DMAS Design [SPDIF-Optical]

[Ferather]

Here is my DMAS (Digitally Managed Audio System) design (lossless up to, into the speaker), for anyone interested. I am trying to eliminate additional components (cheaper, less THD THD-N, lower latency).



----

Given the current standard IEC 60958-5:2021, the current consumer standard 15 channel SPDIF and 15 channel programmable SPDIF transmitter, a DMAS can be built easily.
GPU's (especially with HDMI 2.0+) can do 1536k total sample aggregates (HDA's max, ~37mbps), which is: 8 x 192k = 1536k | 15 x 96k = 1440k (34.6mbps).

If we stick with the tick box system, and update Windows, we can add the formats needed to SPDIF, although globally (shows on all SPDIF devices).
Unfortunately HDA can't support 15 x 192k, as its too much bitrate and total samples, 15 x 192k = 2880k (~70 mbps).

Windows also needs to support more than 8 channels for PCM output, try and set more than 8 speakers using Windows config.

----




----

The DSP amp can have a set maximum decibel output (for example, full volume = 120dB out), speakers can have a set input, and a larger input can be attenuated (losslessly).
Technically speaking the total audio volume could be used to regulate the total power a single speaker can draw, and the same for output for the DMAS.

You can image the main unit of the DMAS as a PCM processor or SPU (sound processor unit), with-without DTS-Dolby-Other PCM enhancements.
32 bit float should make volume and power management at the PCM level, without amp (no Class-D), doable.

---

Please note that a DMAS does not make DTS-Dolby redundant, instead more into PCM processors at the DSP point (external main unit, not software).
The compressed formats can-are used to bypass channel restrictions (such as Windows, 8 channel) and reduced bandwidth.

Note that while compressed audio formats decode to PCM, they passthrough PCM processors (not in PCM form, inside a container), until decoded.
Since the main unit is working at the input level (PCM), formats can be decoded as normal, then continue to processing as PCM.

 

[Ferather]

SPDIF-Optical, HDAudio.

 

[Ferather]

Example speaker PowerDAC, however with a DMAS the PowerDAC is built into the speaker unit, and is mono. Speakers should be the only point consumers can change audio quality.
All speakers should ideally use some sort of standard, or detection could be added at the DSP unit, alternatively use changeable PSU's, similar to computers.

= | PCM @ 100dB > DMAS Main Unit @ [X]% Volume > PCM @ per channel dB (100dB) > PowerDAC > PWM @ 100dB > 100dB SPL
+ | PCM @ 50dB > DMAS Main Unit +50dB (32b float) > Per channel, (100dB) > PowerDAC > PWM @ 100dB > 100dB SPL
- | PCM @ 144dB > DMAS Main Unit @ [X]% Volume > Per channel, (100dB) > PowerDAC > PWM @ 100dB > 100dB SPL

A match input volume and/or match input volume up to XX, feature would also be useful. Amplify and amplify up to XX could also be a mode.



32 bit float can be used for all PCM processing at the main unit.

====

Also note SPDIF does not specific a bitrate, instead the hardware used determines total bitrate (bandwidth).
Optical speaker cables can be moulded with the 3.5mm ends, which can be used to carry power.

 

[Ferather]

In consumer mode, we have up to 15 channels PCM and up to 108 compressed (@ 48k/24b), speaker configuration could be setup based on the input format, or specified by the user.
Since we are not using a multichannel DAC, or any other bottleneck, the full bitrate of TOSLink can be utilized as a standard for parts used to produce a DMAS.

Most likely, a good mobile CPU SoC should be enough to run as the DSP and main processor (@ 32 bit float), opposed to current methods.


HDA (max ~37mbps) > 108ch compressed > DMAS > Uncompressed > 108 Speakers or single TOSLink to another unit.

====

Also note with HDA (HDAUDIO), using Windows (W10), SPDIF can be configured in the same way as Speakers and HDMI (configuration button), and work with the same registry data.
The 15 channel SPDIF transmitter listed in post 1 is programmable in terms of input and output channels (also 15, and don't care mode).

If I add formats to SPDIF (e.g. DTS-HD), and then convert form factor to HDMI, not only do the ticked formats and samples show, but also a 2 channel transmitter (as EDID).

 

[Ferather]

So now the question, did Yamaha update their SPDIF hardware (more than 2 channel optical receiver), and added all formats?
Note the use of dual 8 channel DAC's to get 16 channels, whereas the DMAS does not.

 

[Ferather]

PWM DAC | Open Music Labs

Using a PWM for high quality audio generation

www.openmusiclabs.com

PowerDAC V2.1 | (Low THD+N, PWM, Headphone version)

 

[Metal-Tom]

Hi
That will do much cost of Money, ohhweeee xDD

 

[Ferather]

No analogue parts, not as much as you think, excluding the speaker PSU (power supply), I would guess at the price of a high end Android (Mobile CPU) Smart TV-IPTV 4K/8K box, for the DMAS main unit.
If we then included DTS-Dolby licensing both in terms of decoding and effects, then the price of the main unit would go up. The DMAS main is separate from speakers.

As far as the speakers, that's down to speaker manufacturers, and how they decide to handle the 1 channel PCM input from the DMAS main unit.
The speakers will be the heavy cost point, although the system as a whole reduces parts used (and total price).


I guess two separate forms of end units can exist, an assembled unit (like the Logitech Z906), or main unit + power matched (PSU) user selected speakers (AVR style).
It's also possible for the DMAS main unit to handle analogue in, by instantly converting it to PCM and continuing as normal.


Natively of coarse being digital, HDMI eARC (higher specification, up to ~37mbps) can be supported | Preferably no video as it's a waste on the GPU and power.
Also note, its 100% possible for a GPU to utilize the current consumer standard SPDIF and TOSLink module, GPU's use HDA.

----

Note, the DMAS main unit will also have a THD - THD+N rating, due to components, but it should be very low (non audible). No other real rating other than the PSU.
DMAS compatible speakers don't need matching with the amp, there is no amp, however it must be compatible with the PSU of a DMAS (v??) unit.

Optical is immune to electrical signals such as EMI-RFI, DAB-FM, WiFi, Bluetooth, faulty lights (so on), surrounding power (cables). Light is also faster than electricity.

 

[Ferather]

Based on the Pioneer DJ XDJ-XZ DJ System, as an example (not the price), I would estimate the SNR and THD of the DMAS main to be about: 120+ dB / 0.002% THD.
I selected the units USB specifications as its digital, but still uses conductive circuit, which should mean an improvement with optical (no power).

I think even a cheaper DMAS main unit, or sound processing unit (SPU), at 0.05% or 0.5% THD would be acceptable.
Given the mobile CPU, frequency response should be down to the sample rate.

Technically speaking, if built in a certain way (and working with a mobile CPU), there should be no technical limit to samples, bits and number of channels.

----

The price of the following boards made me laugh, still, lets add some better parts, - the DAC ,and, + PWM module, and of coarse mono, per channel-speaker.




----

Note, in order to use 15 channel SPDIF, its ideal to use the 125mbps optical modules, although HDA can only output ~37mbps (input to DMAS).
Supporting the full rate allows a DMAS to have a firmware update in the future, such as adding formats, or other support.

If an end unit is a stereo unit, it can still accept 15 channels for spatial processing or ignore them, so 15 channels can be a standard.
I am not 100% sure why 15 and not 16 channels, but I guess the 'Don't care' can be utilized as 'Any' channels.

 

[Ferather]

Google search: Lossless Optical HDMI (note, bandwidth is shared, mostly video).
Since a GPU uses HDAUDIO, final audio bitrate is ~37 mbps.

 

[Ferather]

SPDIF Optical - Realtek S1220A (with Crystal Sound 3).

----



That's my [fiber optic] broadband, UK to China ping.

 

[Ferather]

Manchester and Biphase Mark Code Encoder/Decoder Implementation

Implementing Manchester and BMC encoder/decoder on EFM8LB1 1. Introduction The Manchester code and Biphase Mark Code(BMC) are popular l...

markdingst.blogspot.com

Unfortunately SPDIF does define a version, unlike HDMI and USB. Its based on hardware used.

 

[Ferather]

 

[Ferather]

Specs of a high grade Class-D amplifier (not that the DMAS uses a hardware amp), note the THD % for a Class-D (PWM), image below.
Here is an older but similar product to my DMAS design, although it includes an ESS Sabre DAC.



Should translate to 99.5% pure audio. What is True Sound?

----



----

[Lossless True Sound] or [True Sound] should be a badge!

====

Note to OEM's, if you are going to build a device with SPDIF, please use the current [consumer standard], not legacy standard.
This is the equivalent to implementing HDMI at a lower than current standard such as HDMI 1.0

15+ channel SPDIF transmitter > TOSLink (T-R) > 15+ channel SPDIF receiver > [X] Channels > DAC (could be 2 channel).
If you are using ESS Sabre DAC's please check the chip manual for SPDIF consumer support (example).

You should also consider contacting ESS in regards to the 'Don't Care' mode, it should mean 'Any Number'.

https://www.boomplay.com/songs/73560755?from=artists | output device @ 92%.

 

[Ferather]

84 GBd (168 Gbit/s) PAM‐4 3.7 Vpp Power DAC in InP DHBT
An Ultrahigh-Speed Low-Power DAC Using InP HBTs


SPDIF says to HDA: "Going to need more bandwidth"
HDA responds: "Can't do more than ~37 mbps"

====



E1DA PowerDAC V2 Review – Soundphile Review
E1DA PowerDAC V2 Review – AudioDigitale



====

 

[Ferather]

ESS ES9038PRO (Page 42) | Cirrus Logic CS42528 | Both are 8 channel DAC's, not 15 mono PowerDAC's (in the speaker unit).



Full rate of HDA (and HDMI Audio), 1536K total samples max.

----

Molex SMI Optical Interconnects (250 Mbps) | TMDS - HDMI (Transition-minimized differential signalling).





Welcome to the optical age! (Digital > Optical).

====

If a mono headphone PowerDAC is small enough, it could fit per ear, using Optical with USB power or USB.
24 bit can go up to around 144 dB, so a set maximum output amplitude will be needed.

Volume set to maximum (made up example):

[-] Input from PC @ 144 dB (PCM) > Optical/USB > Stereo splitter (PCM) > Mono PowerDAC @ 88 dB (SPL).
[=] Input from PC @ 88 dB (PCM) > Optical/USB > Stereo splitter (PCM) > Mono PowerDAC @ 88 dB (SPL).
[+] Input from PC @ 48 dB (PCM) > Optical/USB > Stereo splitter (PCM) > Mono PowerDAC @ 88 dB (SPL).

 

[Ferather]

2m Toslink (SPDIF) Male Black Cable - Blake UK (blake-uk.com)



Being capable of 125mbps is enough of a spec

Optical Digital Audio Cable (10 Feet) | Mediabridge Products



====

In terms of PCM, and even formats such as FLAC, there is no licensing or copywrite in the way, although SPDIF does now have copywrite protection.
Sound output devices, such as a Realtek, that has SPDIF, will need to support a device attached using the current consumer standard.

Lets says an AVR OEM decides to utilize 8 channel SPDIF for their DAC implementation, how would they receive it?

The best approach for all sound output OEMS, is to stick with the 125Mbps modules, and update to any changes made to the consumer standard.
The total supported channels is really down to the receiver, which has the DAC, or other, such as the DMAS (which is less limited).

In terms of a 2 channel receiver, there is no real reason why the extra channels can't still be received and converted to spatial, directly.
Windows also needs to update its native support for SPDIF, as there is no channel config or additional formats (normally).

As as sound output device OEM, this is not so much your problem, however utilizing a stereo restricted setup is down to you (parts, standards).


In short, the output device should support 125mbps, be ready for updates, and support 15+ channels, all formats, natively.

 

[Ferather]

THX Ltd. Announces New THX Interconnect Cables to Elevate Home Theater Entertainment

THX Ltd., a world-class high-fidelity audio and video tuning, certification, and technology company, today announced it is launching THX Interconnect, designed in partnership with Pixelgen. THX Interconnect are a family of Ultra High Speed HDMI 2.1 cables capable of delivering 100% uncompressed...

www.techpowerup.com

I will giggle if in 12 months THX put out a DMAS spec!

----

Random brand, optical? (I searched for), lossless, DisplayPort 2.1, 80 Gbps, £14 for 3M. (edited, not sure)

Is a Fiber Optic HDMI Cable Worth it for Game Streaming? (avaccess.com)

 

[Ferather]

 

[Ferather]

Note: It should be the case that with consumer configuration, channel number 0x0 (Don't Care) translates to 'Any Number', essentially 16+.
SPDIF professional configuration does not seem to specific a maximum number of channels, or final bitrate.

SPDIF still seems to stick to its original design, without a maximum defined stat, but instead the hardware used for SPDIF.


Current JIS F05 (TOSLink) specifies up to 125Mbps NRZ (PAM2), with PAM4 this would be 250Mbps. Without cable or module changes.



Since the DMAS does not use a single DAC, 216 individual channels can be achieved, as the PowerDAC is per speaker.

 

[Ferather]

OEM based DMAS, in this example a DTS DMAS:

Number of channels: Unlimited, set by physical hardware, total available input (HD Audio, PC), TOSLink (or similar) mono outputs |> mono P-DAC's (speakers).
CPU: What ever is needed to process [x] number of channels in 32 bit float format. Probably ARM (mobile SoC, low power).
OS: OEM based, a made up example would be 'DTS Processing', includes user interface and phone app.
Modes: True Sound (no changes, other than amplitude), DTS Processing (SRS, other).
Processing: Software, well CPU based, similar to Windows.
Acceptable audio delay: 0.1ms maximum.
Number of amplifiers: None.

Speaker power supply: [x] watts for [x] number of speakers (P-DAC's).


Note, If I am sending 102 dB of audio recording (PCM), I should get 102 dB output from the speakers (without amplitude changes).

 

[Ferather]

Seems some speaker PowerDAC's (essentially a DMAS) hit the market a while ago, but they are extremely expensive (other parts).

Linn's flagship 360 is the "finest loudspeaker" it has ever made | What Hi-Fi? (whathifi.com)
Here is an older headphones PowerDAC video (note cheaper, better overall).

It seems even with best possible parts, the best choice is a PowerDAC (and the use of optical over copper).


Note the main unit of my DMAS design will work with any flagship speaker, or cheaper, as long as the power supply is capable.
The DMAS design requires a specific cable, given an optical fiber (audio signal) and two copper (power supply).

It's also possible to allow the main unit to change-replace the PSU (like a computer).

Any digital input (essentially PCM in), can be used, SPDIF (15+ ch), HDMI, DisplayPort, Bluetooth, USB, WiFi, so on.

----

PowerDAC-SX - ECdesigns | This one has built in DA converters, opposed to the DMAS, which does not (cheaper).



----

What is True Sound? The concept explained | Stuff

 

[Ferather]

This is more in relation to the optional modes, not True Sound mode: https://www.soundonsound.com/techniques/new-harmonics
Some people like A/B amps due to THD, and in some cases certain amps. Digitally its possible to replicate.

There is a difference between True Sound, and added THD-other you like.

People can-will pick between Dolby and DTS, because they prefer the sound, well 'added' processing is also not True Sound.
You can also pick between two sets of headphones or speakers of the same value, and get different sound.

====

DMAS main unit (sound processing unit + power supply) > DMAS optical audio cable (1 channel, speaker) > Speaker [PowerDAC].
Software amplifier via multicore CPU, 32 bit float (1528 dB) more than any physical amp, bitperfect-lossless.

24 bit ,144 dB input > DMAS, SPU | 32 bit float > additional processing, adds 24 dB > attenuation to 24 bit, 144 dB.
Note that the additional 24 dB did not add clipping or any kind of distortion, no where near 1528 dB.



Example mock cable, set or custom length.

====

Since humans generally hear around 20Hz-20kHz, and sample rate /2 = frequency range, speaker specs will look like this:

Bit depth: 16-24bit (96-144 dB max).
Sample rate: 44.1-48k (22.05-24kHz max).
Speaker response range: [X]Hz-[Y]kHz (max).
Power draw: [X] watts (DMAS standard).
Total THD, THD+N: [X]%, [Y]%.
Output SPL: [X] dB.

 

[Ferather]

Some additional notes: ECdesigns (General, Philosophy) | Class-D (PWM) | True Sound (Minimal Hardware).
SPDIF: CS8416 (8 Channels) | TAS5558 (PWM Modulator). Combo chip possible?

 

[Ferather]

Someone asked me if this also possible via WiFi, I answered yes as long as the DMAS main unit had a wireless transmitter and the speaker a wireless receiver (and power source).
Not sure I would personally use that option (wired vs wireless), even more so when we include optical, which is immune to WiFi, Bluetooth (other RFI-EMI).

Still, yes. And also a DMAS main unit can have as many input and outputs as needed, so it can replace a traditional AVR.

====

The image below, I created a bi-directional optical system for HDMI-DP. Single specific lanes of single fiber, a multi-fiber cable.

The video lane could start at 100 Gbps (NRZ-PAM2), and later move to 200Gbps (other encoding, example PAM4).
The audio and data lane can-will also have their own bandwidth, 125Mbps audio 10Gbps data.

This would require a new version, due to the built in optical transmitters-receivers (not in the cable or connector).
Blanking periods for audio-other are not required (data island period).

A device can be audio only, video only, data only (network-internet), combination, with-without return.




48k samples (24kHz Nyquist frequency), 24 bit x 108 (channels-speakers) = 124.4mbps.