Welcome, Guest
Username: Password: Remember me

NOTE: This is a "Community" forum. Please be mindful that community members are here to help as part of a community effort. We therefore appreciate your effort in keeping this forum a happy place!

If you have a specific issue (e.g. hardware, failure) and want help from our support team, please use our tech support portal (Support menu - > Contact Us).
Thanks a lot of your help in making a better community.

TOPIC: 3x DA-FPs & 3x miniSHARCs in series & parallel

3x DA-FPs & 3x miniSHARCs in series & parallel 3 years 7 months ago #23880

  • Thunderstruck
  • Thunderstruck's Avatar
  • Offline
  • Senior Boarder
  • Posts: 59
  • Thank you received: 8
  • Karma: 5
Dear community,

I starting this thread to find a way to get stereo SPDIF signals into a mono 6-way setup.

You may not be interested in what I'm doing with a mono 6-way speaker but you will probably find the discussion interesting because your mind will race with new ideas and things you could do if you only knew how to tie these awesome miniSHARCs together in series and parallel.

This thread started from my "miniSHARC/miniDAC8 stacking question" and I thought I should start a new thread to discuss using DA-FPs to get the SPDIF into the miniSHARCs and the processed analog out of them.

I am proposing a setup attached below to get SPDIF into 3 different miniDHARCs and 6 processed analog signals out of the DA-FPs (2 each per 3 DA-FPs).

Please read the thread "3x AN-FPs & 3x miniSHARCs in parallel" to understand why I am proposing this.
I could repeat my reasons in this thread as well and I may add that later.
For now, if you just read my "3x AN-FPs & 3x miniSHARCs in parallel" thread then you will understand.

The following is a start on a discussion of my proposals. I will build this setup if I have to in order to determine if it will work. I must buy 3 DA-FPs and a Vol-FP to try it and I hope to work some bugs out first to save some trouble. I don't care about money much.

1) The Vol-FP is attached to the 1st miniSHARC and used to adjust the overall volume of the whole setup. It controls the input gain to the 1st miniSHARC so all the others will be affected by it.

2) The miniSHARC 4x8 96kHz Plug-in matrix routing is used to sum the full range stereo SPDIF inputs and route full range mono signals to I2S Output 7 & 8 which is also routed as 2 mono signals to the Pin 2 SPDIF TX at 96kHz. The I2S output on channel 7&8 is also found using Pin 19,20,21 and a ground pin but is not useful because all 3 of the miniSHARCs must be I2S In/Out Masters. This is because they are all connected to DA-FP boards which are firm I2S In/Out Slaves. If we could figure out a way to get the I2S signal into the 2nd and 3rd miniSHARCs which are running the OpenDRC 2x2 48kHz Plug-in then that would be wonderful because we wouldn't have to accept multiple I2S - SPDIF conversions. I require the 2x2 Plug-in to get 6148 FIR taps per channel down stream. The 2x2 Plug-in doesn't support being an I2S slave input so there is no way to get an I2S signal into a miniSHARC running the 2x2. You can only get SPDIF into it via a DA-FP.

3) The 1st miniSHARC also puts out a processed output on channel 1 and 2 for the DA-FP's XLR analog outputs. I matrix a mono signal on both channels 1 & 2 and put out FIR limited bands of 2000-6325Hz on one and 6325-20000Hz on the other. This is processed at 96kHz sample rate so it is great resolution for high frequencies.

4) Remember. See my "3x AN-FPs & 3x miniSHARCs in parallel" to see why.

5) Now a tricky part. I propose to take the 2 matrix mono summed channels on the P2 SPDIF TX and a ground on P5 and drive both of the 2nd and 3rd DA-FPs RCA SPDIF inputs. This is a "Y" connection. See Rane note "why not Y". I think it will work because:
A) SPDIF is a very robust signal compared to I2S.
B) P2 SPDIF TX is theoretically 0 output impedance. My guess.
C) The RCA SPDIF inputs are theoretically infinite input impedance. My guess again.
D) I do this often with a single analog output split to several analog inputs. Not a guess but it's analog, not digital.
E) Is SPDIF to delicate for this?
F) Both digital and analog signals are just voltages on a wire used for communication. I.e. Not driving current.

6) Now we have a 96kHz SPDIF signal going into the 2nd miniSHARC/DA-FP running at 48kHz so it gets downsampled. No problem here. It gets converted to 48kHz, the sample rate of the Plug-in. This SHARC is processing lower frequencies so the lower sample rate is fine.

7) The 2nd miniSHARC runs the OpenDRC 2x2 48kHz Plug-in and outputs the 2 mid bands of 200-632Hz and 632-2000Hz on the 2nd DA-FPs 2 XLR analog outputs. We get to use 6148 FIR taps to process linear phase audio output on both ranges. An extremely easy task in this range. See the AN-FP thread for more info.

8) The 2 split mono SPDIF signals enter the 3rd miniSHARC/DA-FP as well. Now the lowest 2 bands are processed. 20-63Hz and 63-200Hz and output on the 3rd DA-FPs 2 XLR analog outputs. The 2x2 48kHz Plug-in is absolutely required to process the 20-63 band and the 63-200 band. See AN-FP thread for more info.

Lets discuss this setup. If it will work and how it could be done better.

Thunderstruck in Marietta, GA USA

File Attachment:

File Name: SPDIF&DA-FP.pdf
File Size:255 KB


File Attachment:

File Name: Speaker-2.pdf
File Size:336 KB

This message has attachments files.
Please log in or register to see it.

Last Edit: 3 years 7 months ago by Thunderstruck.
The administrator has disabled public write access.

3x DA-FPs & 3x miniSHARCs in series & parallel 3 years 7 months ago #24298

  • Richard
  • Richard's Avatar
  • Offline
  • Expert Boarder
  • Posts: 127
  • Thank you received: 68
  • Karma: 13
Looking at your PDF connection diagram for daisy chaining multiple DA-FP boards together from a common S/PDIF input, that may be something you can conceptually do, on paper at least, but each of the S/PDIF inputs will still have to pass through the Asynchronous SRC which is built into every DA-FP board, so digital audio will be locally reclocked by the crystal oscillator of that specific board without any reference to the other neighbouring board's crystal oscillators. So you'll actually get three very closely matched digital clocks to begin with, which over time will slightly drift further and further apart from one another because they're running off three seperate crystal oscillators. So your six mono outputs, or rather your three stereo pairs of outputs won't stay 100% in bit-for-bit digital sync with each other for much beyond several seconds typically.

If you can find a way to extract a master clock timing reference signal from one DA-FP and use it to slave the others to, that would be the solution maybe. Whether there's a significant audible problem to be worried or not isn't easy to say...... but if you're a perfectionist you ought to bear it in mind.
The administrator has disabled public write access.
The following user(s) said Thank You: Thunderstruck

3x DA-FPs & 3x miniSHARCs in series & parallel 3 years 7 months ago #24301

  • Thunderstruck
  • Thunderstruck's Avatar
  • Offline
  • Senior Boarder
  • Posts: 59
  • Thank you received: 8
  • Karma: 5
Dear Richard,

Thank you so much for your very thoughtful, insightful and clever response.
I don't have an answer for your proposal as I am inclined to agree.
Immediately, as I was studying your post, I'm thinking "Oh crap! He's right.".
I have already bought and just received 3 DA-FPs along with a 5th miniSHARC, miniDAC8, miniDC Isolator and VOL-FP.
So I am going to build the preamp as shown in the diagram and see how it works.
I will be wondering of a way to test the drift between the 3 different clocks. At 48,000 and 96,000 samples a second, I wonder how many samples the clocks could drift apart? They couldn't drift too much because the buffer would fill up. I wonder how that works? The source is squirting digits into the inputs at a rate regardless of whether the input can handle it. I can imagine that is what buffers provide for digital signals but I don't know that. I think buffers really provide infinite input impedance and zero output impedance (at least in my analog signal world). But could the digits flow into a tank like a traffic circle and swim around until the processor finally gets around to process them? I assumed there would be some difference in the latency of the 3 different clocks and I assumed that it would be consistent. Assuming they are out of whack time wise but consistently the same then I just adjust the delay to account for it.
Adjusting the delay was my plan going in but your suggestion is really making me wonder.

Now let me tell you that I have been in discussions with the DevTeam about a I2S solution for tying miniSHARCs together in series and parallel. They say they are working on a 2x2 48kHz Plugin with 6148 taps/chan that supports I2S Input Slave and Output Master. I am trying to convince them that the three plugin are good but actually aren't very useful for FIR processing. See, you have to have 6148 taps to modify low frequencies and you need a 96kHz sample rate for the highest frequencies. The pitiful little FIR length of the 96kHz plugin can only deal in the top third of the freq range. So the POWER solution is to use the plugins where they are strongest and tie them together in series and parallel. I offered to pay for the development cost but they said it is on their "to do" list.
In the mean time, I'm going to build my SPDIF solution .

I swear, I really appreciate your comments and thoughts. That is why I posted it, to see what I didn't think of.
Wow. The clocks drifting about here and there screwing up whatever delay I set and phase adjustment I make. I'll go crazy if this really is a problem. I won't sleep a wink tonight.
Thanks dude! I upped your karma.

Thunderstruck in Marietta, GA USA
The administrator has disabled public write access.

3x DA-FPs & 3x miniSHARCs in series & parallel 3 years 7 months ago #24308

  • Richard
  • Richard's Avatar
  • Offline
  • Expert Boarder
  • Posts: 127
  • Thank you received: 68
  • Karma: 13
Hi Thunderstruck,

Before you panic too much, I don't think the absolute master crystal clock drift between three seperate Asynchronous SRC output boards will screw up your speaker's coherence, in this case BECAUSE each of them is a self-contained D/A type board - which means their realtime incoming digital audio (which may be getting reclocked at a slightly different rate) is being then processed and converted into analogue output by the same local master clock in realtime, so even if the new local clock was running slightly faster at 48001 Hz sample rate in an absolute sense, the D/A process on that DA-FP board, under local control of same crystal clock, is also running at same identical 48001Hz rate so the local realtime analogue output is still generated in realtime and doesn't gallop ahead or drag behind because realtime in > realtime out always holds true. It just has more sample data values representing it during this intermediate step than it might otherwise have had on a slower clock. It's conceptually upsampling the input from 48000Hz to 48001Hz, just like upsampling to 96000Hz doesn't make it double-speed helium sound!

However, it would be a problem if you had three DIGI-FP digital in > digital out boards feeding into an external 6 channel DAC where you need all your streams to be locked in sync. But since your DACs are built into the DA-FP boards and thus independently running realtime digital in > ASRC to whatever new local clock rate > realtime analogue out, it doesn't make a lot of difference to your case, and I don't think the tweeter signal will come un-glued from the woofer signal or anything!

I've been banging my head against these technical issues myself since I purchased three OpenDRC boxes to run a stereo 3 way FIR crossover (woofer / midrange / tweeter) and had the same problems to contend with as you - but I was wanting the option to use external DACs. I spent a lot of time testing things from first principles to try and prove to myself if it was working as theory suggested. I had to ultimately expand my theory to incorporate the problems I discovered.

I also discovered that DA-FP board has a manufacturing fault where the analogue output is wired in inverted polarity to the (digital) input. I reported this to DevTeam and a few weeks later they officially acknowledged the fault did exist - "Oh crap! He's right." :laugh: and announced they would issue a firmware fix to correct this. That was back in Nov 2015. I'm not aware of the firmware fix being released yet - unless I somehow missed it ?????? - so I hope they're still planning to fix that soon.

I posted about the DA-FP polarity issue on this thread...

And I posted about the more general Asynchronous SCR re-clocking dilemmas on this thread...

There is one alternative workaround to use a Synchronous SCR device for convolutiuon FIR which is more expensive, and has fewer taps available, mentioned in detail on this thread...

Quote - "They couldn't drift too much because the buffer would fill up. I wonder how that works? "

When a buffer fills up past it's limit, you usully get a nasty pop / click in the audio after it is forced to restart again..
I've done test recordings on OpenDRC with digital in > out locked in sync where I was getting repeated clicks in the audio and figured out it was the ASRC that was causing it.

However, the way an ASRC works normally as a stand-alone device connected in series, means it re-interpolates new data values to continuously re-represent the incoming data at the new output rate, and there are no clicks or glitches intrinsic to the ASRC process. It could keep going all day and night without any pops or clicks. It's not really buffering anything - it's realtime in > out rate conversion..
Only a problem if you try and bit-for-bit align the digital audio before ASRC and afterwards back against each other - then they don't perfectly match - even if they're both nominally the same 48kHz sample rate. There's more detailed explanations of ASRC behaviour found in the links above.

Anyway, I hope your project works out well and if you get anywhere in solving the ASRC issues with this hardware I'd be grateful if you could share your successful technique.

Best wishes,

Richard
Last Edit: 3 years 7 months ago by Richard.
The administrator has disabled public write access.
The following user(s) said Thank You: Thunderstruck

3x DA-FPs & 3x miniSHARCs in series & parallel 3 years 5 months ago #25157

  • Thunderstruck
  • Thunderstruck's Avatar
  • Offline
  • Senior Boarder
  • Posts: 59
  • Thank you received: 8
  • Karma: 5
Richard,

Thank you so much for you most thoughtful reply. I can hardly wait to show you my latest progress. I've been sidelined for a while now because I needed to care for my mom and dad who are both 89 and have been in and out of hospitals and rehabs. I hope to finish my setup soon and post a success here. I have read all your posts and I truly admire you and appreciate your support here.

Thunderstruck in Marietta GA
The administrator has disabled public write access.
The following user(s) said Thank You: Richard
Moderators: devteam