What X-over for a hybrid ESL?

[CharlieM]

Calvin just made some very important points (as he usually does).

From your post I figured you had a preference for analog over digital, and an analog X-over can certainly sound great when compatibly matched up to the other gear.

And I forgot to mention that I had to use a separate multi-band EQ in conjunction with my old DBX analog crossover, to EQ out my panels' dipole phase-cancellation and to tweak the sound to my personal hearing characteristics.

My old DBX analog unit used 24db/oct slopes and my speakers are hybrid ESL's that have a nasty-loud diaphragm resonance around 70Hz-- so, in order to not excite this resonance, I set the crossover frequency a full two octaves higher. But now that I have the DSP crossover with it's marvelous flexibility, I have the option to use a steeper (48/db) slope, which allows dropping the crossover frequency a full octave lower without exciting the diaphragm resonance-- this takes practically all of the woofer's energy out of the midrange, where it performs poorly but the electrostat excels. Not to mention, the DSP has parametric EQ's on every band (so I no longer need a separate EQ), and digital time-delay which allows me to precisely time-align the woofers/panels and left/right speakers. Also, I sometimes get some annoying and scary-*** transient "pops" over internet radio, and the DSP has limiters I can set to prevent those "pops" from damaging my gear and giving me a coronary.

Admittedly, my old analog crossover setup sounded wonderful too, but for all the reasons I just mentioned, you'd have to pry my DSP from my cold dead hands before I would ever go back.

 

[DjSinae]

the reason why i'm investigating both options is that

I can always replace the NE opamps of the crossover later with very good AD797.

I can also change the 4 volume pots for passive volume control..

I wonder what is better the 8 channel dac with 8 internal digital volume control of the minidsp 2x8 or 4 passive volume in the opamp Xover..

Latency is really what pains me with something like minidsp... Since it's not the one having the asio drivers, i wonder if it communicate to the source board(with asio drivers) for asio latency compension.

I watch, listen to everything on my computer and i also want to play games.. I have at least a rme that give me like 3ms latency on outputs at

i still don't have the minidsp but i'm afraid too much latency can be added to the signal path and that no compensation is done so every movie + games + computer djing + music production + will lag.

As for EQ correction on the analog part, the best is to use a FIR EQ before the co3 crossover. That FIR EQ can be in the computer or soundcard..

Of course, my idea is to first use the minidsp to analyse my setup and afterwards build myself a lower latency one with the Co3.. afterwards, once money flow and that i know i like this setup, i will buy the AD797 myself and change the NE opamps.

Also note that minidsp will not give you phase linear xovers.. and that brick wall FIR xovers have pre-echo(they say lower FIR xover curve doesn't have this problem). Also i will add even more latency to the chain..

So i'm stuck and that's why i asked for comments. I will try the minidsp way but too much latency will kill it for sure. Also i wonder how much phase linearity impact the sound at the crossover point.

of course with analog crossover.. Time-alignment of the driver + baffle step correction will be necessary..

I'll measure the time-alignment needed with minidsp and measure the best distance of the drivers that way.. and i will have 1/2" or 1" rockwool on top of the baffle to absorb most of the baffle irregularities and baffle corners diffraction.

I never saw my idea on the web.. I wonder if any of you have toughts on this..

I will do an 2 on-wall - open baffle - on a corner of my room..

My room is shape like a L and at the back of the corner i have the open kitchen.. so lots of diffraction there.. Also in this position, I will have less reflection from the side and more balanced L-R reflection than if one side would be near a wall and the other in open space..

I will also have a superchunk basstrap on the corner behind the speaker and wall mounted 2" rockwool on the side of the speakers.

The thing i'm not sure is how bad is it to have an OB in a corner. (Like how much rear wave reflection is needed and if absorbing to much of it is bad (because of superchunk and other wall panel)

At the same time, I mostly sure the best is to absorb the most i can since is very near the corner and that rear wave reflection will be totally strange in a corner.. (like one monitor rear wave will bounce and go to the other wall and bounce back on me..)

Maybe in closed box would be better.

 

[Andersonix]

My thoughts on XO for ESL today are based on my listening experience with the stock passive XO in my SL3: It seems sufficiently transparent in the ESL, but apparently let down by the inductor for the 10 inch woofer muddying up the bass. (Although I have recently retrieved some bass clarity by improving the power supplies everywhere in the chain and thus generally better resolving the upper harmonics -- where the realism lives.)

So my next step will be to go active on only the bass (as I have a cheapie 12dB/oct sitting idle) while keeping the passive XO on the ESL only. Probably a MiniDSP could instantly replace all the passive stuff in there and yield a sudden improvement in overall performance, but I want to experience for myself the effects of every tiny upgrade step along the way and not get ahead of myself, so my upgrades come in baby steps.

But if I were to start from scratch today, I would choose the MiniDSP (as it appears very flexible and quite future proof) and then later obsess over cleaning up its power supplies and analog outputs.

 

[DjSinae]

I have another question for you... if i have a

3 way system with bass on one side and mid/tweeter vertically on the other

7" bass 50hz
4" midrange btw 100-400 to 3-5khz
ribbon 3-5khz

I know time-aligning the 4" and ribbon is essential but

what about the 7" bass if it stick to bass duty?

Is this essential since bass waveform is more than 1meter long anyway

Also anybody know how to program minimun phase IIR crossover with minidsp?

 

[Calvin]

Hi,

I can always replace the NE opamps of the crossover later with very good AD797.
I can change...

In contrast To good ol´ Barrack O. You really can
But this wouldn´t be the solution to the problem in first case.
Replacing OPamps with OPamps doesn´t omit with the general flaws of the analog Xover and it certainly won´t let the technical/artificial note in every OPamp I came across vanish. The co3 is quite cheap. As such the power supply chain might not be adequate for the AD797. I wouldn´t be surprised if the replacement of the OPs would result in oscillations.

I wonder...minidsp 2x8 or 4 passive volume...

Volume control in the digital domain is basically just calculations. As such the results are consistent and precise. Interchannel differences are negligible. Ageing, temperature, dust/dirt and noise are not problematic. The useful volume range is lower. As a rule of thumb, if it takes more than 10dB to set the level correctly You might think about an alternative. On the other hand is a analog poti best when hardly used at all (low attenuation).

Latency is really what pains me with something like minidsp..

Why?? Following the signal chain there´s is just an amplifier left and the speaker. You don´t feed the signal back into a signal processiong chain.
So the DSPs Latency time is nothing but a small delay. If playing music alone latency is of no importance. If gaming or watching vids how much delay may be ok, before You can see the non-synchronousity of speech to optical impression? Compare the latency to the blinking of an eye, that takes 300-400msec.

As for EQ correction on the analog part, the best is to use a FIR EQ before the co3 crossover.

That is very debatable. Since speakers are minmum-phase systems (amplitude and phase are linked) a mirror like amplitude equalizer will correct for the phase response at the same.
This is exactly what analog filters and digital IIR filter do. The most remarkable property of digital FIR filters is that the linkage of phase and amplitude doesn´t apply. This is useful as a finalizing step if You correct for the time response of the complete speaker system or to do room correction.
It is alot more convenient though to leave the filtering/equalizing of the single speaker channels to IIR filters.

Of course, my idea is to first use the minidsp to analyse my setup and afterwards build myself a lower latency one with the Co3.. afterwards, once money flow and that i know i like this setup, i will buy the AD797 myself and change the NE opamps.

Again, check if the small delay of the digital part plays any significant role. BTW. analog filters also introduce latencies....they are just called groud delay or delay As to OPamp swapping, I´d invest the money in something more ´efficient´

Also i wonder how much phase linearity impact the sound at the crossover point

One thing is absolutely impossible... that the analog co3 gives You a linear phase response! It could do this only working into a pure resistance but not working into a real loudspeaker (the power amp in between may be regarded transparent as to amlitude- and phase-response).

...and i will have 1/2" or 1" rockwool on top of the baffle to absorb most of the baffle irregularities and baffle corners diffraction.

Then use a large ESL panel and omit with the baffle alltogether

At the same time, I mostly sure the best is to absorb the most

As with all highly directive systems focusing will be very high, but the image of a natural stage might probabely be missing. It is a major advantage of dipolar systems that they offer one degree of freedom more regarding the acoustic image influencing mechanisms. While the highly directive frontal acoustic waves lead to precise focussing, the delayed reflected rear waves create the ´room or stage´. A Corner as reflector and diffusor for the rear waves should be fine. Care needs to be taken that no direct reflections back through the speaker occur and that the diffuse reflections hit the ear with sufficient delay.

jauu
Calvin

 

[Andersonix]

While Calvin's replies are pertinent to ESL XO and interesting in general, I would encourage DJSinae to ask his thoughtful but specific questions about his proposed ribbon system in a separate thread here in the 'exotics' forum, as I think this would be more helpful for all (and also to help keep this thread from going too far off topic)?

 

[Chii]

Guys great topic for me as I have just bought a c03 pcb board to build one (yesterday before i saw this thread). I wasnt aware of the mini DSP and will now try both over the next 6 months starting with the C03 moving to the minidsp 2x8 at a later date.

For me this does cause some questions over sound quality on the minidsp and the c03 as I am using some quite good kit at every level with one of the source's actualy being 96/24 (and capable of going higher if the media ever becomes avalable) I hate the idea of re digitising a source when the rest of the system is analog and after its left a pair of WM8741's already. at the same time 26 opamps in the signal path doesnt seem like a good idea either so i am unsure ware to go with this long term.

I will of cause update when some of the kit comes in.

 

[bentoronto]

To begin with, lots of good reasons today to plan to be tri-amped if you are DIY. Easy to buy great second-hand amps cheap. Nothing else makes much sense unless spouse-factor or economics predominate design issues.

A whole lot of design and crossover issues disappear or solutions get readily determined when you have a single main driver/panel that covers a large middle frequency range. That range would be from sub-woofer crossover (at 24dB/8ave, that can be as high as 120 Hz or even a bit higher) to say, 3500 Hz.

Keeping crossovers out of the main music range save a lot of problems and sounds good, as the "full range" crowd correctly think. Then adding one purpose-built mixed-bass sub for below and some ordinary good tweeters for above is relatively easy and without many heart-rendering issues to debate.

Question: I'd like to learn about distortion in DSP equipment; can someone provide links for me.

Ben

 

[Audionutz]

Hello to all !

I am a recent Acoustat Model 6 owner, after a long and varied history of speakers spanning a great many different models.

I'm really enjoying what the 6's have to offer, but wanted to integrate two subs into the system.

I would like to know if its possible, and if so, how to go about rolling off the bass in the Acoustat panels at 100Hz at 18db octave. I would like to do this passively, without using an active crossover, buffer or network of any sort. I dont want to try cap(s) on the amplifier inputs, as this will only give me a 3db slope(?)

I'm not as technical as most who post here, but can figure most things out, with a bit of help !

Thanks in advance.

Scott

 

[bentoronto]

You don't need two subs (one great sub sounds better than two poor subs) and you do, inevitably need bi-amping. So you'll then need an electronic cross-over. That's easy.

If you have some reason to be purist about the Acoustats, you might find the new sub (or subs) sound can be shaped to be quite congenial to the natural unfiltered Acoustat low-end roll-off.

Ben

 

[Audionutz]

Thanks Ben,

As I have the two subs anyway, I'd prefer to use them. Is there a reason I need to biamp, if the existing amp drives the Acoustats without problem ? Also, both subs are active, which I probably didnt make clear.

I would prefer to use passive, as I've always found it sounds better.

Scott

 

[bentoronto]

Thanks Ben,

As I have the two subs anyway, I'd prefer to use them. Is there a reason I need to biamp, if the existing amp drives the Acoustats without problem ? Also, both subs are active, which I probably didnt make clear.

I would prefer to use passive, as I've always found it sounds better.
Scott

There is no practical way to cross-over at 100 Hz except electronically, and less feasible if trying to match a pair if you have two woofers.

One woofer signal is all anybody needs. But, your woofer sound will be far better if you have two (playing the same signal or the normal stereo way) but putting the two speakers in distinctly different parts of the room (like corner and mid-wall).

I don't think you'll find a lot of support around here for your belief in the superiority of low-level crossovers. Except maybe coincidentally from those who find shallow curves better... since low-level are always shallow.

Ben

 

[georgehifi]

There is no practical way to cross-over at 100 Hz except electronically, and less feasible if trying to match a pair if you have two woofers. Ben

Not so sorry, Martin Logan have rolled off their ESL panel to the 12" bass driver at 125hz with their passive xover at 12db per octave.

Cheers George

 

[bentoronto]

Yes, many of us are aware that there do exist (in theory) large enough (iron-core?) coils and large enough (electrostatic?) capacitors to cross-over at 100 Hz. Of course at a mere 12dB/8ave, you end up with less cut (and less protection) to the ESLs at 50 Hz and below than the "rumble filter" on my pre-amp... so why bother?

I'll stick with saying that for Audionutz it isn't practical, makes no construction sense, hard to measure, expensive, can't be tweaked without heroic efforts, the design requires coordinating with the unknown present frequency roll-offs of the drivers (as ML has done), and likely won't sound right at first if ever.

Does it make sense (at 125 Hz) for a manufacturer of speaker systems you take out of a box - who at least can make, measure, and closely-pair (and closely-pair again for stereo) giant iron-core coils and electrostatic capacitors? Seems ill-advised - but that's a different question.

Ben

 

[georgehifi]

Sorry again they are all air core for the ESL and for the the bass xovers, I have the bass active though at 24db and bypassed the passive xover for it.
I tried active for the esl with both discreate and opamp based active 24db, but they always sounded electronic, they could not compete with the soothness and transparency of the good quailty passive xover that came with the ML's, but the bass I aggree is better when done active.

Cheers George

Cheers George

 

[bentoronto]

Sorry again they are all air core for the ESL and for the the bass xovers, I have the bass active though at 24db and bypassed the passive xover for it.
I tried active for the esl with both discreate and opamp based active 24db, but they always sounded electronic, they could not compete with the soothness and transparency of the good quailty passive xover that came with the ML's, but the bass I aggree is better when done active.

Cheers George

Cheers George

Perhaps there are a dozen differences using recognized principles of electricity between your electronic crossover system and the ML system that might account for the difference.

Ben
It is important to make a the low-level crossover wires using an organic insulator locally sourced for smoothest sound. Actually, pro-biotic insulation makes the sound more digestible.

 

[Audionutz]

Lots of different opinions it seems. Hard to sort the "wheat from the chaff" !

Every active crossover Ive heard has always had a signature (except where the system masked it). With something as accurate and transparent as the Acoustats, Im sure there would be an audible difference.

As I said, I'd like to go passive and it seems from comments here and elsewhere that it is definitely possible. And with very good results.

Now I just need someone who has done it on my speakers, or at least has the freq and impedance graphs for this model ?

Scott

 

[georgehifi]

Found a very bad quality x2 impedance graph done by some German company, looks like at 100hz at where you want to do the xover it's about 3.5ohm.
So now you need a guru to work out how many uF series capacitance and how many millihenry air core choke shunt to ground, and you'll have a second order xover at 100hz.

Cheers George

 

[invisible force]

Any serial component, be it resistor , cap or coil connected to the ESL transformer primary winding will raise distortion : the nonlinear current caused by the core will lead to a nonlinear voltage drop across the serial component - ESLs are best driven active with short speaker cable to the transformer...by the way it makes sense to keep resistive part of the primary winding low for this reason...
to sum it up : core distortion ideally is only in the current, not in the voltage..
the amp should be able to drive an almost pure cap with low distortion...most of the existing amps unfortunately fail to do so..
Greetings to all Dayton Wright owners...
Philipp

 

[bentoronto]

OK.... we're fast off to a bad start. I'll try not to say "told ya so...."

The intractable (and intrackable) impedance curve (and other comments) are just a start. You also need to know just what the speaker output is in the crossover range and below. Traditionally, one of the ground-rules is that the driver/speaker is pretty flat across the transition band, for very obvious design reasons. That band is monsterously wide at 12dB/8ave. Do the numbers.... sort of like 50 to 200 Hz.

Let me add, it IS possible to do what your want. ML tries to do it, albeit 125 Hz is a bit easier. But the development effort to get something even half-passible is enormous and needs test gear. On the other hand, using an electronic crossover is as simple as could be.

I know that not everybody shares my belief that inexpensive second-hand amps from 1980 bought at pawn shops are more than good enough to use with great speakers. But if you did, bi- and tri-amping would look more attractive.

As you can see from my signature, I firmly believe have a great speaker cover the range right down to low-woofer frequencies is wise. In terms of rooms and stereo, 130 Hz (at 24dB/8ave) is just fine for a single mixed-bass woofer and maybe even higher if the woofer is directly behind one of the mid-ranges.

Ben