A how to for a PC XO. - diyAudio
Go Back   Home > Forums > Loudspeakers > Multi-Way

Multi-Way Conventional loudspeakers with crossovers

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 24th August 2005, 01:56 PM   #1
diyAudio Member
 
ShinOBIWAN's Avatar
 
Join Date: Feb 2004
Location: UK
Default A how to for a PC XO.

I see that there is a little interest in using a PC as a fully fledged and highly configurable crossover so I thought I'd make a thread to save repeating information that is already scattered throughout various threads on here.

Don't mistake this method with the inferior Foobar/Winamp method. This route allows you to connect external sources to the PC, watch video and DVD's, play games and so forth. It is in essence a self contained XO box which will filter any audio signal be it analogue or digital. Its real benefits lie with digital audio though and users of Vinyl and Tape would be better served I suspect by sticking to more traditional methods.

It functions just like any other standalone XO such as the DEQX, DCX etc. Plug in what you want filtered and it does the rest.

First off I'll highlight the disadvantages and then you can make a decision as to whether is actually worth reading the rest of this thread.

Right here's the cons:
  • NOISE: Your using a PC so background noise will be an issue in the form of fans. You can of course spend money and make the PC silent by adding things like fanless PSU's, large 120mm fans of the highest quality running at low RPM. Passive cooling on the GFX and CPU. Don't whatever you do start to use very high power GFX cards - they are like radiators and very noisy.
    Another option is water cooling and is very viable with products such as the Zalman Reserator.
  • LATENCY: Because of the nature of the inner workings and sheer number of computation required to pull off FIR filters and EQ your going to be looking at around 100ms of delay or 1/10th of a second. This is no problem whatsoever with just audio but if you have video concerns then you'll clearly need a lip sync delay to avoid this driving you crazy. Players such as the fantastic Theatertek 2 offer this as well as some of the best onboard decoding of DTS and DD.
    I'd highly recommend you build a dedicated PC JUST to do the XO duties and nothing else if you plan to do things like multichannel and gaming otherwise you'll be experiencing drop outs in audio because the PC can't handle both video and audio - I have an Athlon 4000+ and it was brought to its knee's trying to do this so I went with a dedicated box alongside the main PC.
  • COST: This isn't cheap. The main cost is software and the soundcard(s). The actually PC will be a fraction of what these costs. Because I can only comment on the success or viability of my setup I'll list that as the configuration. Other choices of soundcard may work or you may just have wasted money - be careful! To help you out I'll list the features that you should look for in a soundcard that are essential for all this to work.
  • Technical background: The learning curve is fairly steep at first and the most confusing part is configuring the entire thing to actually do what you need. But spend a few days with it and you'll be using it just as simply as you would a volume control on your pre-amp.

If your still reading then the pluses are going to really interest you:
  • FIR filters: These are linear phase filters which means zero phase distortion. The traditional minimum phase IIR cannot claim any such thing - it distorts the signal.
    You've got the ability to create truely symetrical roll off around the XO points for each driver and at a specified acoustic rolloff rather than electrical which could mean anything after the driver has had its way with the signal.
  • Digital Room Correction: The biggest advantage of the PC XO route is DRC. Simply put with the methods and software I describe you'll be able to use a microphone connected to the PC to measure the room response and then the software calculates an inverse of this response which is applied to the signal and when this modified signal reaches the speakers the results are quite dramatic. Before you did the DRC your in-room response curve was all over the place with peaks and dips because of room interactions but afterwards your looking at a truely flat curve on the listening axis. I don't need to tell you what this does for imaging, realism and coherncy.
  • Perfect Phase Control: You'll have no problem getting a minimum phase response using the digital delays.
  • Easy of Use: The whole thing is incredibly quick and easy to use once the initial technophobia has gone. You can make adjustments on the fly and measure the response of the speaker using a Real time Analyser that takes seconds to do a sweep, then you go back tweak and re-measure until your happy. Its very easy and quick to get great results and with more time and patience, getting the very best out of the drivers is more possible than any passive or analogue circuit.
  • Flexibility: I'll show later on just what can be done with the system as a whole at the end of this thread but simply put if you can imagine it then its probably possible with a little money.
  • Upgradability: Every component is upgradable and can be bent to new digital formats. DACs, Clocks, Number of Input/Outputs, Soundcard(s), Filters, EQ's, Software, Processing Power etc. etc.
  • Multichannel: One of the biggest benefits to this system is creating a coherent multichannel system be it for DTS, DD, SACD or DVD-A. You not only can phase align individual speakers but also all speakers within reference to each other, this bring improvments in coherency. Then the DRC offers tonal matching between speakers because placement affects the tonal quality of the sound regardless of matched drivers and in general offers further speaker to speaker coherency as a whole. The flexibility is quite incredible and is superior to anything I know of in a real world living room rather than a treated studio.
  • Fun Factor: With any PC based player these days you can add reverb and effects but the quality is often left wanting. You can effectively create your own ambience with studio mastering grade plugins such as Pristine Space which allows you to create stereo reverb in a highly realistic way. You can even create surround matrix from stereo sources if you should desire.
    The creative types will enjoy this more than I but if you like to play then you've got more options than you could possibly imagine.

Please don't reply to this thread just yet since I'll continue the discussion on another post. This way it doesn't seperate the information - Thanks
  Reply With Quote
Old 24th August 2005, 03:06 PM   #2
diyAudio Member
 
ShinOBIWAN's Avatar
 
Join Date: Feb 2004
Location: UK
Hardware:

Right, I'll list the most basic components required to create a stereo 3-way XO. I'll elaborate on my configuration for multichannel just after but I think this would be best to keep things from getting overly intimidating and complicated.

The PC:

As always its a case of bigger is better regarding the CPU, the more power the more plugins you can run and the lower the latency or delay of the signal.
Also remember that you really need a dedicated PC to do DVD, HiDef video, multichannel audio or gaming otherwise you'll start getting dropouts on the audio and other strange artifacts.

I use the following for the dedicated XO PC:
  • AMD Athlon 4000+ 'San Diego'
  • Asus A8N-E Motherboard with passive cooling
  • 1Gb PC3200
  • 80Gb Hard Drive
  • A basic Geforce 6200 with passive cooling
  • Windows XP

Thats really all you need, you don't need much hard drive space since all the PC does it filter audio. You also don't need any graphics power since it doesn't do anything but display the Windows desktop.
What you do need is plenty of memory, a good stable motherboard with a minimum of data transfer bottleneck potential and the quickest CPU you can afford.

Also remember that you can use any components in their, I just listed what works well for me. Keep in mind that you want it quiet or better silent so add things like quite fans, sound proofing or passive cooling and a fanless PSU.

Soundcard and accessories

I use an RME HDSP 9632 in the dedicated XO PC.

Click the image to open in full size.

Info here:
http://www.rme-audio.com/english/hdsp/hdsp9632.htm

Its a pro-level card that supports all the features we need. If your looking for something different then make sure it supports the following:
  • Internal routing of digital signals
  • Multiclient drivers that support simultaneous use of ASIO, DX and MME clients

I can guarantee the RME supports both of these but anything else would be a guess and the operation of various details will obviously change depending on the card drivers etc.

The RME offers a few features that aren't essential but are nice all the same:
[list][*] 192Khz 24bit DAC's with an SNR of 110dB[*] DSP based 40bit precision volume controls for all digital volumes means less bits lost when attenuating signals.[*] Low latency and zero CPU load technologies

One thing I am dissapointed with though is the clocking on the RME - its OK by soundcard standards in its price range but pretty rough compared to what we would ideally like. I bought an Antelope Isochrone OCX master clock to combat this problem - it works great but isn't essential in the way that the whole thing won't work without it.

I almost forgot, the RME only has 2 analogue outputs in standard configuration, clearly not enough for a stereo 3-way which requires 6 channels. Fear not though because you add in expansion boards to flesh out the number of outputs. Depending on your requirements you can have a maximum of 14 analogue outs and 14 analogue in's.

If you want to maintain 192Khz 24bit capability then use the best DAC expansion board, the 4 channel AO4S-192:

Click the image to open in full size.

Adding this will upgade you to a total of 6 outputs that are 192Khz 24bit capable and have the same 110dB SNR that the two outputs on the soundcard have. Should you require even more channels then you can add the 8 channel AEB 8 Out on top of the AO4S-192. This brings the total number of outputs to 14 and is what I use. Bear in mind though that AEB 8 Out doesn't do 24bit 192Khz and has a lower SNR. So you definitely want to take the AO4S-192 over that one to start with.

You can also add inputs in a similar fashion but using the equivalent input boards to the ones listed above. On top of this you can add things like Wordclock modules (needed if you use an external clock such as the Antelope).
  Reply With Quote
Old 24th August 2005, 04:35 PM   #3
diyAudio Member
 
ShinOBIWAN's Avatar
 
Join Date: Feb 2004
Location: UK
Software:

Here's what you need for the software side of things:

PS. Make sure your running the latest RME drivers also.

Console:

This is the very heart of the system, its a plugin client that allows you to pass audio signal routed from the RME through various plugins which process the signal and then after this the processed signal can be routed to the outputs on the RME ready to be sent onto your amplifiers and finally the speakers themselves.

It supports ASIO, DirectX and VST plugins or 90% of what's available now.

Its also easy to use with an intuative drag and drop scheme for routing signal to various plugin components, this also provides a good visual reference of what is going where. A quick pic to show it working:

Click the image to open in full size.

And a shot of it running:
Click the image to open in full size.

You can download a free 30-day trial from here as well as view the features:

http://www.console.jp/eng/


Plug-ins

This next bit will be all about the plugins you need to actually get the whole thing to behave like a crossover.


Waves LineEQ

First up is Waves Linear Phase EQ, an FIR filter. This is actually a very well respected studio grade piece of software used in mastering, so that obviously inspires confidence in the level of performance.

Click the image to open in full size.

Info here:
http://www.waves.com/content.asp?id=137

This plugin actually creates the XO slopes and can easily create any order of rolloff based on Q and combining filters. Like I said in the advantages bit at the start of this thread, you can create perfectly symetrical acoustic slopes either side of the XO point by simplying playing with the values in the plugin and measuring the resulting response with the Real Time Analyser until you get acoustic 4th order or whatever.

Its also entirely possible to add in notch filters at this stage to tame breakup resonances in metal cones or an all pass to erridicate a rising response or baffle step.

Its easy to use because the output is visual and in combination with a PC based RTA you've got the ability to measure and adjust on the fly within seconds to get your perfect response, if its not right go back into Waves LineEQ adjust and re-measure until your happy. Powerful stuff indeed!

Finally the pricing on the Waves stuff is steep and there's no demo available.

NOTE:

I'll mention this here since its related to the filters. If you don't want to use FIR filters you can use the Waves C4 plugin that offers IIR filters, its not as flexible though since you can't add all pass or notch filters easily.


Voxengo Curve EQ

This is used for the Digital Room Correction component of the system and is again an FIR linear phase based EQ.

Click the image to open in full size.

Info here:
http://www.voxengo.com/product/curveeq/

You can also download a demo from the above link but the audio periodically drops out to encourage you to buy the real thing.

Its level of required scope in the context of our application is quite limited compared to its true capabilities so much of what it offers is left untouched and turned off.

But what it does for us is very competant despite a lot of wasted features. Plug your mic into the soundcard, then run a frequency sweep from 20hz - 20Khz and watch it draw your rooms response. Hit a button in CurveEQ and that same response is turn into an inverse of its original ready to be applied to the signal.

The up shot of this is that room interactions on frequency response are virtually eliminated.

------------------------------------------------
NOTE:

DRC isn't and shouldn't be applied until after you've got sorted XO filter - don't try to fudge mistakes with Curve EQ.
------------------------------------------------

Just to be clear let me detail the process I recommend:

First off just play with the XO points and slopes, try to get something that sounds best to you. Then after your satisfied, fine tune the whole thing like so:

PS. When I mention 'measurements' below I'm refering to using a RTA in conjunction with a sinewave generator and/or Speaker Workshop for phase etc.
  1. First off - do steps 2 - 5 outside!
  2. Measure each individual driver in the finished left cabinet, get a nice flat response with the proper acoustic rolloff, or rolloffs for a bandpass, that you require at the desired XO point(s) - don't be afraid to experiment for the best results! Also add in any notch filters and all pass components at this point if needed. Basically just get the response flat with proper and matched acoustic rolloffs for each driver.
  3. The do exactly the same for right speaker but tune it in regards to the response you got for the left you already setup in step 2. This insures a matched response regardless of driver variations.
  4. The measure the left again but with all drivers working in tandem. Check the summing of the drivers at the XO point and adjust as neccessary.
  5. Do the same for the right as your did for the left in step 4 but reference the response curve of the left as a target for the right. Again this ensures matching.

Then bring the speakers inside and correctly set them up in your listening room.
  • Run a frequency sweep with both speakers running and note the variances compared to the flat response you got outside!
  • Now allow CurveEQ to do its magic by measuring the response of both speakers running together in room and calculate the resulting curve required to 'remove' the frequency irregularaties the room is creating.
  • Once that is done run a frequency sweep with a sinewave generator and marvel at how flat your response is.

The benefit of this method is that you can swap rooms or positions for the speakers and know that anechoically they are flat in response. All you need to do is run the DRC again to compensate for the room.
Another method would be get a flat response just using the XO filters but swap rooms or positions and you've got a hell of a lot of setting up to do again. This method just allows you to do the DRC which is very quick ie. <5mins.


Voxengo Sample Delay

This is a simple way of phase correction.

Click the image to open in full size.

Info Here:
http://www.voxengo.com/product/sampledelay/

Self explanatory stuff, adjust dials phase changes, simple as. The bonus is that this one free just download it from the link above.

Now there's also another method that yeilds greater control over phase by allowing frequency specific delays. But the results are virtually inaudible for my speaker project and requires lots of work and measurements with Speaker Workshop ie. very time consuming, actually very very time consuming!

There's a couple more bits I'd like to talk about regarding software such as the RTA software and sinewave generator that are essential to aid proper setup. But I've spent a good couple of hours on this already so I'll leave it for today and continue tommorow if I have time.

I'll also provide a more thorough explanation of the exact options and setting to use in the software and RME drivers for proper operation and best performance.

Feel free to ask questions in the mean time though since I don't expect that to be up for a while.
  Reply With Quote
Old 24th August 2005, 04:42 PM   #4
diyAudio Member
 
Join Date: May 2005
Location: NC
just awesome... just awesome

about this ....

can that program be used from any cards or is it RME specific??

also does it always "work" in the backgroud? meaning if I'm watchign a movie it will still have the xover and such??
  Reply With Quote
Old 24th August 2005, 05:16 PM   #5
diyAudio Member
 
ShinOBIWAN's Avatar
 
Join Date: Feb 2004
Location: UK
Quote:
Originally posted by Audiophilenoob
just awesome... just awesome


Cheers m8.

I did it mainly because you wouldn't stop asking

Quote:
about this ....

can that program be used from any cards or is it RME specific??

also does it always "work" in the backgroud? meaning if I'm watchign a movie it will still have the xover and such??
For our purposes the program can be used with any card supporting internal routing of digital signals and multiclient drivers.

Also bear in mind though noob, that you need a card that can support at least 6 analogue outs and preferably double that for future expansion, for example a 5.1 system with 3-way fronts and the rest with self contained XO's that are of no concern to the PC XO then your looking at 10 outputs needed. Or to break it down: 6 for the fronts, 2 for the surround, 1 for center and 1 for the sub.
Remember though that even though its only the fronts that have XO duties performed by the PC in this configuration you still have full control over delays and more importantly the DRC can be applied to each of those channels also.

I actually use DRC for each individual speaker and each with its own tailored DRC response for its position relative to the listening spot, rather than a generic response spread over all the speakers. Tailoring it to each speaker this way brings even more improvements.

Like I said earlier, multichannel is awesome on this setup.

Regarding your other question: Console will run in the background at all times and perform XO duties regardless of what your doing. Remember that you'll need to have light tasks for it such as 2 channel audio for it to work flawlessly if you use an all in one solution. I can't stress that enough because if you start playing DVD and especially high definition video and games you will start to hear all manner of stranger noises from the speakers that could actually potentially damage the HF drivers. Those strange noises are things like pop's, stuttering, cracking, hiss etc.

I'd say stick with the 2 channel only if you want DRC and a 3-way XO all on one PC regardkess of power - my main rig is an Athlon 4000+ and it can't handle it, which shocked me as I went in with the attitude that I'd be fine. The setup I have now with dedicated rig hasn't glitched once and why should it since its sole purpose is to process the audio.
  Reply With Quote
Old 24th August 2005, 05:28 PM   #6
diyAudio Member
 
Join Date: Jan 2002
Location: Puget Sound
Very interesting. Thanks for laying this all out.

Can you provide a diagram that shows how your sources are interfaced? Having DVDs and CDs on a computer is fine but how are you connnecting analog sources?

Also, how is CPU utilization on your XO box? What components seem to use up the most?
  Reply With Quote
Old 24th August 2005, 05:38 PM   #7
diyAudio Member
 
ShinOBIWAN's Avatar
 
Join Date: Feb 2004
Location: UK
Quote:
Originally posted by ultrachrome
Very interesting. Thanks for laying this all out.

Can you provide a diagram that shows how your sources are interfaced? Having DVDs and CDs on a computer is fine but how are you connnecting analog sources?

Also, how is CPU utilization on your XO box? What components seem to use up the most?
I'm going to draw a block diagram to show how I've got everything connected up.

I don't want to confuse people because the way I've got mine setup is quite complicated spread out over two PC's. It doesn't need to be that complicated because a lot of the complexity I have is due to extra niceties.

I can state that for analogue sources, you simply connect them to the dedicated XO's analogue in's and its all automatically processed the way I've got mine setup. For example I've got a digital TV box that's just plugged into the XO PC and when its turned on the sound is processed regardless of if the main PC is sending digital audio also - it doesn't sound good playing two sources at the same time though

Wait a few days and all will be revealed. Afterall there's still a fair amount that I haven't said, so please bear with me.

Likewise with the CPU resources needed - Console actually provides such a monitor to keep an eye on things. With the dedicated box I use around 15-20% for stereo and around 65% for multichannel using FIR filters. Again I'll post shots when I continue the rest of it.
  Reply With Quote
Old 24th August 2005, 05:41 PM   #8
BillH is offline BillH  United States
diyAudio Member
 
BillH's Avatar
 
Join Date: Mar 2005
Location: Wisconsin
ShinOBIWAN: Thanks for the information. It comes at a very good time for me. I've just started to do a PC crossover using the KX drivers and an old Soundblaster Live! 5.1 Value soundcard.

As I understand it, the KX crossovers are IIR. There was some discussion on the KX forum about FIR filters, but it seemed that it could not be implemented.

I'm running a Pentium II 350Mhz computer with 384Mb ram and getting ~15% CPU usage while playing a CD and a 3-way crossover in KX.

I'm at work, but have some questions for you later about Console.

BillH
  Reply With Quote
Old 24th August 2005, 05:42 PM   #9
diyAudio Member
 
Join Date: May 2005
Location: NC
Quote:
Originally posted by ShinOBIWAN


Cheers m8.

I did it mainly because you wouldn't stop asking



For our purposes the program can be used with any card supporting internal routing of digital signals and multiclient drivers.

Also bear in mind though noob, that you need a card that can support at least 6 analogue outs and preferably double that for future expansion, for example a 5.1 system with 3-way fronts and the rest with self contained XO's that are of no concern to the PC XO then your looking at 10 outputs needed. Or to break it down: 6 for the fronts, 2 for the surround, 1 for center and 1 for the sub.
Remember though that even though its only the fronts that have XO duties performed by the PC in this configuration you still have full control over delays and more importantly the DRC can be applied to each of those channels also.

I actually use DRC for each individual speaker and each with its own tailored DRC response for its position relative to the listening spot, rather than a generic response spread over all the speakers. Tailoring it to each speaker this way brings even more improvements.

Like I said earlier, multichannel is awesome on this setup.

Regarding your other question: Console will run in the background at all times and perform XO duties regardless of what your doing. Remember that you'll need to have light tasks for it such as 2 channel audio for it to work flawlessly if you use an all in one solution. I can't stress that enough because if you start playing DVD and especially high definition video and games you will start to hear all manner of stranger noises from the speakers that could actually potentially damage the HF drivers. Those strange noises are things like pop's, stuttering, cracking, hiss etc.

I'd say stick with the 2 channel only if you want DRC and a 3-way XO all on one PC regardkess of power - my main rig is an Athlon 4000+ and it can't handle it, which shocked me as I went in with the attitude that I'd be fine. The setup I have now with dedicated rig hasn't glitched once and why should it since its sole purpose is to process the audio.

thanks Shin

really great info

what about the MAudio 1010 .... 32 channels of outputs... external DAC and such... I just can get it slightly cheaper than the RME

either way thanks for the info
  Reply With Quote
Old 24th August 2005, 06:05 PM   #10
diyAudio Member
 
ShinOBIWAN's Avatar
 
Join Date: Feb 2004
Location: UK
Quote:
Originally posted by Audiophilenoob



thanks Shin

really great info


No probs, hope others will see the light of this type of setup since its superior to anything I've heard in my system before.


Quote:
what about the MAudio 1010 .... 32 channels of outputs... external DAC and such... I just can get it slightly cheaper than the RME

either way thanks for the info
Like I said back in the hardware bit of this thread. I've only got experience with the RME and with that in mind I can assure folks that what I have is stable, reliable and performs very well. Other cards I have never tried so I'd hate to recommend something and then have someone come back to me and say they spent $xxx amount and it doesn't work!

From a pure SQ perspective, folks in the HTPC forum on AVS have put the RME up against cards like the Delta and Lynx TWO B. From the results it was ranked Lynx top closely followed by the RME and the Delta was a fair way back from the Lynx.
Also the Delta supports 32 outputs but how many analogue outs does it have as standard? What expansion boards are available? How many of those 32 outputs can actually be used for analogue? The RME claims 32 outputs but only 16 of those can be used for analogue with a single card and as standard it comes with 2! If you need more than that with the RME you have to add a second card into the PC and link it to the first - you can add upto 3 x RME HDSP 9632's into one PC for upto 48 analogue outs if the need took you.

All these questions need answering if you choose another card and the answers largely depend upon what you expect from the system. If you want 2 channel 3-way then 6 is enough, if you want 2 channel 4-way then 8 and at the other end of the scale, if you needed enough ouputs for a 3-way 5.1 multichannel system then you'd need 16 outputs along with a dual processor dedicated PC for the sheer huge number crunching task!
Don't forget that the card also MUST have internal routing and multiclient drivers.

The only soundcard related questions I can answer with certainty are those regarding either the RME HDSP 9632 or the HDSP 9652 since I own both of these. Anything else would be a guess and at the prices these cost you wouldn't respect me for guessing!

Hopefully others will add definite working examples here also as I have done and hopefully some cheaper configurations that still allow the same core plugins and software to be used - please folks no talk of foobar or winamp stuff since that's very much restricted in flexibility and performance. What we need is people to test it on other soundcards and then report back here with the results. This will help everyone I feel.

Also Noob, you've probably noticed that the software and plugins alone come to around $2000 or more. Do you have an email address?
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



New To Site? Need Help?

All times are GMT. The time now is 12:24 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2