BI-AMPING

[thoppa]

Problem solved ?

I didn't want passive bi-amping 'cos of the phase shifts, or active 'cos of more circuits, noise, distortion (albeit small if I pay enough). I really wanted a digital filter thingamy and was thinking of getting a DSP to go in my DAC. However it seems there is a plugin for Foobar that creates a high pass and low pass channel for the left and the right channels - four channel output and bit perfect I hope. So now I need to play with this and if it works I need a new soundcard and maybe a second DAC. Not the cheapest option but at least it is an option.

So if anyone out there is looking for a digital crossover to bi-amp....the answer is Foobar.

....maybe.

Signing off,

Tom

[DrWho]

Wow, where to begin...

An engineer once told me, "start with the assumptions" when measured results don't correlate to the theory. I would propose you test your theories and lets us know your results...using measurements of course - the placebo effect would be rather annoying in this case.

The only other thing I can propose would be to find a good analog circuits textbook and familiarize yourself with the design issues instead of reinventing the wheel...

You're making a lot of statements that "aren't valid" and the only way I can see accurately addressing them would be to provide a complete nonlinear electrical model and then address its implications. I'm finishing up my 4th year in college studying electrical engineering and it's taken me over 3 years to even start thinking about tackling such models. Amplifiers are inherantly designed such that the output bandwidth matches the input bandwidth, regardless of the (complex) impedance of the attached load. It is only by reducing the input bandwidth that an amplifier might improve its performance - the only way to achieve this is with a crossover before the amplifier. In other words, an amplifier approximates an ideal voltage source. In fact, they were intentionally designed this way so that we could use any amplifier with any speaker and with any preamp.

Btw, you are going to have the exact same phase shifts regardless of the method of altering the frequency response - whether it be the natural roll off of the speakers, a digital crossover, passive crossover, etc...in fact, a crossover wouldn't work without the phase shift [] A second important point is that the other half of the crossover will acoustically sum and cancel the phase shift - so in the end you (ideally) still have the original waveform. And for what it's worth, because the phase response is going to be the same, the group delay is also going to be the same (unless you're doing some of the crazy trippy stuff that some of the engineers at Shure are doing, but it involves very long propogation delays and can only be done digitally - and as far as I know that algorithm is not public yet).

But if you can achieve a quality digital crossover in your computer with foobar, then you don't have to worry about all of this [] What soundcard are you using? I would be very surprised if it's capable of two seperate stereo digital outputs. You probably also want to consider using ASIO if you're on a windows machine so that you can bypass the very crappy internal processing that windows does (it samples everything to 48kHz with a really crappy alogorithm). You won't be able to do bit perfect when you do digital filters. You're going to lose at least 1-Bit of resolution due to rounding errors by employing a digital filter (nothing to worry about though).

[thoppa]

Hi,

I fully appreciate and agree with your comments re. bandwidth. I did a year of Physics at uni, incl. a course in electronics so I did circuit theory, Norton's Thevenin's and all that. We might not be on the same page but at least I have an idea about what book you're using.....Before work took over I used to build amps and speakers as a hobby. Sadly, I have forgotten almost all of my learning experiences and only recently had the chance to ease off work and get back to my first love ! Anyway, I don't claim to know or understand more than just the basics. It may well be my assumptions that are clouding my thinking.

I've done some more reading on the fool's thing - seems the phrase was coined by a guy trying it with tubes. The consensus seems to be that it provides only minor improvements (intermodular distortion ? not sure what this is...) so it isn't worth the cost, given themeagre benefits. Interestingly though, no-one says it won't provide some benefits and is a class above simple bi-wiring, which does next to nothing it seems.

As you know I'm sure, class-T is totally different from other forms of amp because it uses PWM, amplifies this, pumps in current, filters out the clock. This results in low output impedance, high damping , high efficiency, low distortion. So perhaps some of the theories that apply to linear and tube amps might not be valid for class-T ? I know class-T has only started to become popular but I thought someone must have tried out all the 'old theories' on the 'new technology' if you see what I mean.

I am also completely open to finding the best and simplest solution; within a budget. I am aiming to build a sytem for around US$1200, not incl the PC. So far I have spent US$540 on the Klipsch, US$250 on the amps, US$150 on the DAC (which is dead - taking it back...), $80 on the soundcard. I am not interested in ultimate SPL - what I am searching for is awesome dynamics and soundstage - which is why I bought the Klispch.

My souncard is ASIO but two channel. I have an old SBlaster Extigy which is 6-channel but it also re-samples to 48Khz and claims it is 24 bit but actually upsamples from 16. I don't want to use it but I could, just to experiment with the Foobar dsp digital crossover for bi-amping etc. Here's a link to it -

http://xover.sourceforge.net/

I read abut a Linkwitz-Riley active crossover that is 24dB/octave and, although as you say phase shifts, they are 360 degrees (or 180...I forget ..) so the soundstage won't be seriously affected.

Anyway, thanks again for keeping on at me - I'm on a learning curve....

Do you bi- or tri-amp ?

Cheers,

[thoppa]

This combination married Mcintoshs recomendations of using the transformer taps to control output levels with Klipsch's recomendations of using their crossover circuts to protect their drivers.

Why not just use the volume controls on the amplifiers and reduce the number of parts in your signal path?

The only parts in the signal path is a lenth of wire. One lenth for each section of the crossover, which attaches to the drivers.

To get +3db with the LF, 0db change with the MF, and -3db with the HF.

My implementation was to tri-wire using the 8ohm tap to the LF, 4ohm tap to the MF, and 2ohm tap to the HF.

The 8ohm tap has twice the turns of the 4ohm tap, twice the voltage, +3db.

The 2 ohm tap has half the turns of the 4 ohm tap, valf the voltage, -3db

The amp is rated at 135 watts per channel.

So what's the result ? What benefits have you experienced ? Any drawbacks ? Please tell me if you think it was worth the effort !

Most of the bi-amping stuff says I should bypass the speakers internal crossover but there is no way even on a sunny day that I'm gonna open up my new Klipsch - they haven't even settled in yet.

Incidentally, they are really loosening up. They sounded much harsher out of the box than the demo models in the shop and I was concerned about acoustics but actually my Klipsch and I are getting on very well indeed. The horn is very impressive and the 5.25" drivers exceeded my hopes. I'm made up, as they say back home.

Cheers

[DrWho]

I've read about a reduction in intermod distortion, but I've never seen any derivation for it so don't know what to believe. I know there is a reduction when using an active crossover. You also need to consider that you're halving the impedance seen by your preamp, so it's going to be delivering twice the current - so double the distortion from the preamp for the same voltage. It's probably not a big concern, but if your gains are tiny, then you're just shifting equal compromises...

Btw, class T amps are still "ideal voltage" sources. [] Sounds like you've got a rocking setup going on there. I'm still learning too and probably will be for the rest of me life.

No, I don't bi-amp, but only because I don't have the budget. I'm running the digital output from an Audigy 2 into a Denon 3300 receiver and then into an HK 630 that I'm using as an amp for my Chorus II mains. I mentioned earlier the option of going with a receiver for the DAC. I know it's buying a lot of features you don't plan on using, but the DAC and preamp stages they're using are usually the same thing as the top of the line products (it's just cheaper to use the same parts everywhere). There really aren't that many different chips available on the market either (unless I'm totally looking in the wrong catalogs). I know my 3300 was the flagship at one time and only cost $200 used. One feature you get with a good receiver is the bass management allowing you to properly integrate a subwoofer. You also have the ability to decode Dolby and DTS, which is beneficial for using the digital output from your dvd player.

One last note - if you ever get around to biamping your speakers with an active crossover, then it would be extremely beneficial to put a cap inline with the tweeters to prevent any turn on thumps from bottoming out the tweeter and destroying it.

[thoppa]

Hi,

The bi-amping efforts progress...and then get immediately stuck....

So I'm now thinking to get a DAC that has great op amps - then maybe I can add another op amp circuit of the same quality to do an active crossover; easy and not much worse for noise, distortion etc. I'll also try a first order passive line level crossover (once I find out the input impedance of the amps), and of course, a fool's bi-amp too. Thing is, I've got to wait - my speakers are still settling in and that will take a full month before they can be eliminated as a variable.

Anyway, first step - I opened them up - very well damped (with the foam egg-box-style stuff) inside for a reflex cabinet. Also well-braced. Coloured black too. Nice. Caps are custom coated with a Klipsch logo. The crossover is a second order butterworth with, I think, polyester caps. The circuits are entirely separate.

The HF goes into a 5W 11RJ resistor, then to a small cap (can't read the value because of the glue and positioning) which is about a 3uF I think. After this the signal goes on to the horn via a 68uF 100VDC cap. A 0.36mH air core inductor with, I think 18AWG wire, is wired in parallel to the horn.

The LF signal goes into a 1.4mH iron core coil, also 18AWG, then to the drives which are wired in parallel. A large cap (again, can't read) is wired in parallel. Shame about the iron core but hey ho. An aire core replacement with the same resistance would be huge and need something like 40metres of 16AWG. I don't fancy making that so the iron core will stay !

So now I'm confused - 1.4mH @ 8 ohms is 1286Hz crossover. 1800Hz, the stated frequency, would mean the drives are 11 ohms at the crossover frequency. 11 ohms for two drives in parallel ? 22 ohms each ? Hmmm, my maths must be wrong...?

I also opened up my Edirol - the controller is fine and there is a 60Mhz clock, the DAC is okay; AK4114 (24bit/192Khz, dynamic range 108dB, S/N108dB, THD+N -94dB) but then the op amps - horrors NJM2100. I'd expect this in a crappy discman or something. So now I'm looking for someone in HK that can replace them with opa2134s (I think...)

Anyway, I'm stuck until I can work out what's what with the crossover frequency and impedances - little help anyone ?

Cheers,

Tom

[thoppa]

You also need to consider that you're halving the impedance seen by your preamp, so it's going to be delivering twice the current - so double the distortion from the preamp for the same voltage.

One last note - if you ever get around to biamping your speakers with an active crossover, then it would be extremely beneficial to put a cap inline with the tweeters to prevent any turn on thumps from bottoming out the tweeter and destroying it.

Thanks for the advice - I had not considered this at all ! Maybe I should modify the output op amps in the DAC to form a Linkwitz-Riley crossover instead of adding another circuit ?

On the second point, I doubt I will disconnect the HF crossover. I'm too scared of connecting an amp straight up to a sensitive compression driver. I realise I won't get all the gains I hope for and there might be phase issues if I dsconnect the LF one but not the HF one, but there we are. Peace of mind is a factor that I cannot ignore.

Many thanks again,

Tom

[thoppa]

Problem solved ?

it seems there is a plugin for Foobar that creates a high pass and low pass channel for the left and the right channels - four channel output

Here's the foobar link again for digital bi-amping.

http://xover.sourceforge.net/

The only decent soundcard I can find for my laptop is the Creative professional E-MU 1616 US$399 or E-MU 1616M with better DACs for another US$100.

This is out of my budget so the digital bi-amp option is too sadly. Should have got a desktop....

Here are some links for the pros and cons of bi-amp with and without an active crossover. Opinion is divided and it seems that, to broadly generalise, tube-dudes maintain it's almost pointless and linear-lovers seem to think it's worthwhile. Although there are theories, no one has posted any hard data to support it seems. And no class-T anywhere.....

http://www.soundstage.com/synergize/synergize031998.htm

http://www.fluance.com/wiring.html

http://sound.westhost.com/bi-amp.htm

Man I love my Klipsch RF52 s ! I haven't done much else except work, eat and listen, listen, listen....who would have thought I'd be ust as happy to listen to Michael Jackson as Also Sprach Zarathustra as Norah Jones as ....the list is endless !

Cheers Mr Klipsch ! U R A * !

Tom

[DrWho]

So now I'm confused - 1.4mH @ 8 ohms is 1286Hz crossover. 1800Hz, the stated frequency, would mean the drives are 11 ohms at the crossover frequency. 11 ohms for two drives in parallel ? 22 ohms each ? Hmmm, my maths must be wrong...?

The impedance of the woofers aren't going to be purely resistive at every frequency (inductance from the voice coil). The raw frequency response of the driver isn't going to be perfect either, so it's important to think of the passive crossover as also providing a certain amount of EQ. The quoted crossover point is going to be the acoustical crossover, not an electrical one.

[thoppa]

Thanks - good to know it's not my maths, but it is a bit of a headache.

Here is an article on passive line level crossovers :

http://www.t-linespeakers.org/tech/filters/passiveHLxo.html

And another two on crossovers generally :

http://www.silcom.com/~aludwig/Sysdes/Crossove_Design.htm#Bi_amping

http://sound.westhost.com/lr-passive.htm

So, after much reading I've decided to keep Klipsch's RF52 crossovers and replace the iron core inductor with an air core (on order from Solen in Canada). I plan to vertical bi-amp (separate the left channels from the right channels competely with separate amps) and use individual filters for all amp channels. The amps are small so they will hide behind the speakers and have very short cables to the speakers. The combination of short cables and a much better coil should make a difference to the LF section especially.

I'll use a first order passive line level filter on the exit of the DAC and bi-wire from the DAC pre-amp output to the amps with some tasty cables. I'll do my best to match the desired op amp output to the crossover frequency and input impedance of the amps. Although the gains will be not be as good as an active filter without a high level crossover, there are no real drawbacks either, except a slight reduction in the pre-amp level into the amps. It also avoids the horrors of another set of op amps dragging down the noise and distortion performance before it gets amplified. The amps will still receive a narrower bandwidth and also have to drive a less complex load. It is also far far far simpler and I don't have to re-engineer the crossover Klipsch has designed, which would require test equipment almost as expensive as the amps it seems.

I saw someone's quote after their post was "it's all about compromise". How true.

Many thanks to Dr Who and Speakerfritz for your help and pointers. Great guys ! Thank you.

Cheers,

Tom

[thoppa]

Ah, the quote was by Dr Who....! I hope the studies go well. All the best and thanks again.

[Guest " "]

thoppa

looks like a plan.

Just curious what type of DAC?

I am on the hunt for a DAC and was having trouble picking out a nice one....so I decided to buy the cheapest one I could find to set a base line and compare to the internal DAC's of my system.  I was surprised that the cheap DAC performed better than the DAC in my sat reciever, and about the same as my computer DAC.  It did not fair better than the DAC in my media server and HT reciever.  I spent 50 bucks for the cheap DAC that I'm using in the short term.....still looking for a long term solution.

[thoppa]

Hi,

Yes, a plan...love making plans....but is it a good one ?

DACs are everywhere in Hong Kong and, as you say, choosing can be hard. I'm off today to visit vocative audio and diy kits to demo some.

The DAC in my Onkyo FR-V77 goes into its "discrete" amp stage and the two together sound very brittle compared to the DAC in the Edirol UA-1EX "Eddy" combined with the DIY Paradise "Charlize" class-T amps.

I also have an old Sony MD313 which is balanced but flat sounding.

Also have a Soundbaster Extigy USB card which is okay but I have some issues with it. I saw Dr Who is using an Audigy 2 - how is it Doc ? Unfortunately the USB in the Extigy has been very temperamental since an electrical storm a few months back so I also have a cheap CM106 USB card that I've been using to pass through AC3 into the Soundblaster since the storm.

Of all the combinations above, Eddy and Charlize are the happiest couple, despite the atrocious op amps in Eddy (given in an earlier post).

So, yeah, DACs.....hmm.....tricky. I'll post my findings on Sunday probably and in another thread ; and unless I fall in lust with one (e.g. the Musiland MD-10) I'll try them all (Zhaolu (said Jow-loo with a slight z in the J ! in Putonghua) 2.5C, Great March 2, Great March modified, Musiland MD-10 and the Lite models too maybe.))

I won't listen to the tube ones. I have never been into tubes for two reasons - they need to warm up and they have a short(ish) life. They do sound really good but simply don't appeal to me.

So how is the passive bi-amping you use with your MacIntosh ? Any noticeable improvements or problems ?

Cheers,

Tom

[DrWho]

I run the digital output from the Audigy2 into my Denon 3300 receiver, and use it as a DAC to feed the amplifier of an HK 630. The Denon has a lower noise floor, which reveals itself as sounding less harsh (most of the differences between DACs are in the noise floor). The advantage of the Audigy2 over other sound cards I've played with is the ASIO functionality, which means I can bypass the stupid kernel windows uses that resamples everything to 48kHz before sending the signal to the sound card.

I wouldn't be happy with the normal analog outputs, but the digital output has me very satisfied.

[thoppa]

Hi

Is the Audigy PCI or USB ? 6 channel ? Do you have any problems with the driver ? And are you using optical or co-ax connections ?

Seems to me you might have a v sweet setup going there ! No wonder you plug Denon eh ?

Cheers,

[DrWho]

It's the PCI version with a fancy front face plate for additional connectivity. I use the coax digital output from the front panel.

[thoppa]

Hi,

Your system is sorted eh ? No fuss....

I have set up my system with one channel "fool's" bi-amped and the other running both HF and LF through one channel. I have listened and listened and tried test tones too but I cannot hear a difference. I believe my ears. If I can't hear it, it may as well not be there right ?

So now it is time to think about the crossovers. The speakers use a 2nd order - but 1st orders are much better - so how about a passive line level crossover (for 6dB per octave and only using capacitors and resistors - no awful coils) and then another passive first order crossover at the exit of the power amp - will this give 12dB per octave overall ? Is there any way to build a speaker crossover without using coils, as there is for line level crossovers ?

Any / all help welcomed !! Thanks,

Tom

[DrWho]

1st orders are better???

[thoppa]

Apparently so - reduced phase issues, constant output over the crossover frequency, fewer components enabling better damping. Not a great difference I imagine , but better.

Tom.

[DrWho]

I don't buy it.

So things like intermodulation distortion, interaction between drivers, power compression, etc etc... all take second place to phase rotation which is classically considered inaudible? [] What about the natural roll-off of the drivers in question? (which add another pole or two to the effective acoustical response). What about the EQ built into the stock filters designed to address driver specific issues?

For what it's worth, higher-ordered filters also maintain constant output through the crossover region, but that's not necessarily the goal either...I'm of the opinion that the optimal crossover topology is going to be dependent upon a system of least compromises. An ideal crossover in crossover world isn't necessarily the best crossover in light of the rest of the speaker...(heck, even in light of a perfect speaker).