advertisement


MDAC First Listen (part 00110011)

Status
Not open for further replies.
edit: In other news, h3 codec support landed in linux-next a few days ago, so I'll be trying that when my m1 board arrives. Proper i2s support for h3 in vanilla kernel might still take a few months, but it's getting there.
Somehow I suspect we have time to wait for support in the vanilla kernel :)
 
That's great news, given the lack of alternative boards with all the necessary connectors on the same side. Thank you for your continued efforts Jirij. It sounds like a rather challenging project; your perseverance is appreciated!

Yes indeed :) Thanks Jiri for your support and help :)
 
I have done some trials with SoX upsampling to 352.8khz and comparing DXD (PCM) and DSD256 (Native) output with Foobar2000+AsioProxy. I definitely like the DSD output more. And i like both more than just plain 44.1khz output, which really does not sound bad either. I used to be against upsampling, but i think DSD with a good upsampler is a game changer.

This is with MSB The Analog Dac with Premium Quad USB 2 module. I think if the FDAC will have a good DSD512 support it has a lot of potential. Upsamling+DSD256 takes about 5% of my [email protected] and it works fine when playing games too. I'd expect the CPU load of DSD512 conversion to be around 10%.

I think more people should experience the DSD x2/x4/x8 (and with proper upsampling, SoX and the like, not windows upsampler). There seems to be a fundamental difference to the better even tho the original material is "just" CD-quality.
 
We keep being asked about the RPi - what is workable is a updated "Compute Module" with the latests Quad core CPU used on the RPi3.

There is talk of such being developed (CM3), but it was meant to be released in Q2 2016...

This would be a great solution (despite the use of the more complex USB interface) from a software and future hardware availability standpoint, provided the CM3 becomes available in time.

It must be very frustrating to you that we are 'close' to a few solutions, yet without guarantee that any of them will work in the end!
 
I have done some trials with SoX upsampling to 352.8khz and comparing DXD (PCM) and DSD256 (Native) output with Foobar2000+AsioProxy. I definitely like the DSD output more. And i like both more than just plain 44.1khz output, which really does not sound bad either. I used to be against upsampling, but i think DSD with a good upsampler is a game changer.

This is with MSB The Analog Dac with Premium Quad USB 2 module. I think if the FDAC will have a good DSD512 support it has a lot of potential. Upsamling+DSD256 takes about 5% of my [email protected] and it works fine when playing games too. I'd expect the CPU load of DSD512 conversion to be around 10%.

I think more people should experience the DSD x2/x4/x8 (and with proper upsampling, SoX and the like, not windows upsampler). There seems to be a fundamental difference to the better even tho the original material is "just" CD-quality.

Hi Elviz,

Did you compare your foobar2000 set up versus HQPlayer? Could you provide links to the components you're using to upsample to DSD256 please? I found Resampler V which has SoX option but it only seems to do PCM upsampling.

http://www.computeraudiophile.com/f11-software/resampler-v-sample-rate-convertor-foobar-23241/

https://sourceforge.net/projects/resamplerv/

Thanks

PS I'm assuming you're upsampling on the fly as HQPlayer?
 
We keep being asked about the RPi - what is workable is a updated "Compute Module" with the latests Quad core CPU used on the RPi3.

There is talk of such being developed (CM3), but it was meant to be released in Q2 2016...

https://www.raspberrypi.org/forums/viewtopic.php?f=98&t=137952

https://www.raspberrypi.org/forums/viewtopic.php?f=98&t=146948&p=967458#p967458

So lets say we design the SOC Daughter PCB with CM3 in mind (with eMMC fitted), but we test it with the CM1. The daughter PCB will have the SOC related rear panel connectors arranged to fit the MDAC2 rear panel and the USB to XMOS solution this should be a viable RPi3 based solution :)

NEC recently announced new displays which will be using a modified CM3: http://hackerboards.com/rpi-compute-module-3-revealed-tapped-for-nec-signage/

And the Compute Module Datasheet already includes the CM3 and CM3L (for use with your own SD/eMMC): https://www.raspberrypi.org/documentation/hardware/computemodule/RPI-CM-DATASHEET-V1_0.pdf
 
The NEC is scheduled for launch in January - so I suspect a CES launch, but no sign of the CM3 module ANYWHERE!

Still, nothing to stop us design and test the SOC daughter board using the CM1 - by the time MDAC2 hits the production line CM3 hopefully will be more really available - or can be shipped with CM1, and customer upgrades when CM3 is released,
 
Hi Elviz,

Did you compare your foobar2000 set up versus HQPlayer? Could you provide links to the components you're using to upsample to DSD256 please? I found Resampler V which has SoX option but it only seems to do PCM upsampling.

http://www.computeraudiophile.com/f11-software/resampler-v-sample-rate-convertor-foobar-23241/

https://sourceforge.net/projects/resamplerv/

Thanks

PS I'm assuming you're upsampling on the fly as HQPlayer?

Oh, the source is PCM, 44.1khz 16bit. I am using this SoX resampler, because it supports 352.8khz (its possible to write the desired samplerate). The DSD output is done via foo_dsd_asio and asio proxy.

I haven't tried the HQPlayer. I might give it a try.
 
We keep being asked about the RPi - what is workable is a updated "Compute Module" with the latests Quad core CPU used on the RPi3.

There is talk of such being developed (CM3), but it was meant to be released in Q2 2016...

https://www.raspberrypi.org/forums/viewtopic.php?f=98&t=137952

https://www.raspberrypi.org/forums/viewtopic.php?f=98&t=146948&p=967458#p967458

So lets say we design the SOC Daughter PCB with CM3 in mind (with eMMC fitted), but we test it with the CM1. The daughter PCB will have the SOC related rear panel connectors arranged to fit the MDAC2 rear panel and the USB to XMOS solution this should be a viable RPi3 based solution :)
would this solve the problem you identified that the pi could not be slaved to the dac? I took this to mean that the pi itself could not be clocked by an external master (not that it could not work in i2s slave mode)
 
would this solve the problem you identified that the pi could not be slaved to the dac? I took this to mean that the pi itself could not be clocked by an external master (not that it could not work in i2s slave mode)
I read it like it will use USB to XMOS and not i2s.
Guess that is what everyone else is using.
I rather have that than waiting for a i2s solution that might never happen.
At least for MDAC2
 
If anyone else is curious about what are Compute Modules and CM3s ... they are small (Raspberry Pi) computers which are slightly modified by Raspberry to slot into the circuit board of some other equipment - such as an FDAC.

See the end of
http://arstechnica.co.uk/informatio...i-fi-and-bluetooth-64-bit-chip-still-just-35/

which leads to

http://arstechnica.com/information-...uino-y-with-new-open-source-modular-hardware/

But - I still don't know what John would use them for.
Same thing the original Nanopi M1 board was meant to do - a built-in "streamer". Basically a "RPi" with the MDAC2 connected on one of its USB ports, internally. A "single-box" solution.
 
If anyone else is curious about what are Compute Modules and CM3s ... they are small (Raspberry Pi) computers which are slightly modified by Raspberry to slot into the circuit board of some other equipment - such as an FDAC.

See the end of
http://arstechnica.co.uk/informatio...i-fi-and-bluetooth-64-bit-chip-still-just-35/

which leads to

http://arstechnica.com/information-...uino-y-with-new-open-source-modular-hardware/

But - I still don't know what John would use them for.
Thank you, helpful post.
 
Remember some discussion early on but please remind me what provisions are being made for a power supply for the DETOX?

IF a SMPS is being provided, would a LRPS enhance the performance?
Why or why not?

What are the Voltage and Amperage requirements?

Thanks.
 
Same thing the original Nanopi M1 board was meant to do - a built-in "streamer". Basically a "RPi" with the MDAC2 connected on one of its USB ports, internally. A "single-box" solution.

Sounds like my guess was correct:)
 
Jan didden's silent switcher could be a solution for the detox. Very good multi output reg.
 
We keep being asked about the RPi - what is workable is a updated "Compute Module" with the latests Quad core CPU used on the RPi3.

There is talk of such being developed (CM3), but it was meant to be released in Q2 2016...

https://www.raspberrypi.org/forums/viewtopic.php?f=98&t=137952

https://www.raspberrypi.org/forums/viewtopic.php?f=98&t=146948&p=967458#p967458

So lets say we design the SOC Daughter PCB with CM3 in mind (with eMMC fitted), but we test it with the CM1. The daughter PCB will have the SOC related rear panel connectors arranged to fit the MDAC2 rear panel and the USB to XMOS solution this should be a viable RPi3 based solution :)

Hi John,

Is it planned to connect the RPi CM to MDAC2/FDAC via I2S or USB or both ?.

please ignore AFAIK RPI as I2S slave (clocked by DAC) cannot de done (due to limitations of the Broadcom SoC). But there are things like this https://volumio.org/product/allo-kali-i2s-reclocker/ that claim to re-clock and clean up the samples. If the RPI connection is to use I2S do you plan to implement something similar on the SoC daughter PCB ?.

If you do not plan to use I2S for the RPI do you plan to bring out the I2S / I2C interfaces on the daughter board header - so that it may be possible to use them later on without having to re-work any of the MDAC2 main PCB (to keep the door open to RPI alternatives in the future) ?.

By "USB to XMOS solution" do you mean an external USB connection (from say PC or MAC etc.) or is this a dedicated "internal" RPI-DAC USB connection (and if so will there still be an external XMOS USB connector for PC/MAC etc.. via DETOX) ?.

If the RPI is connected via USB to the DAC what will be the difference in SQ compared to connecting the DAC via an external USB to a PC/MAC via the DETOX ?.
 
AFAIK RPI as I2S slave (clocked by DAC) cannot de done (due to limitations of the Broadcom SoC). But there are things like this https://volumio.org/product/allo-kali-i2s-reclocker/ that claim to re-clock and clean up the samples. If the RPI connection is to use I2S do you plan to implement something similar on the SoC daughter PCB ?.
I find this very confusing. I was under the impression that the Kali reclocker could only be used with pi dacs which work in slave mode (pi as master) ie that there was several pi dacs which do work in master mode (Hifiberry?)
Perhaps I have misunderstood. Is it just that the latter sort of dac reclocks somehow?
 
I find this very confusing. I was under the impression that the Kali reclocker could only be used with pi dacs which work in slave mode (pi as master) ie that there was several pi dacs which do work in master mode (Hifiberry?)
Perhaps I have misunderstood. Is it just that the latter sort of dac reclocks somehow?

Yes you are correct. I was confusing I2S master /slave mode with an external master clock. Appologies for my wrong statement.

For further info :

Looking here https://www.raspberrypi.org/wp-content/uploads/2012/02/BCM2835-ARM-Peripherals.pdf on pages 129 -131.

For RPi CM1. The CLKM and FSM bits in the Mode_A register of the BCM 2835 can be set to change the PCM clock mode and frame clock mode respectively from master to slave. Enabling lock onto the incoming frame sync signal and PCM bit clock.

Presumably its the same for the BCM2837 as used on the CM3.
 
Status
Not open for further replies.


advertisement


Back
Top