Thanks Pos for your comments. However could you clarify?
I understand coming up with the XO needs significant effort. Are you also suggesting that the daisy chained approach actually makes it more complex?
I was thinking in terms of developing the IIR XO is najda, measuring the response and applying correction/compensation to the original stream in Jriver convolution using rePhase, to get a linear phase/magnitude response overall.
Thanks Orangeart, it does sounds like something which will need lots of measurements and tweaking.
No, not at all.
That is building a complementary and phase coherent minimum-phase 4-way system that is a feat in itself.
Phase linearization is the easy part.
Low-pass phase shift
Hi Thomas,
Do you remenber what was the version (0.9.x) with an 180° phase shift on the LP LR section ?
I've checked the amplifier polarity with a scope (input/output) to understand this phase shift.(all amplifier are non-inverter,exept one).
it was easy to reverse banana plug on the low section speaker.(or inverting polarity in rePhase).
mic measurement showed a null with "a normal polarity" in rePhase.
and other electric measurement too.
this is explaining why.
i was thinking there was something wrong with my method/mind.
Hi Thomas,
Do you remenber what was the version (0.9.x) with an 180° phase shift on the LP LR section ?
I've checked the amplifier polarity with a scope (input/output) to understand this phase shift.(all amplifier are non-inverter,exept one).
it was easy to reverse banana plug on the low section speaker.(or inverting polarity in rePhase).
mic measurement showed a null with "a normal polarity" in rePhase.
and other electric measurement too.
this is explaining why.i was thinking there was something wrong with my method/mind.
Last edited:
An English language idiom. See here(why 9 on smiley....i do not know).
https://en.wiktionary.org/wiki/on_cloud_nine
similar to - "être au septième ciel"
You seem to plan running 2 channel output from a computer via USB. Why not finish DSP treatment in the computer and run all 8 channels to a multichannel DAC? It would seem a lot easier to do alle the DSP in one step.
Yes. I had indeed considered this option in some detail. In fact even posted a thread comparing DSP on general purpose computer versus dedicated DSP processors.
http://www.diyaudio.com/forums/digi...sus-dedicated-dsp-platforms-loudspeakers.html
http://www.diyaudio.com/forums/digital-line-level/276412-uad-2-dsp-accelerators.html
Few options for 8 ch DAC are
1. MiniDSP USB streamer + Buffalo DACIII + IV
2. MiniDSP USB streamer + MiniDAC8
3. MiniDSP UDAC-8
USB Audio Streaming : U-DAC8
4. Really impressive Blackbird DAC
http://www.diyaudio.com/forums/grou...kbird-multi-channel-es9018-dac-group-buy.html
5. 8Ch option from ACKO DAC
https://sites.google.com/site/ackodac/home
6. Motu USB interface (Appears excellent)
MOTU.com - Overview
My main concern has been cost of a powerfull HTPC + one of the above compared to an old laptop coupled with a well integrated dedicated DSP like Najda.
But najda's new board is delayed and no word on when its expected. Maybe i should go back to the PC option.
How does a modern PC compare to Najda DSP in terms of processing power for loudspeaker management?
1. MiniDSP USB streamer + Buffalo DACIII + IV
2. MiniDSP USB streamer + MiniDAC8
3. MiniDSP UDAC-8
USB Audio Streaming : U-DAC8
4. Really impressive Blackbird DAC
http://www.diyaudio.com/forums/grou...kbird-multi-channel-es9018-dac-group-buy.html
5. 8Ch option from ACKO DAC
https://sites.google.com/site/ackodac/home
6. Motu USB interface (Appears excellent)
MOTU.com - Overview
My main concern has been cost of a powerfull HTPC + one of the above compared to an old laptop coupled with a well integrated dedicated DSP like Najda.
But najda's new board is delayed and no word on when its expected. Maybe i should go back to the PC option.
How does a modern PC compare to Najda DSP in terms of processing power for loudspeaker management?
@jojip
I tested both solutions with same audible results. Today any PC has enough power to run a 4-way stereo of any kind.
Finally, I choose to keep DSP separated of the PC after experimenting awful crackling when going back from PC sleep mode. Also on one occasion a Windows or Adobe patch messed up soundcard driver or jRiver filter stack and I have to rebuild it from scratch. There is to much trouble having DSP run by PC and keeping it up to date or trying to keep it frozen on a specific patch level.
I tested both solutions with same audible results. Today any PC has enough power to run a 4-way stereo of any kind.
Finally, I choose to keep DSP separated of the PC after experimenting awful crackling when going back from PC sleep mode. Also on one occasion a Windows or Adobe patch messed up soundcard driver or jRiver filter stack and I have to rebuild it from scratch. There is to much trouble having DSP run by PC and keeping it up to date or trying to keep it frozen on a specific patch level.
A modern PC does probably not compare to Najda in processing power - a PC is far more powerfull. I personally run 8 channels FIR with each 65536 taps on a 4 year old laptop. CPU usage is less than 5% even for these quite long filters.
JRiver is very stable and convolution works excellently. The only problem is Windows updates, which can be annoying, but it never gives real troubles. JRivers WDM drives enables using the FIR filters seamlessly for playback of netradio and other PC programs.
The MOTU 1248 uses the ESS Sabre 9016 chip, which is a little brother to ESS 9018S. It sounds very good. Even though I am completely satisfied with the MOTU 1248, I am building an ESS Sabre 9018S DAC from scratch together with a friend in the Copenhagen HIFI Society. The urge to press limits is never ending
Hi Dorin,
You are severely underestimating the power of contemporary PC.
The 2in – 8out configuration you mention, is far surpassed by 8in – 16out configuration I use in this system: http://www.bodziosoftware.com.au/Signal_Routing_New.jpg
I built several other systems before (Bodzio Software ) and this system about a year ago and it has been working seamlessly from day one.
Each component has it functional destination.
1. OPPO 105D provides HDMI switching from 3 external sources and three USB 3.0 sources, and it extracts 7.1 audio from each HDMI/USB stream. Obviously, it plays BluRays and up-scales to 4k. It also removes the top and bottom black bars from 3D movies.
2. HTPC is a DSP processor + WAV-Player. It does 8in – 16out channel processing, including filtering, HBT equalization and linear-phase for all channels. WAV Player plays 2.0 24/96kHz and 5.1 24/96kHz wave files.
3. AES/EBU amplifiers are modules from miniDSP. ICE-PWR modules can be configured to various power levels and output configurations. You will need a special software version from miniDSP that by-passes on-board DSP.
HTPC needs to be configured for audio. You can google internet for instructions how to do this. My HTPC is a free-standing device (no internet connection) and has no virus programs on it. It works as solid and reliable as if it was a “hardware device”.
Best Regards,
Bohdan
Hi Bohdan,
Sorry that I didn't make myself too clear, but I do not intend to underestimate the power of the PC, just complaint of its SW "volatility" and HW kind of instability. Or maybe I cannot afort using a perfectly silent HTPC "frozen" and isolated just for DSP duty. Instead, I prefer having a stable and dedicated DSP unit for audio duties and using my shiny HTPC in any way I want, without fear of burning my tweeters. Other than these, the audible result will be the same in both cases.
Regards,
Dorin
Sorry that I didn't make myself too clear, but I do not intend to underestimate the power of the PC, just complaint of its SW "volatility" and HW kind of instability. Or maybe I cannot afort using a perfectly silent HTPC "frozen" and isolated just for DSP duty. Instead, I prefer having a stable and dedicated DSP unit for audio duties and using my shiny HTPC in any way I want, without fear of burning my tweeters. Other than these, the audible result will be the same in both cases.
Regards,
Dorin
Hi Dorin,
No, it will not be the same.
1. There is no DSP board, that runs 8in - 16out configuration with 16384 / 32768-long impulse responses for each channel. So, no surround sound processing of 16 output channels.
2. There is no other software that runs HBT equalization for each driver in the system.
Best Regards,
Bohdan
Hi Thierry,
That bug was corrected in rePhase 1.0.0.
From the release note:
Code:
- bug correction in Minimum-Phase Filters tab: the polarity of low-pass
Linkwitz-Riley filters of order (2*n+1)*2 was reversed
(eg 12dB/oct, 36dB/oct, 60dB/oct, 84dB/oct, etc.)
A warning will be emitted when loading correction files from prior
versions using an odd number of such filters, as the polarity will now
be correct and reversed compared to the prior bogus correction.There is an explicit warning box when you load into rephase 1.0.0 an older correction file where this bug is detected.
We are on the same page on this.
I can see a perfectly tuned PC being perfectly stable (ie never updated
) and used only for filtering form its digital input. After all this is what trinnov does.On the other hand using a PC as a media player (local and web) and filter in the same time is too much a risk IMO.
It is tru that today's hardware solution still lack taps, but is is just a matter of time...
(and in the meantime I am perfectly happy with several openDRC units, and addtional IIR filtering when needed...)
Hi Pos,
First of all thank you for your great work on this!
Until now I have always used my own matlab toolbox to do these things, but your program has a much better UI and a better iterative engine, so I will try this now.
I have however one thing I don't understand;
- when I spec a minimum phase LR24 highpass filter at 1kHz, with a centering of 0ms (because I don't want added latency which is not nessecary with a minimum phase filter) and long enough (16384 at 44100sr) taps, The rolloff on the low frequencies stops at ca. -46dB and doesn't drop any further.
It doesn't matter what window I use, it always has approx. the same results.
This I find strange; if no windowing (rectangular) is used and the taps are long enough it should give a "perfect"result, at least it does with my matlab tools.
If I change the centering to 0.5ms, the results track well to -127dB, but again this shouldn't be nessecary with a minimum phase filter
Do you know what is going on here? Is there some "hidden window" applied on the first couple of ms?
many thanks for your reply.
Kees
First of all thank you for your great work on this!
Until now I have always used my own matlab toolbox to do these things, but your program has a much better UI and a better iterative engine, so I will try this now.
I have however one thing I don't understand;
- when I spec a minimum phase LR24 highpass filter at 1kHz, with a centering of 0ms (because I don't want added latency which is not nessecary with a minimum phase filter) and long enough (16384 at 44100sr) taps, The rolloff on the low frequencies stops at ca. -46dB and doesn't drop any further.
It doesn't matter what window I use, it always has approx. the same results.
This I find strange; if no windowing (rectangular) is used and the taps are long enough it should give a "perfect"result, at least it does with my matlab tools.
If I change the centering to 0.5ms, the results track well to -127dB, but again this shouldn't be nessecary with a minimum phase filter
Do you know what is going on here? Is there some "hidden window" applied on the first couple of ms?
many thanks for your reply.
Kees
Last edited:
Hi Kees
There is indeed a tiny bit of ringing just before the impulse peak, probably due to some rounding errors in the 32bits FFT engine (the impulse is obtained from an iFFT of the magnitude and phase response). Centering at 0 samples will chop it and rise the noise floor.
I might try to use a 64bits FFT engine and see if that improve things.
In practice I don't think this is a problem as adding a few samples of delay will solve any issue, and the ringing is so short and low that it does not really matter. But I agree that this would be nice if there was no pre ringing at all.