BI-AMPING

[thoppa]

Hi !

It sort of makes sense to me - all components degrade or create a loss because none are perfect, not even OFC wire.

I don't think the argument is that 1st order crossovers are always better - as you say it depends on the speaker. But if a speaker set-up is great with a 1st order then this is better than a higher one.

So I'm coming back to this idea that the last thing I want to do is use an active line level crossover because op amps colour the sound, and add noise and distortion. I can't afford one that doesn't suffer these problems.

So I'd really appreciate your opinion on this idea :

I would like to reduce the bandwidth going into the amps to improve their performance, especially the woofers, and, if possible, I would like to remove/minimise all components connecting the power amp to the woofers for better damping. So I was thinking about a passive line level 2nd order low pass filter. It's easy - two capacitors and 2 resistors matched to the frequency and amp input impedance (20Kohm). There's an insertion loss but my DAC has a very high output and my amps have high gain too. It seems to me to be the best solution for my set-up.

I will always have to have a capacitor in line with the tweeter, even if only for protection, so it might as well be part of a high pass filter. So I think it will suit my set up (and my lack of knowledge*) just to leave the HF section alone and add a slight attenuation at the pre-amp level if needs be. I don't think I'll need one because the insertion loss on the LF might be compensated for by the loss of the coil in series with the woofer.

Can you see any matching problems with this ? Is my theory cockeyed ?

Thanks,

Tom

*The high pass filter is not 'standard' - there is an attenuation resistor, 11ohms, then a capacitor 2.2uF, then a coil to earth 0.36mH, then a final series capacitor of 6.8uF. It seems to me that the 6.8uF cap is the correct value for 1800hz and the rest is a compensation and attenuation circuit, which I do not know how to analyse.

[thoppa]

Hi,

Thanks to this excellent post :

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

I have found out the tweeter uses a 3rd order 'dual' series crossover. Not my earlier assumption about compensation - I see light ahead !

Cheers,

Tom

[thoppa]

Er...scratch that...quite wrong .

The woofer uses a 2nd order crossover and the values indicate it is Linkwitz-Riley for 4ohms at 900Hz, half the stated crossover frequency. The values don't fit any other model. I have replaced this with a 2nd order passive line level crossover set at 1800Hz and matching the amps input impedance of 20Kohms and this works well it seems.

The tweeter crossover has me at a complete loss. It is 3rd order; the cap values fit Butterworth but the coil value doesn't, by a factor of 6. I have given up trying to work it out and have used another passive line level crossover similar to the above and also 1st order butterworth in line with the tweeter.

Initial impressions is that the unmodded speaker sounds very similar to the modded one so I think maybe I have got lucky with the maths. If someone reading this decides they want to mod their RF52s I would actually say don't bother unless you are into the fun of it. Positioning the speakers for the room acoustics has made more of a difference IMO.

BTW I ran looped mono test tones through the speakers and switched from left to right and back again to test output and response. As far as musical performance goes, however, it is too early to say if all this effort has proved worth it.

In fact, I have become very doubtful about the real world benefits of bi-amping. I certainly don't have ears capable of appreciating the subtle differences or pockets deep enough to really gain from it.

Cheers,

Tom

[mas]

For what is fundamentally such a simple and valid concept, I am amazed to watch the topic of bi-amping morph into a subject just as convoluted and screwy as the many perennially cable debates! And in all of it, the concept of more closely aligning the acoustic centers is all but lost. Oh well.

But it is always entertaining to watch people go to such incredible lengths to avoid doing what would be much less work to begin with to simply do it correctly!

And regarding "fool's bi-amping"...Its nice to finally see an audio technique appropriately labeled!

[thoppa]

fundamentally such a simple and valid concept

the concept of more closely aligning the acoustic centers is all but lost.

Simple concept, not so simple to do. I found out that my Klipsch RF52 have a 2nd order low pass Linkwitz-Riley for 4 ohms at 900Hz, and a 3rd order Butterworth high pass for 12 ohms at 4kHz. The stated crossover is 1.8Khz. The bi-amp has made a small difference - bass is better defined, treble more crystal. I am pleased with the results now it is done but it was a real headache and I'm not sure I'd recommend it all things considered.

I have no speaker cables as the amps connect into the internal speaker wiring (which I might upgrade one day) so I can avoid that debate !

So what is "aligning the acoustic centres" ?

Cheers

[mas]

fundamentally such a simple and valid concept

the concept of more closely aligning the acoustic centers is all but lost.

Simple concept, not so simple to do. I found out that my Klipsch RF52 have a 2nd order low pass Linkwitz-Riley for 4 ohms at 900Hz, and a 3rd order Butterworth high pass for 12 ohms at 4kHz. The stated crossover is 1.8Khz. The bi-amp has made a small difference - bass is better defined, treble more crystal. I am pleased with the results now it is done but it was a real headache and I'm not sure I'd recommend it all things considered.

I have no speaker cables as the amps connect into the internal speaker wiring (which I might upgrade one day) so I can avoid that debate !

So what is "aligning the acoustic centres" ?

Cheers

[:|]

Its Deja Vu all over again.

Without re-reading all of the posts, I believe it was Tom who provided a clear diagram (See the post in this thread of 11/28/2006, 10:21 PM # 830371 by TBrennan for a diagram of a 2-way multi-bandpass system).

Additionally, my references to 'bi'-amping are not limited only to 2-way systems. Rather I use the term as slang to refer to all multi-bandpass systems dependent upon the topology of the speaker system - be they 3 way or 4 way, etc.

Bear with my wordy description best shown graphically:

You run the output of the preamp stage into an external active crossover with as many bandpasses as are appropriate to eliminate internal passive crossovers. (In employing this configuration I question the purpose of retaining any passive crossovers.)

Depending upon the topology of the system and crossover design, the appropriate crossover topology (e.g.: LR, Bessel, Butterworth...) and slopes are chosen.

The delay for each bandpass is based upon the time offset determined by measurement of the impulse response of the system. This time offset is specified in the crossover for each bandpass. This will align the acoustic centers (the apparent 'source' reference of each 'element' in the speaker system which is not necessarily the voice coil! Heck, its not necessarily even a stationary point! But that subject is far beyond the scope of this issue.).

Proper gain structure is observed dependent upon the design of the variously employed elements.

Each bandpass output is then sent to an individual amplifier and the output of the amplifier is sent to the appropriate driver elements.

This basic general configuration is very simple and straightforward and has been the norm in pro SR and many other environments for 'a long time'.

I guess that I am not sure what you mean by this not being 'simple' to implement. All you need is a little information, an active crossover, amplification, speakers, and a few cables. It is actually very simple from my perspective.

But judging from the many fascinating(sic) descriptions and ideas that many have posited regarding what they think 'bi'-amping is, I would agree with Tom. I suspect that many are not familiar with what 'bi'-amping is. And I must admit to being a bit surprised (he said with a very quizzical look on his face) at the plethora of opinions regarding topologies. This is a well-behaved and, if I do say so, a well understood topic to undergo such a 'reinvention of the wheel'.

And except as an experiment, and not meaning to criticize as each person has their own preferences and applications- and i certainly do not desire to denigrate anyone's quest to understand various technologies and/or first principle issues, but if you are driving this with a traditional PC (as opposed to an optimal Avid or ProTools configuration), there are other areas which will yield a greater return than bi-amping, as the sources of noise, etc. in a 'normal' PC have some fundamental limitations.

[thoppa]

Hi,

Thanks for that....I guess alignment is why the bass used LR and the treble used B'worth and why they are different orders and so far apart in terms of frequency ?

The bass was easy - 2nd order LR passive line level is no problem to match to the Klipsch 2nd order LR in the speaker. What I don't know is whether a passive line level filter creates a phase change that continues on through the class-T amp - I assume it does. I just don't know for sure.

The tweeter was not so easy mainly due to a lack of information. So I have gone for a 2nd order LR passive line level and then a 1st order B'worth in line with the tweeter. I have probably messed up the alignment so that is why I say it is not easy. I also don't know the exact input impedance of the amp but it is around 20K so I used that in the calculations. Easy ? Not for me.

If phasing, alignment and input impedance were not issues then I would not hesitate to say it is as easy as making a fool of myself.

If you know how to replicate a 3rd order butterworth passive line level and 2nd order passive line level please let me know as I only have the maths for LR.

Cheers,

Tom.

[mas]

You got that from what I wrote?????

[:|]

[thoppa]

You said it was time offset and the crossover....

The delay for each bandpass is based upon the time offset determined by measurement of the impulse response of the system. This time offset is specified in the crossover for each bandpass. This will align the acoustic centers

....so from this I think an impulse test is used to measure the time offset which in turn is used to configure the crossover so that the acoustic centers are aligned. So alignment is, I guess, about using the crossover to align the acoustic centers. That's why I talked about the crossover in my speakers...and why it seems Klipsch have mixed orders and designs in the one crossover.

Obviously from your response I have misunderstood. Would you explain ? Here is how I worked out my RF52 crossover.

[thoppa]

So now I've got the speakers working musically and tonally, I'd love to get them sorted for their 'alignment' as, I guess, this will improve the staging and make the position of the listener less critical. Am I wrong about this too ?

Here is a link to the site that helped me calculate the PLLXO.

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

It says the values it gives are close to Linkwitz-Riley but I have no idea if they are, or how much different Butterworth would be...

Info is what I really need....

Cheers,

Tom

[mas]

In what I said, their is no employment of "fools" bi-amping. Nor are there any passive crossovers.

You cannot introduce the type of time delay I refer to in a passive crossover. And the impulse response and ETC measurement to which I refer are time domain measurements performed acoustically with, say, a TEF, EASERA, or any of a number of acoustical time domain measurement systems. They are not electrical measurements done with an oscilloscope, such as looking at a square wave response or checking for 'ringing'.

My point was not to debate the values of a capacitor, rather it was to address the basic topology.

[DrWho]

"Time offset" refers to the output of the tweeter needing to travel a further distance than the woofers because the compression driver is mounted to the rear of the horn. In other words, the sound from your woofers is arriving before your tweeter. Yes, it's a very short period of time, but that's the number one advantage to real bi-amping (which involves an active crossover). You have to look outside the typical steady state analysis of audio and look at the transient behavior (in other words, your frequency response isn't consistent over time).

If I may make an observation...

Right now you are trying to get rid of the passive crossover in the speakers by duplicating the effects of the crossover at line-level to the input of your amplifiers. When properly implemented this isn't fool's bi-amping - it's just a passive crossover implementation. But to stress the point - you're trying to copy the current engineered crossover...so at best you'll be able to achieve the same acoustical response (with whatever benefits might be achieved by bi-amping). I doubt you'll be able to achieve an identical crossover, but since there's no measuring involved who is to say that the difference is indeed better? My experiences with bi-amping have always indicated that a proper crossover transition is much more important than any distortion reduction in the amplifiers. We're talking orders of magnitude difference if you've ever sat down to measure it (I personally haven't measured, but I've read plenty of articles on it).

You've got a computer handy and there are a few free measurement programs out there. For around $100 you can purchase the necessary hardware and land yourself some powerful measurement capability. And then you'll be better equipped to re-engineer your system to your heart's content.

[thoppa]

It might seem a dumb thing to be doing but it is because I don't have test equipment that I'm relying on my ears and my math, so I have tried to change the set up and keep the sound the same or better. One day I shall invest in some test gear - and an active crossover - but at the mo I'm strapped for cash.

Anyway, thank you for explaining the time offset. I realise now this will need an active crossover - it isn't to do with phase changes in a passive network. My bad.

I have assumed the passive phase changes will pass through the amp but as the signal is converted to pwm in the amp and then back again, is this assumption incorrect ? Is the only phase change due to the cap in line with the tweeter ? As you said before, will I not hear these changes anyway ?

Thanks again,

Tom

[DrWho]

Relative phase and amplitude remain constant through any device that doesn't affect the frequency response. In other words, the phase going into your amp equals the phase going out of the amp. Any device that changes the frequency response will have the same change on the phase response - so it doesn't matter where the filters occur. There are rare exceptions to these rules, but none of them apply here. (in fact, you need active processing to affect them differently).

[thoppa]

Thanks again for teaching me ! I've been looking at a Behringer digital crossover - are they any good ? It has the various crossover types and orders, attenuation and delays , which I now understand the need for...:-)

Cheers,

Tom

[DrWho]

You're going to find very polar opinions about the Behringer stuff - I'm gonna cop out and not comment []

I believe sfogg on the forum is using the Behringer for his overkill bi-amping setup - using digital amps too I believe. There are some threads somewhere where he details all the minor details he worked out to maximize the SNR and resolution of the DACs in the Behringer. I'm yet to hear his system, but I'm sure it sounds amazing. He's provided a great example for implementing active crossovers correctly - though some might argue that he's gone to the extreme...at least he's not playing with fancy power cords (yet). []

If he doesn't chime in I'll go hunt him down and have him post his links.

[thoppa]

Hi,

Thanks - I've done a search for his posts and will trawl through them.

The ideal for me would be a digital crossover that takes an optical feed and a feed from a HD DVD/Blu Ray and has top notch DACs and a sweet headphone amp....and is "affordable"......fat chance eh.....

Anyway, my system is sounding quite good and I'm getting a decent-sized stereo sweet spot so now it's not 'urgent'. It might be an idea to wait until the latest format war has settled a little and 24bit/192Khz is common....

Many thanks for your help - you have saved me from several mistakes - I really appreciate it []

Tom

[thoppa]

For anyone following this thread and hoping to build a 'good' system for $1,500 here is my effort. Comments more than welcome...

Musiland MD-10 DAC - PLLXO - Class-T Bi-amp - Klipsch RF52 (no crossover)

PLLXO - Matched to low-mid output impedance of Musiland MD-10 DAC headphone output (40 to 80 ohm seems to work best so I've gone for the higher end to minimise distortion) and 20K input impedance of Charlize class-T amps :

PLLXO Caps are Sonicap Gen I and resistors are PRP .5W from Sonic Craft (I'm upgrading from the Solens and carbon film resistors I have been using to set things up)

Low pass - 82 ohm and 820 ohm series, 2.2uF and 0.22uF parallel. This 12dB per octave at 900Hz. Higher crossover frequencies make the speakers sound shallow and really really horrible. I started to hear awful resonances that has made me interested in better damping for the cabinet side walls and the internal braces. There's plenty of well-defined bass across the range so I don't mind losing a little if I might get a smoother overall response.

High pass - 0.47uF and 0.047uF series, 82 ohm and 910 ohm parallel (910 will be increased to alter the roll-off - actually the sound is not how I'd like it yet so I expect I'll increase this but it is a close match to the original Klipsch in theory). Two 12K ohm resistors are used as a potentiometer to replace the power level attenuator. Again, I expect to play with these to vary the overall bass/treble balance (being careful to keep the overall value including the amp's input impedance at 20K !).

Power level capacitor will be 3.3uF Mundorf M-Cap Supreme silver/oil when the postman comes....I'm waiting by the box.....

I'm still fussing over stuff like phases, optical cable and power supply caps but I need to burn-in everything together before I do more changes. As for time-shifts, well, I need to hear if there really will be an improvement that would far exceed the problems of adding an electronic circuit in the path. I feel, but don't know, that positioning everything well will minimise this and it is overkill. Am I way off again ?

Total cost is a shade under US$1,500 including power conditioner, ASIO soundcard, cables etc.

Very pleased overall.

Cheers,

Tom

[Guest " "]

Probally need to start a seperate thread since this looks like a SS/op amp thread.

I just accuired a second decent tube amp to partner with the one I have now. 

 So next steps will be for me to either build a tube crossover or modify a tube pre-amp kit to function as one. 

 

I found a tube pre-amp kit that uses fully balanced XLR connections with +,-,G signal path using a seperate tube for the +,-.

I know there are tube crossovers out there, but for the price of them, there's considerable cost savings to DIY.

[thoppa]

Hi,

If you do the the DIY route please post copious details as I'd be very interested in this !

I read posts about the benefits of XLR and it would be a good step up for me to replace my RCAs with XLRs as I have six meters of RCA cables in total. Where is the tube pre-amp kit you mentioned ? Can you post a link please ? Thanks.

Tom