A how to for a PC XO.

[BlackCatSound]

Linear phase filters mean that there is no time lag due to the filter.

But doesn't a linear phase FIR filter have a constant phase delay? The delay is non-zero and the phase angle changes with frequency but it changes in a linear way, no lumps or bumps.

 

[ewildgoose]

Well, I was trying to give a laymans definition, but this is certainly true

Strictly speaking I think we are after zero phase filters here, which are a subset of linear phase filters. People seem to use zero phase and linear phase interchangeably, but of course this is not quite correct

Ed W

 

[ShinOBIWAN]

Shin, no doubt the Fireface is incrementally better spec'd than the 9632, but you need to compare both in a listening test, the specs are not the whole story. I doubt there is so much difference in practice and the fireface is basically their 8 way DAC with a firewire input I think?

I agree Ed.

I have however compared the Lynx/Apogee to the 9632 and then the fireface to the Lynx/Apogee. So I've got points of reference for performance. My thoughts on the 9632 vs. Lynx/Apogee were posted earlier in this thread.

I was looking at the 8-way ADC/DAC when I bought the Apogee. Its DAC's are spec'd somewhat similar to the 9632 with 112dBA SNR etc., although its ADC section is superb with 117dBA SNR. Seems to me that the RME ADI-8 is around 5 years old and getting on now. Newer DAC's like the AKM's top of the line AK4396 in the fireface have since appeared with 119dBA SNR etc.

Bottom line is: There's a very definite improvement from RME9632 -> Lynx/Apogee but from that to the fireface there's little difference between them, though I'd give the edge to the Lynx/Apogee setup with a slighty more focused sound.

 

[ShinOBIWAN]

double post

 

[geofstro]

Ed,

From what you are saying it would be definately worth my while employing a linear phase filter(s) between my mains and subwoofer.

Shin,

If I buy the Fireface I will be able to use console and Waves or the free LP equalizer to do this, right?

The only part I didn`t understand is the RTA (real time analysis) part.

Does this involve employing a mic and some additional software to measure the result of the crossover filter(s) you`ve created? If so, could you please explain the software required and the exact procedure?

Thanks

geoff

 

[dwk123]

Geoff,

Console won't work with your EMU card or at least it won't do what you need. The reason for this is that you have no internal loopback function on this card.

Whoa! I don't believe this is correct. The EMU DOES have multi-client drivers that should work with console, as long as the playback client uses the waveout device. ie the waveout I/O is simply another strip in PatchMix , and can be routed to an ASIO input or to a physical output. You can't have multiple ASIO clients, but I believe that is a somewhat common limitation.
There are some limitations in how they implemented multi-channel waveout drivers but for simple stereo it should work fine.

 

[ShinOBIWAN]

Whoa! I don't believe this is correct. The EMU DOES have multi-client drivers that should work with console, as long as the playback client uses the waveout device. ie the waveout I/O is simply another strip in PatchMix , and can be routed to an ASIO input or to a physical output. You can't have multiple ASIO clients, but I believe that is a somewhat common limitation.
There are some limitations in how they implemented multi-channel waveout drivers but for simple stereo it should work fine.

Very sorry if I gave the wrong impression. I was going with another user reporting that they couldn't get an EMU card to do what is required with console for DRC/XO.

The best thing to do Geoff is to download the trial of Console and try routing audio from foobar, winamp etc. through it and see if it will actually pick it up by using the level meters within console. If it works your all set to go, if not then my original comments stand.

Can someone please confirm if the EMU cards work and if so I can add them into the list of compatibles.

 

[ShinOBIWAN]

Ed,

From what you are saying it would be definately worth my while employing a linear phase filter(s) between my mains and subwoofer.

Shin,

If I buy the Fireface I will be able to use console and Waves or the free LP equalizer to do this, right?

The only part I didn`t understand is the RTA (real time analysis) part.

Does this involve employing a mic and some additional software to measure the result of the crossover filter(s) you`ve created? If so, could you please explain the software required and the exact procedure?

Thanks

geoff

Are you referring to generating DRC filters or analysing your loudspeaker response for XO setup?

 

[dwk123]

Very sorry if I gave the wrong impression. I was going with another user reporting that they couldn't get an EMU card to do what is required with console for DRC/XO.

Well, the Emu PatchMix app isn't the most intuitive thing in the world, so it's possible there was simple confusion. It's also possible that 'simple' multi-client functionality isn't quite enough for Console.

The best thing to do Geoff is to download the trial of Console and try routing audio from foobar, winamp etc. through it and see if it will actually pick it up by using the level meters within console. If it works your all set to go, if not then my original comments stand.

Can someone please confirm if the EMU cards work and if so I can add them into the list of compatibles.

I've been meaning to do this, and will try to do so soon-ish. My multichannel/xo/drc efforts stalled a while back when I unexpectedly ended up with a very good set of off-the-shelf speakers and spent some time getting a 'normal' 2-channel system going. I have a new target for my DIY 3/4-ways, though, so I'll try to find some time to explore the capabilities of the Emu card (I have an 1820M, but all the cards use the same drivers)

 

[ShinOBIWAN]

Thanks dwk,

If you get it to work, I'll add it to the list and that will certainly open up another price point for others interested.

 

[dwk123]

Well, a quick sanity check indicates that it works ok. I d'ld the demo console to try and set it up as follows:

- add waveout strip to PatchMix
- add 'send' in waveout strip to send to asio in 1/2
- set foobar up to use the waveout device
- in Console, patch the input straight to the output with no plugins

Result is sound output as expected on the ASIO outputs.

Now, the only thing I'm not sure of is whether foobar via waveout will be bit-perfect, or whether Windows will get in the way.

 

[geofstro]

dwk123 and Shin,

I really appreciate your help on this. And Shin, no need to feel bad about thinking EMU wouldn`t work. I second what dwk123 says about Patchmix being not exactly the most intuitive app out there. At least that`s true at first. Once you get the hang of it, the design can actually start to make sense.

Shin,

I`m concentrating soley on creating the XO filters at the moment. In an earlier post you mentioned something about RTA and I though it was related to XO`s. Do you simply listen to the XO filters to determine if they are to your liking, or is there a more technical step I need to take?

I`m in Switzerland on a Ski trip at the moment. The conditions have been absolutely fantastic today. Currently struggling with a Swiss keyboard. I`m continuing to watch this space though.

Thanks again!

Geoff

 

[ShinOBIWAN]

For the XO thing to work at its best, yes you'll need some form of acoustic measurement software. The depth and accuracy of these measurements is roughly equal to the depth of your pocket though

A really nice measurement system would be something like an Earthworks, B&K or GRAS measurement microphone together with a decent linear mic pre-amp and WinMLS level 6. For that lot though, its silly money for a single loudspeaker setup. I your doing it for profit or many designs I'd recommend it though. Also bear in mind that you need some thing like anechoic conditions for accurate measurements, for most it means going outside in the middle of a large back garden or preferably in the middle of a school field. Buy a long power extension cable too

A cheaper but still reasonable system would be a Behringer ECM8000 mic with one of the cheap Behringer mic-pre's and finally something free like speaker workshop. It isn't as slick or as accurate as the above but it gets the job done and is similar to what I use for now.

A step by step guide though? I'm not sure I could devote that much time for such a thing. Creating a crossover isn't something that's easily taught or at least creating a decent crossover isn't.

 

[geofstro]

Thanks again Shin,

I already have that Behringer Mic and Pre-amp, which I've used when doing DRC using the DRC program.

Any pointers to info on how to create a high quality crossover will be much appreciated.

Where did you learn to do it? Is it connected with your profession?

BTW is the Mic pre-amp still necessary with the RME Fireface?

Geoff

 

[ShinOBIWAN]

Any pointers to info on how to create a high quality crossover will be much appreciated.

Hi again Geoff,

The building blocks are the important thing and they've already been outlined in this thread and over on the DRC wikipedia. The rest comes from measuring, evaluating, implementing, listening and re-measuring. Keep going through those cycles until your happy. The best advice is to get stuck into it and then return here with the results. If you measure the various linear systems(phase, magnitude, impedance, distortion, groupdelay etc.) of the loudspeaker and take screen grabs of the results folks will be able to advise on changes to make to improve those measurements. It should be noted that the last 10-20% of the sound has to tweaked out by ear and preference so nobody can help you there but if your loudspeaker measures well in the beginning you can make these subtle changes with a good reference point to fall back on and evaluate the differences. Another benefit of digital is that you can compare in seconds two different crossover scheme's and instantly hear the differences.

Where did you learn to do it? Is it connected with your profession?

Just through experimenting, reading and then trial and error. Its the long way around and if you've got somebody that already knows all the tricks then your much better prepared. Its easy to scratch the surface, harder to really understand and master but given time and dedication anything is possible.

BTW is the Mic pre-amp still necessary with the RME Fireface?

No the Fireface has better quality mic-pre's than the Behringer ones. Sort of a freebie of sorts when you buy the Fireface.

 

[Builder Brad]

Hello Shinobiwan!

Great posts!

I have read your thread on Active Room Correction with interest!

I am currently using a 4 way active system consisting of a pair of orions and a pair of Thor subs. I am also running 1 x AKSA25N+ 1 x AKSa55 and 2 x AKSA100s with N+ parts for LF duty. I am using the AKSA GK 1R as a pre-amp and also a Red Wine Audio modded Squeezebox2!

from what I have read about PC XO and active room correction I am now thinking that this will allow me to really realise the full potential of my current equipment.

This is quite exciting as my system sounds awesome already!

I just have a cuple of quick questions

1)It will be really difficult for me to set my system up outside, is there an alternative option?

2)The Linkwitz Orions have already been equalised using the Orion ASP circuit board, can I easily replicate all of the filter blocks in order to create the Orion ASP on the PC? This would be a great starting point for me. I remember reading in a number of threads that digital filters strugle to create the gain required for some of the drivers in the Orion.

3)I have been using the Soundcheck 2 Studio test CD to measure my systems room response details of the Cd can be found here: http://www.interstudio.co.uk/sndchk.htm is this test CD a good source for the sweep tones ect. that are required for active room correction?

4)I am currently using a modded Squeezebox2 as my main source for music and have been very surprised by the performance of this "cheap" piece of kit. I am even using the DAC in the Squeezebox as it just sound so good. what type of performance can I expect from the DAC in the RME soundcard?

Cheers

I have had a quick read of your posts in the AC re.AKSAs and would be happy to let you listen to my 100s vs the Ucd digital amps, if it would be helpfull b4 you go ahead and build ur full system

Brad

 

[ShinOBIWAN]

Hello Shinobiwan!

Great posts!

I have read your thread on Active Room Correction with interest!

I am currently using a 4 way active system consisting of a pair of orions and a pair of Thor subs. I am also running 1 x AKSA25N+ 1 x AKSa55 and 2 x AKSA100s with N+ parts for LF duty. I am using the AKSA GK 1R as a pre-amp and also a Red Wine Audio modded Squeezebox2!

Excellent setup Brad, I particularly like your choice of amps

This is quite exciting as my system sounds awesome already!

DRC + an already awesome setup will really knock you back. You'll be walking around slobbering at how great you system sounds for months, I did Even the super expensive designs that I've heard at hifi shows sound so-so compared to what I hear now.

1)It will be really difficult for me to set my system up outside, is there an alternative option?

There's a measurement method called 'gated' which allows you to take measurements and mainly remove the room from the equation althought it isn't perfect. Basically what you do is take measurement just a couple of cm's from each driver in isolation so as to avoid reflections from room boudaries. You've got to do this for each driver (treble,mid,bass for 3-way) and then these responses are 'summed' to give you a total frequency response without large room interactions. Like I said before, its not perfect but it does work better than in-room far field measurements. The downside is that you need to test each drive unit individually and then sum the responses afterwards so your looking at 3x more work than if you did this in semi-anechoic conditions.

2)The Linkwitz Orions have already been equalised using the Orion ASP circuit board, can I easily replicate all of the filter blocks in order to create the Orion ASP on the PC? This would be a great starting point for me. I remember reading in a number of threads that digital filters strugle to create the gain required for some of the drivers in the Orion.

Piece of cake this part. If you know the exact transfer functions even better. If you email SL and ask him the exact nature of the EQ and filter slopes these can easily be translated into equivalent digital ones in the PC. I highly suspect you'll still need to tweak the sound out though as were using FIR filters rather than IIR, although you could simply use an IIR filter in the digital domain and include DRC. It will be interesting to see how DRC copes with a dipole as well. I haven't seen it done before but the principles are the same ie. correct amplitude and time domain anomolies.

3)I have been using the Soundcheck 2 Studio test CD to measure my systems room response details of the Cd can be found here: http://www.interstudio.co.uk/sndchk.htm is this test CD a good source for the sweep tones ect. that are required for active room correction?

No pretty much useless for DRC. The best way would be to use... actually my brain isn't in gear today. There's an excellent guide floating around the web by Rush Jones which details how to create the neccessary stimulus to capture an impulse response of your room. It involves CoolEdit and the Aurora plugins. Well worth a look and its easy to understand and implement. Another and quicker method would be Ed Wildgooses 'Simple Impulse Response' program available from the DRC wiki on wikipedia. I've only tried the Rush Jones method but Ed's seems even quicker if it works with your soundcard.

4)I am currently using a modded Squeezebox2 as my main source for music and have been very surprised by the performance of this "cheap" piece of kit. I am even using the DAC in the Squeezebox as it just sound so good. what type of performance can I expect from the DAC in the RME soundcard?

Its hard to quantify in words but I put it up against £3500 worth of Lynx/Apogee equipment and they sounded very similar. All in all I'd say the DAC's in the squeezebox are standard fare looking at the specs on their website. You'd have to hear whatever you were going to replace it with before making a critical judgement.

I have had a quick read of your posts in the AC re.AKSAs and would be happy to let you listen to my 100s vs the Ucd digital amps, if it would be helpfull b4 you go ahead and build ur full system

Thanks Brad, are you in Kent? I'm in Chesterfield, Derbyshire so its a little far for me to travel but I'll keep the offer in mind. The favour is extended likewise if you fancy hearing and seeing the digital stuff in action.

 

[Builder Brad]

Thank you Shinobiwan,

I don’t like the idea of the “Gated” test so I guess that I will have to get those speakers outside at some point!

Funnily enough SL refers to the response of various drivers in the Orion when tested outside and he implements various filters in the ASP to deal with the known predicted room response ie. The LF signal reflected by the floor and the 6db roll off of each dipole driver below resonance are corrected. The time and effort invested in creating the Orion and its ASP has given us a speaker which already deals with some of the unwanted in room responses that DRC also addresses and the fact that dipole speakers excite less room nodes than conventional speakers could mean that there will be less of an improvement that say with conventional box speakers.

Here is a link to the Orion Asp filter specs/ slopes and functions;

http://www.linkwitzlab.com/orion_asp.htm

Note the cautions from SL re. duplicating the functions of the ASP

Could this be easily repeated with the system you are proposing?

I may sound a bit negative, but I am actually buying into this......!

Brad

 

[ewildgoose]

I have some slightly different thoughts to Shinobiwan, but broadly agree. See inline:

from what I have read about PC XO and active room correction I am now thinking that this will allow me to really realise the full potential of my current equipment.

The room is DEFINITELY the limiting factor in anyones system. This is why you have been less than impressed at all those demos of commercial stuff at shows. I'm sure they are actually decent systems, but they never sound good in a small hotel room

1)It will be really difficult for me to set my system up outside, is there an alternative option?

I don't think most people do this. At least not until a fairly serious level.

More to the point though, DRC is about equalising the system *in-room*. You *want* to see what all the room reflections do to the system. Getting the anechoic measurement is great if you want to know how your system performs in theory (which is somewhat useful actually).

So measure in room and at the seating position for use with DRC.

2)The Linkwitz Orions have already been equalised using the Orion ASP circuit board, can I easily replicate all of the filter blocks in order to create the Orion ASP on the PC? This would be a great starting point for me. I remember reading in a number of threads that digital filters strugle to create the gain required for some of the drivers in the Orion.

Yes, keep the analogue equalisation in place. There is no sense in wasting a ton of DAC resolution just to compensate for driver rolloff.

The digital crossover advantages are significant, but not universal. Neither is it necessary for a digital active system to get world class performance. Basically, active offers:

a) easy setup of arbitrary filters
b) repeatability (try and get several large inductors of the same and correct value)
c) size of components is smaller (often)

You are already doing "active". On top of that "digital" offers:

- New crossover types which can compensate for phase distortion introduced as a result of the crossover.

Basically all normal crossovers introduce a small delay at the crossover freq (which is not necessarily detectable!). With digital crossovers we can eliminate this distortion (but at the cost of another kind of distortion, which may also be detectable...)

3)I have been using the Soundcheck 2 Studio test CD to measure my systems room response details of the Cd can be found here: http://www.interstudio.co.uk/sndchk.htm is this test CD a good source for the sweep tones ect. that are required for active room correction?

No need. For serious analysis you want an "impulse response". See my wiki for more info on what this is, but basically you can put all the info about your entire room and speakers into what looks like a burst of static. Hard to get your head around but the maths work out.

I have some software on my wiki site which should work under windows, linux and mac and makes it quite easy to measure this impulse response. See http://www.duffroomcorrection.com

4)I am currently using a modded Squeezebox2 as my main source for music and have been very surprised by the performance of this "cheap" piece of kit. I am even using the DAC in the Squeezebox as it just sound so good. what type of performance can I expect from the DAC in the RME soundcard?

I seriously doubt the DACs are really the limiting part of anyones system. I haven't yet tested, but I am prepared to put up DRC using the the really nasty built in soundcard on my PC up against a non-eq squeeze or RME card... Perhaps I am wrong, but whilst I think the DAC is a link in the chain I just don't see it being the weakest link. Don't worry about it (yet) to be honest

I have had a quick read of your posts in the AC re.AKSAs and would be happy to let you listen to my 100s vs the Ucd digital amps, if it would be helpfull b4 you go ahead and build ur full system

I have 2 UCD amps and 2 Zappulse amps. I would be very interested to run your AKSA against either or both of these. The digital switching amps have a very strong clean bass and mid-range which is way better than my benchmark Meridian 557 power amp. However, the treble is not so superior (which not to say its bad), but I wonder if a top end traditional amp or the AKSA might not suit the treble end of a really high end system better...?

Lest this be damning with faint praise the UCD and Zappulse are full time in my system and the 557 is no longer used. For overall imaging and presence and clean sound I have heard nothing better (at all). However, my first switching amp had a too small powersupply and it acquired a very hard treble and since then I have been a little sensitive to improvements in the treble end (and these seem to have come as I built steadily better and better powersupplies for the amps)



Now, coming back to the question asked before as to how you design a crossover. First get a book like Vance Dickason's Speaker Design Cookbook. From this we learn that clean impulse response and smooth off axis response are what makes a really good speaker. ie it's not enough to just have direct sound from the speaker to the listener - you also need to think about what gets radiated sideways and bounces off the sidewalls.

OK, so 101 time. Assuming you hear the direct sound plus the sidewall echos (etc), then it's clearly important to have really smooth off axis performance. But now check the response plots of that nice driver you were looking at. See the off axis performance... eughh. VERY few pistonic speakers have a really nice off-axis response (the Orion is pretty decent I think?) Bass is omni directional and treble is highly directional and so the off axis performance is not as smooth as the direct sound (often with horribly peaked response at some frequency as the comb filtering from treble and mid combine)

That's the worst of it. However, you also have to factor in the linear operating mode of each driver. Little drivers can't repro bass, and large drivers have problems reproducing treble.

So:

- Combining two drivers causes comb filtering at the crossover frequency (so we want steep crossovers)
- Steep crossovers can cause a large phase distortion (time lag) at the crossover frequency. This can be somewhat fixed using digital crossovers under some circumstances (linear phase)
- Shallow crossovers sound great but may allow the driver to enter into a range where it's not optimal, increasing distortion

So it's a tradeoff, but your Orion uses carefully optimised 4th order crossovers called Linkwitz-Riley which the designer things cause inadible phase distortion. You *might* improve the speakers by using steeper linear phase crossovers which would keep the crossover interval much smaller.

So in conclusion look to DRC to improve your speakers in room. Also look to the PC crossovers to give you a chance to play with some steeper crossovers than the ones you use at the moment (but be prepared to decide it's not a good enough advantage to be worth it).

The author of the DRC software believes that his correction is sufficiently good that it can correct for the problems in the LR crossovers you use, meaning that he doesn't think it's worth playing with digital crossovers... So DRC alone should be a high bar to beat..

Good luck

Ed W

 

[ewildgoose]

I don’t like the idea of the “Gated” test so I guess that I will have to get those speakers outside at some point!

Gating is quite easy and measurement software does it for you. DRC uses this quite extensively as well

I don't think that you should carry your speakers outside until you understand what it is that you are measuring. Ultimately it's the *in room* response which needs fixing...

Funnily enough SL refers to the response of various drivers in the Orion when tested outside and he implements various filters in the ASP to deal with the known predicted room response ie. The LF signal reflected by the floor and the 6db roll off of each dipole driver below resonance are corrected. The time and effort invested in creating the Orion and its ASP has given us a speaker which already deals with some of the unwanted in room responses that DRC also addresses and the fact that dipole speakers excite less room nodes than conventional speakers could mean that there will be less of an improvement that say with conventional box speakers.

He know what he is talking about. Basically try and keep the digital correction to correct small things, or the other stuff that it's good at like linear phase crossovers. The Orion is already a very good design

DRC works at a whole deeper level and actually tries to generate a certain amount of "cancelation" against the "echos", rather than just reducing the level of the initial output to compensate for the same. This is what gives it a huge extra expansive stereo image. However, no one is quite sure how it will work with open baffle speakers to be honest... It may need quite a bit of tweaking to work well. It's easy to experiment though - like I said, start with the nasty internal soundcard on your laptop - I doubt it will be so bad that it's not a good test bed!