HQPlayer/microRendu working with Topping DX7s for DSD512

[Miska]

35 minutes ago, jabbr said:

Topping uses the Thesycon driver which works with Windows 10 for DSD. Folks are working on getting Thesycon with DSD into the mainline Linux kernel but for the moment you need to get the manufacturer to patch the Linux they distribute. You can try running @Miska s images on an NAA — I’m working on getting this running with the Espressobin

 

My patches plus the Thesycon support is already in mainline kernel. I wonder what is the thing they  are "working on", because Thesycon/XMOS implementations do work with the support I already upstreamed. If it is just listing one extra USB VID, then it is certainly not something that needs to be "worked on" longer than an hour or so...

 

Where do they have the patch/sources for the kernel?

 

Topping DACs are not that expensive, I could probably buy one for testing purposes.

 

[Miska]

26 minutes ago, jabbr said:

In my own case, I’m working in getting the Espressobin working with the Topping DX7s and needed to compile my own 4.18.7 kernel because Armbian doesn’t include ALSA nor snd_usb by default.

Each board has its own bootloading mechanism/weirdness that has to be taken into account so testing out your own images is more straightforward on amd64 builds than arm64 ... I plan to test your image with an Intel based NAA but have been pulled in another direction for the past couple of weeks and haven’t gotten to it yet — hopefully this weekend.

 

Yeah, ARM is always pain in that sense... Lot of BSPs are stuck to some ancient kernel because they didn't upstream their changes and couldn't maintain their custom things either. Or made some new shiny SoC and totally dropped the ball with previous ones. Some companies are better than others, for example Atmel has been quite OK with mainline kernels what I've used. Maybe some minor editing of device tree source for a particular setup, but that's it. Many others certainly not, while many are somewhere in between.

 

26 minutes ago, jabbr said:

My arm64 4.18.7-rt build works great with the iFi Micro and it’s driver, so there must be some issue with the Thesycon or else I’m doing something wrong ??‍♂️

 

(iFi is also XMOS)

 

And iFi nowadays falls back to the same detection method since I tested and removed the explicit VID:PID to alt-setting mapping which was wrong and seems to be problematic for example with iFi Pro... And the similar strict mapping probably broke Oppo BD players when they updated firmware.

 

But be careful with -rt kernels on ARM, in many cases it breaks things on the BSP because they didn't do things properly.

 

[Miska]

I've tested WiFi by running HQPlayer on a laptop connected wirelessly to local network and then NAA is connected to the local network using cable connection. DSD512 goes through just fine. I have not personally tested it the other way around where NAA would have a wireless connection. Maybe I should do some tests and measurements that way too.

 

[Miska]

Can someone just try it with my 4.5.1 HQPlayer Embedded image? That's the most recent stuff I have.

[Miska]

Those noise effects usually happen due to either of two reasons:

1) Transfer errors on USB.

2) Clocking problems causing DAC to loose sync.

If this happens, check with standard USB HiSpeed certified cable (not "audiophile" cable) and try without extra USB gadgets if any of such are used. Some USB interface implementations are more sensitive to these than others. For (2) also check the device grounding layout to make sure there are no ground currents disturbing clocks. With NAA make sure the ethernet cable is NOT shielded type with metal covered connectors, instead use UTP cables that have fully plastic connector bodies.