Should be no clicks or pops even from a cold start
cheers
I can't wait to throw money at you
Total number of taps depends on the sample rate and the tap rate which is 1.25 nS per tap for this dsp. This gives us a total number of taps of:-
4K taps @ 192KHz
8K taps @ 96KHz
16K taps @ 48KHz
but that is with the core running flat out processing the FIR and not doing much else. Distributed over 2 channels you can divide those figures by 2. With 1:4 decimation it is possible to achieve 16K taps at 192KHz or 8K per channel for stereo or 2K taps per channel for 8 channels.
Now it is possible to offload the FIR's onto the accelerator which requires almost no core intervention and has similar tap execution time to the core. One can expect a similar tap execution time but there are some other overheads involved with this. I am still looking into the pros and cons of this feature. I wrongly assumed that I could process 8 channels concurrently with this feature but there are only 4 MAC units assigned to the whole of the 32K taps and not to each 1K segment as I initially assumed
cheers
Last edited:
If I understand this correctly, at 96kHz there's 4k taps for Left and Right; freely distributable among the channels. So in a multiway speaker setup, one could use for example 1.8k taps for woofer lowpass, 1.8k for mid bandpass, and 400taps for tweeter highpass.
And in addition to that the "standard" IIR stuff.
A feature I would really like would be a limiter with user definable sidechain filter. So one could design the limiter to act based on driver excursion, not just input volume.
And LED outputs to visualize, when the limiter is engaging.
And another cool thing would be mid/side EQ.
And lots of presets, with control outputs to switch amps, and quick switching times for usable AB comparisions between presets.
Last edited:
Interesting design!!!
I'm currently using a miniDSP with DIRAC + DBX 4820 to manage my 2x4-way active speakers. I would like to replace this by an integrated system like yours with digital and analog inputs and with the possibility of using DIRAC in order to do DRC. ¿Do you consider feasible to integrate my current minidsp in-between your system? May be, at the point just at the output of the digital input and after the AD converter.
Thanks
I'm currently using a miniDSP with DIRAC + DBX 4820 to manage my 2x4-way active speakers. I would like to replace this by an integrated system like yours with digital and analog inputs and with the possibility of using DIRAC in order to do DRC. ¿Do you consider feasible to integrate my current minidsp in-between your system? May be, at the point just at the output of the digital input and after the AD converter.
Thanks
Yes there are 7 programmable status leds and the board should be able to handle your FIR requirements too. The other DSP functions would be a work in progress but still doable even with all of the FIR stuff happening in the background.
The board has one 12 volt trigger output to switch on one amp but this can be expanded to another four more with an expansion board. Most amps have the ability to chain triggers to power up more amps - hopefully with a time delay
If not I can accomodate this feature along with optional time delays.cheers
Yes. Please do.Most amps have the ability to chain triggers to power up more amps - hopefully with a time delay
cheers
I suppose there is an input selector that permits to chose between the different digital inputs and the output of the ADC converter (all of them, digital signals) I was speaking about to connect the miniShark board of the miniDSP DDRC-22D, I have, just in this point. Preferably through an I2S connection. ¿Is it possible?
Thanks
Thanks
Reading LTD's Question... Not sure if this is what he was thinking... but a good feature/possibility either way...
(based on my limited technological experience)...
Having an i2s out/in (think effects loop in pro gear) in the signal chain prior to the on board DSP would open up a lot of possibilities today and in the future for either third party solutions like using the minisharc board inside of the miniDSP DDRC-22D for Dirac Live or for future in house 2channel DSP upgrades/solutions.
Thoughts ???
GM
(based on my limited technological experience)...
Having an i2s out/in (think effects loop in pro gear) in the signal chain prior to the on board DSP would open up a lot of possibilities today and in the future for either third party solutions like using the minisharc board inside of the miniDSP DDRC-22D for Dirac Live or for future in house 2channel DSP upgrades/solutions.
Thoughts ???
GM
If I have understood correctly, the project of Tranquility Bass is like that:
Inputs -> Selector -> ADC (If Analog) -> DSP (Crossover) -> DAC -> Outputs
What I want to do
Inputs -> Selector -> ADC (If Analog) -> DIRAC -> DSP (Crossover) -> DAC -> Outputs
Where DIRAC is meant by a miniDSP (one miniSHARC card) with DIRAC installed. This board would be connected via I2S connection to the DSP that brings the "Tranquility Bass" system and, in my case, this DSP would work as crossover.
Is this the DIRAC unit you are talking about ?
DDRC-22D | MiniDSP
It only has TOSLINK or AES-EBU input or output. There are no I2S inputs or outputs.
Currently I have TOSLINK Input/Output and one AES-EBU input but not an output.
regards
I believe the minidsp DDRC-22D is built around a minidsp minisharc kit with a custom/locked Dirac Live firmware. If this solution is doable/feasible, the ability to switch between the four minisharc presets would be useful as well (this is currently handled by the minidsp Vol-fp).
I must say though... besides the Dirac Live partnership, Tranquility Bass' system reads to be a much higher quality spec/feature dsp preamp processor solution... certainly approaching and in the end hopefully competing/surpassing the usual suspects in the "high end" audio circles.
GM
I must say though... besides the Dirac Live partnership, Tranquility Bass' system reads to be a much higher quality spec/feature dsp preamp processor solution... certainly approaching and in the end hopefully competing/surpassing the usual suspects in the "high end" audio circles.
GM
Yes it is. Inside there is a board like this http://www.minidsp.com/images/documents/Product Brief-miniSHARC.pdf programmed with DIRAC. This board has I2S inputs/outputs. The idea is to extract the board from the DDRC-22D and to connect it to your system as I explained before. The question is whether it is possible or not.
Thanks
Before of seeing the system of TB in this foro, I was thinking about to make a system similar to the TB's system. Specifically, I was thinking to integrate an ADC based in SABRE ES9102, two miniSHARC boards, and DAC buffalo. The system of TB is very similar to I was thinking about. I like how DIRAC does DRC. This is the main reason of my interest in the possibility of integration with TB's system.
Please add the necessary analog inputs and loopbacks enabling realtime optimizations during the setup using pink noise audio :
- need to measure the transfer function of the loudspeaker, in realtime
- need to measure the transfer function of the correction, in realtime
- need to measure the global transfer function (correction + loudspeaker), in realtime
Three audio signals to monitor :
- pink noise
- DSP filter output
- measurement mike output
Three transfer functions to monitor :
- correction = DSP filter out / pink noise in (gain and phase)
- overall = measurement mike / pink noise in (gain and phase)
- loudspeaker alone = measurement mike / DSP filter out (gain and phase)
Need to get rid of the massive delays introduced by the ADC's and DAC's.
An easy way is to provide analog loopbacks. Doing so you don't need to fiddle with ADC's and DAC's delay compensations. Or course you'll remain in charge of taking the FIR filters delays in account.
If there are enough MIPS you can specify a target transfer function (in gain and in phase), and attain it by running a LMS (Widrow-Hoff) during the setup phase. This would lead to a purely FIR filter correction, say three times a 8k FIR filter in a 3-way system. Or possibly, if you want to spare MIPS : a 8k FIR filter for the bass, a 1k FIR filter for the midrange, and a 128 FIR filter for the treble.
Need to do the bass setup, the midrange setup, and the highs setup, all separately.
Finishing with a global setup (all channels working), providing some global equalization / pre-processing for shaping the deep bass range, say from 10 Hz to 100 Hz using a Linkwitz Transform (IIR-based deep-bass equalizer requiring the 32-bit precision), hopefully with variable Fc in function of the volume setting.
This would be unique and powerful, while remaining elegant and simple, conceptually speaking.
Need to ensure a decent communication (using USB) between the DSP and the PC (running Windows) for guiding the user, through the setup, in a clever way, letting the user learn and understand what the system is doing. The system needs to export three measurement curves to the PC, at the same time (see above). If you want to spare DSP MIPS, the system needs to export three kinds of audio frames (see above) to the PC, letting the PC calculate and display the three transfer functions.
In 2015, a dedicated audio DSP platform must provide this.
- need to measure the transfer function of the loudspeaker, in realtime
- need to measure the transfer function of the correction, in realtime
- need to measure the global transfer function (correction + loudspeaker), in realtime
Three audio signals to monitor :
- pink noise
- DSP filter output
- measurement mike output
Three transfer functions to monitor :
- correction = DSP filter out / pink noise in (gain and phase)
- overall = measurement mike / pink noise in (gain and phase)
- loudspeaker alone = measurement mike / DSP filter out (gain and phase)
Need to get rid of the massive delays introduced by the ADC's and DAC's.
An easy way is to provide analog loopbacks. Doing so you don't need to fiddle with ADC's and DAC's delay compensations. Or course you'll remain in charge of taking the FIR filters delays in account.
If there are enough MIPS you can specify a target transfer function (in gain and in phase), and attain it by running a LMS (Widrow-Hoff) during the setup phase. This would lead to a purely FIR filter correction, say three times a 8k FIR filter in a 3-way system. Or possibly, if you want to spare MIPS : a 8k FIR filter for the bass, a 1k FIR filter for the midrange, and a 128 FIR filter for the treble.
Need to do the bass setup, the midrange setup, and the highs setup, all separately.
Finishing with a global setup (all channels working), providing some global equalization / pre-processing for shaping the deep bass range, say from 10 Hz to 100 Hz using a Linkwitz Transform (IIR-based deep-bass equalizer requiring the 32-bit precision), hopefully with variable Fc in function of the volume setting.
This would be unique and powerful, while remaining elegant and simple, conceptually speaking.
Need to ensure a decent communication (using USB) between the DSP and the PC (running Windows) for guiding the user, through the setup, in a clever way, letting the user learn and understand what the system is doing. The system needs to export three measurement curves to the PC, at the same time (see above). If you want to spare DSP MIPS, the system needs to export three kinds of audio frames (see above) to the PC, letting the PC calculate and display the three transfer functions.
In 2015, a dedicated audio DSP platform must provide this.