Xilinx) have steered clear of XMOS. 

 

So far, most seem to have gone to STM32 which I'm also very familiar with.

 

59 minutes ago, Superdad said:

And given Schiit having a wide range of cost-conscious DACs with USB input, it sure makes sense. A $3 PIC is a big savings versus a $13 XMOS when you are talking thousands of units! 

 

I personally wouldn't go with PIC for USB. I'm fine writing assembly for it to control DAC chips and S/PDIF / AES/EBU receivers. That's how I've been using PIC16 series. But bigger code bases is pain with such, need to keep stack depth etc in mind. From pricing point of view it is easy to understand. Also the electronics needed to run the controller is minimal. Typically you don't even need external clock, but you can rely on the built-in clock generator (4 MHz for PIC16). Also managing LCD and similar displays and some hardware buttons is simple and straightforward.

 

[taipan254]

While a big chunk of this conversation is above my head, here are a few points I remember hearing about that may contribute to this conversation:

• Taiko builds custom drivers to optimize interaction between the Extreme and various DACs (including those with XMOS controllers). So at least Taiko believe that sound gains can be had by modifying USB drivers. 
• I have read in numerous forums that the Amanero USB interface is generally not the best - hence why Chord folks that upsample have adopted Dan Mance's / AudioWise's SRC DX instead of going via Chord's integrated Amanero USB board to their sources.
• I am less sure on this, but I BELIEVE Jason Stoddard said on the DARKO podcast episode discussing Urd that Unison USB does not support UAC 2.0, and that is why it is driverless and only works through 32/192.  NOT TRUE

[The Computer Audiophile]

8 minutes ago, taipan254 said:

While a big chunk of this conversation is above my head, here are a few points I remember hearing about that may contribute to this conversation:

• Taiko builds custom drivers to optimize interaction between the Extreme and various DACs (including those with XMOS controllers). So at least Taiko believe that sound gains can be had by modifying USB drivers. 
• I have read in numerous forums that the Amanero USB interface is generally not the best - hence why Chord folks that upsample have adopted Dan Mance's / AudioWise's SRC DX instead of going via Chord's integrated Amanero USB board to their sources.
• I am less sure on this, but I BELIEVE Jason Stoddard said on the DARKO podcast episode discussing Urd that Unison USB does not support UAC 2.0, and that is why it is driverless and only works through 32/192. 

Urd ONLY supports UAC2. UAC1 will not function with Urd. 

[taipan254]

Post modified / corrected for the record! Thanks.

[Miska]

9 hours ago, taipan254 said:

I have read in numerous forums that the Amanero USB interface is generally not the best - hence why Chord folks that upsample have adopted Dan Mance's / AudioWise's SRC DX instead of going via Chord's integrated Amanero USB board to their sources.

 

It is good with the latest firmware versions that support also DSD the same way as other interfaces. But there are older firmware versions around that are not as great.

 

Amanero is quite a bit less noisy though than XMOS. This was interesting testing result for example with my DSC1 DAC design. XMOS needs some special care on DAC design to not leak it's noise through the DAC.

 

[Currawong]

On 10/1/2023 at 9:26 AM, Superdad said:

A $3 PIC is a big savings versus a $13 XMOS when you are talking thousands of units!

Especially when your cheapest DAC is $129.  Actually, for some fun, plugging it in to an old iPhone power supply (which are low noise) and running HQPlayer has quite decent results. Overkill, but interesting all the same. 

[miguelito]

My 2c on transports... It seems to me that most of the transports produced these days are fairly good. The Hegel Mohican used the same CD-only Sanyo transport used in many cheap Cambridge CD players/transports. 

 

I read that Teac is resurrecting the VRDS transport in some shape or form?