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: OpenDRC-DI @ 96 kHz

OpenDRC-DI @ 96 kHz 1 year 9 months ago #34847

  • xnwrx
  • xnwrx's Avatar
  • Offline
  • Senior Boarder
  • Posts: 56
  • Thank you received: 11
  • Karma: 0
Hi all,
OpenDRC-DI datasheet states that the OpenDRC-DI (or mini-Sharc kit) can run at 96 kHz when in mono signal mode : "Mono signal: FIR filter with up to 12228 taps @48kHz, 6144 @ 96kHz"
This scheme is not described in the OpenDRC-DI user manual nor it is in the OpenDRC 2x2 plug-in.
Can someone could please confirm that the OpenDRC-DI can run up to 6144 FIR filter taps on one channel (mono) at 96 kHz. Or at least that it can run at 96 kHz ?
Thank you.
The administrator has disabled public write access.
The following user(s) said Thank You: neatlyfolded

OpenDRC-DI @ 96 kHz 1 year 9 months ago #34887

  • devteam
  • devteam's Avatar
  • Offline
  • Administrator
  • Posts: 8533
  • Thank you received: 1052
  • Karma: 157
xnwrx

If you want 96k, you can run the miniSHARC 4x8 plugin on the OpenDRC-DI but it won't be that many taps. For 48k, the max is 6144 FIR

DevTeam
MiniDSP, building a DIY DSP community one board at a time.
The administrator has disabled public write access.
The following user(s) said Thank You: neatlyfolded

OpenDRC-DI @ 96 kHz 1 year 9 months ago #34895

  • Richard
  • Richard's Avatar
  • Offline
  • Expert Boarder
  • Posts: 127
  • Thank you received: 68
  • Karma: 13
I'd be interested as well to know if it's possible to do this somehow.
Was it ever conceived as part of the OpenDRC design, even if the current 2x2 plug-in doesn't support it?
Being able to run both 6144 tap channels together as one mono 12288 tap @ 48kHz would be very nice. So would downsampling options...

The alternative MiniSHARC 4x8 plug-in can only import max 2048 taps per channel, so still doesn't help us reach either of these DSP targets the OP quoted above from the datasheet.
Last Edit: 1 year 9 months ago by Richard.
The administrator has disabled public write access.
The following user(s) said Thank You: neatlyfolded

OpenDRC-DI @ 96 kHz 1 year 9 months ago #34901

  • xnwrx
  • xnwrx's Avatar
  • Offline
  • Senior Boarder
  • Posts: 56
  • Thank you received: 11
  • Karma: 0
Thank you for these answers.
From what I understand then, no miniDSP product can run at 96 kHz and provide sufficient number of FIR taps to allow low frequency channel processing.
The administrator has disabled public write access.

OpenDRC-DI @ 96 kHz 1 year 9 months ago #34915

  • Richard
  • Richard's Avatar
  • Offline
  • Expert Boarder
  • Posts: 127
  • Thank you received: 68
  • Karma: 13
xnwrx wrote:
Thank you for these answers.
From what I understand then, no miniDSP product can run at 96 kHz and provide sufficient number of FIR taps to allow low frequency channel processing.

No, as DevTeam said, you CAN run 96kHz on OpenDRC (by using the 4x8 plugin) but you'll be limited to 2048 taps.
The Dirac DDRC-22 series which run native 96kHz in stereo 2x2 is MiniDSP's most powerful product for 96kHz use, and offers 3072 taps per channel at 96kHz. That's exactly half the 6144 taps per channel offered by OpenDRC 2x2 running native at 48kHz, because they're both built from the same MiniSHARC engine, but have different firmware - in Dirac's case the more expensive proprietary kind!

You can always process low frequencies to a dramatic extent with even a very modest number of taps, but for very fine precision corrections you may not have enough taps, and that's why 48kHz DSP is better than 96kHz DSP for the woofer channel because you'll get more taps.
If you're running broadband audio through FIR, then you won't be correcting for the speaker's multiple driver's outputs properly anyway.

Exactly what is a sufficient number for low frequency processing depends on the complexity of target EQ / phase curve you're trying to achieve in that frequency range. If it looks like the teeth of a comb, then no - you probably won't have enough taps to match it, but I'd probably question the measurement method that resulted in this graph , as it's highly unlikely your woofer's natural behaviour is that erratic, and more likely you're seeing comb filtering across the measurement due to wall / floor / ceiling reflections in the room.
If however, you've mounted your measurement microphone to the physical speaker cabinet on axis with the driver using a clever home-made metal bracket so it's rigid in a fixed position relative to the drive unit and repeated your tests 64 times while both rotating the whole speaker + mic rig around to numerous different angles and moving location through the room, and then taken a mathematical average of all the measurements, then it's likely you'll see a much cleaner, smoother, purer graph of your woofer's true response (loaded inside that cabinet) and the nasty room reflection comb-filter peaks and dips will be significantly averaged out and almost disappear.
Then the woofer's target response will be much easier to match with an FIR filter correction, because it should be more gentle rises and falls in level which is a less complex target.
Last Edit: 1 year 9 months ago by Richard.
The administrator has disabled public write access.
The following user(s) said Thank You: neatlyfolded

OpenDRC-DI @ 96 kHz 1 year 9 months ago #34917

  • xnwrx
  • xnwrx's Avatar
  • Offline
  • Senior Boarder
  • Posts: 56
  • Thank you received: 11
  • Karma: 0
Hi Richard,
Happy new year.
Thank you for this detailed and clear reply. I 100% agree with you.
My project is not only speaker/room correction but also multi-amp with sharp cross-over and phase correction for each driver using FIR filters.
I'm currently using an OpenDRC-DA8 (modified to get one SPDIF stereo output) and my current system consists of two full-range line-array complemented with one 18" Sub for the low end. OpenDRC line out 1 drives a class-D amp for the Sub. SPDIF out drives both full-range drivers through an FDA amp (FX D802). The 2048 taps per channel, are just enough to cross Sub and Full-range channels at 125 Hz, applying phase correction and drivers equalisation using the 48 kHz plugin. Just enough is even not enough as it limits cross-over slope to 48 dB/Oct (I would like to apply 96 dB/Oct) and prevent correct Sub equalisation.
I'm going to add tweeters (Mundorf's AMT ones) to my full-range Line-array crossed around 3 to 4 kHz. Time o upgrade my FIR system :
- 96 kHz sampling for 96 dB/Oct crossing and phase and amplitude equalisation of the tweeters with FDA amp (OpenDRC-DI or SPDIF modified OpenDRC DA8 can do the job as it doesn't require more than 1024 taps per channel to do so)
- 48 kHz sampling for 96 dB/Oct crossing and phase and amplitude equalisation of the full-range drivers with FDA amp (OpenDRC-DI will do the job with 6000taps per channel)
- 48 kHz sampling for 96 dB/Oct crossing and phase and amplitude equalisation of the Sub driver (OpenDRC-DI will do the job with 6000 taps or more, but need to go to digital input amp).

As you can see, this project requires 3 OpenDRC devices, plus a NanoDigi necessary to dispatch SPDIF signals to the OpenDRC.... that's a lot !

We need a more powerful device, or at least an OpenDRC-DI-4 with 4 digital outputs, and 48kHz and 96 kHz pluggin and 12000 FIR taps total, 6000 per channel possibly.
The administrator has disabled public write access.
The following user(s) said Thank You: neatlyfolded

OpenDRC-DI @ 96 kHz 1 year 9 months ago #34918

  • Richard
  • Richard's Avatar
  • Offline
  • Expert Boarder
  • Posts: 127
  • Thank you received: 68
  • Karma: 13
Happy new year to you too.
Yes, if they made a more powerful OpenDRC type variant with more taps, I'd be first in the queue to order one too.
I myself own four OpenDRC boxes (with both DA-FP and DIGI-FP output boards so I can swap analogue or digital as needed.) I use 3x OpenDRC for my tweeters / midranges / woofers as an FIR crossover setup.
I just wish MiniDSP made one big box perhaps 1U rackmount that did the whole job as a 2x6 or 2x8 FIR crossover / loudspeaker processor, but I've been asking for that for 5 years or more... Maybe there's not enough people out there like wanting stuff like that. Makes you wonder how DEQX can sell their stuff for £4000+ which does that with only 4800 taps per channel. OpenDRC solution is much cheaper and better performance, but feels like a work-around with all the separate boxes and extra cables needed.

If you really need a LOT of taps in a piece of FIR hardware, you need to look around for a Logidy EPSI convolution processor, which is really a guitar effects pedal costing about $199 that's intended for reverb or amp cabinet simulations, stereo in / out via unbalanced jacks, but it can run 65,536 tap FIR filters!!! The trouble is, if you made a linear phase response with the impulse peak in the middle of that filter, you will then have a very long audio latency delay to deal with, but for playback only (not live audio or video sync) applications it could be workable.
The administrator has disabled public write access.
The following user(s) said Thank You: xnwrx, neatlyfolded

OpenDRC-DI @ 96 kHz 1 year 9 months ago #34920

  • xnwrx
  • xnwrx's Avatar
  • Offline
  • Senior Boarder
  • Posts: 56
  • Thank you received: 11
  • Karma: 0
Hi Richard,
I was not aware of this EPSI convolution processor. Interresting. But it's an analog in/out engine.
I prefer MiniDSP, and hope for a digital in/multiple out miniDSP capable of running 48k as well as 96k with 12000 taps overall at least, capable of minimum 6000 taps per channel. OpenDRC is not far away from that, but its not.

The NanoSharc kit seems to be perfect with 8 in channels and 8 out channels, a more powerfull DSP than miniSharc kit, but the plugin limits to 4096 FIR taps, 2048 max assignable per channel.
Last Edit: 1 year 9 months ago by xnwrx.
The administrator has disabled public write access.
The following user(s) said Thank You: neatlyfolded
Moderators: devteam