I'll see if i can make it work, but I have some doubts
could be its a different matter with fullrange drivers
My experience is quite the opposite - best with fairly steep band limiting. I'm listening to an elliptic band-limiting filter at the output of my DAC right now. So far the best sound I've got out of this DAC - the elliptic just replaced a 4th order Burtterworth (the current one is 7th order, 17kHz cut off, -50dB by 22k).
Lavry makes multibit too. I agree that ears are the arbiter, not meters but 'measures bad' is highly context dependent.
Ouch! no treeble.
Between drilling treeble and attenuation I would also choose the latter (and i did with the FR), but with a good dac and speakers it shouldnt be needed. I'm afraid something is wrong in your chain.
The real problem with early lowpass filters is the phase change in the middle of the audio band. I guess that your 7th order @17khz would be more or less equal to a 4th order at 30khz, or better so not that bad in the audio band (but i suck at math so probably i'm wrong ).
Yes, this explains why it sounds good to you.
I didnt dislike much the old brickwall filters and digital is now allowing us to play with very steep filters without much degradation... but I'm still afraid of anything higher than 4th order with passive components.
). Yes, this explains why it sounds good to you.
I didnt dislike much the old brickwall filters and digital is now allowing us to play with very steep filters without much degradation... but I'm still afraid of anything higher than 4th order with passive components.
Ha you need to be careful in judging the phase performance from just the order - this is an elliptic, the group delay plot looks fairly benign (30uS or so) up until 14kHz where its doubled (60uS) and after that it goes totally ballistic, reaching over 250uS at the cut off. So you think I enjoy this kind of phase distortion? How does phase distortion improve the soundstage do you think?
D.Self did an active crossover with adjustable frequencies way back in the 80's.
This Forum ran a few Threads on adjustable crossovers. There were 3 or 4 different versions on their own PCBs. Read Mox.
These can all be done as plug and play.
A case with the plug board facing the bottom could be very quickly changed by removed the protective cover and swapping a DIP plug with the new frequency components. Or use DIP switches.
No problem at all since the components which determine frequency and slope are on little plug-in cards and I have a few spare ones.
I can change and A/B in a matter of seconds.
Stochastic interleave crossover-filter alignment
Just came across this variation on active crossovers. I don't know whether it works as advertised, but if it did, it might be something worthwhile that passive crossovers simply can't do. Has anyone tried this? (Unintentionally, I may have done something very similar in one experiment, and it sounded good, but I realised my 'mistake' and 'corrected' it...)
http://www.essex.ac.uk/csee/researc...Analogue and digital crossover alignments.pdf
Just came across this variation on active crossovers. I don't know whether it works as advertised, but if it did, it might be something worthwhile that passive crossovers simply can't do. Has anyone tried this? (Unintentionally, I may have done something very similar in one experiment, and it sounded good, but I realised my 'mistake' and 'corrected' it...)
Because in the principal crossover region at least two drive units are radiating significant energy, then interference will occur especially when the path lengths change as a function of angle as is the case with non-coincident drivers. In this Section, the incorporation of a static random vector in the crossover filter response is explored as a means of reducing the subjective significance of polar response errors. This random sequence together with selected frequency shaping is termed the interleave function.
A design requirement of a crossover filter is that the loudspeaker should have a well-behaved off-axis frequency response. This is normally described in terms of the polar distribution which is influenced both by the frequency dependent spatial characteristics of a drive unit, the physical spacing of the drive units and the baffle size and profile. The frequency region in which the crossover has influence is defined in terms of the filters' attenuation characteristics together with their relative phase response. The effect of non-coincident drive units is to introduce a time delay between the soundfield radiated from each drive unit. This in turn can be represented by a frequency dependent phase difference, which creates interference.
To implement a stochastic crossover alignment a zero-mean random vector rd
The effect of this process is to leave the composite response unmodified but the difference response, that determines the off-axis lower interference bound, is noise like. Consequently, inter-drive unit interference that occurs in the filter transition band is distributed more broadly with narrow bands of constructive and destructive interference. An example stochastic interleave crossover is shown in Figure 5-1, while in Figure 5-2 four off axis composite plots are shown with a progressive increase in time delay between drive units. Complementary randomization in both the high-pass and low-pass filters should be observed together with the destructive envelope. It is conjectured that with music signals or any non-periodic signal which must have a broader spectrum, that there are beneficial effects to be gained from this randomization especially as the micro-structure of the interference varies with angle and is not localised to a discrete frequency null. In this sense, a similarity with a distributed mode loudspeaker (DML) should be observed [NXT White paper, 1996]. Finally, it should be noted that the proposed noise additions are static and consequently the resulting changes in frequency response are also static and do not result in actual noise signals.
http://www.essex.ac.uk/csee/researc...Analogue and digital crossover alignments.pdf
This is a possible (theoretical) advantage of series over parallel passive crossover topologies, but with active the amplifier does the same thing even better . . . it "washes out" the effect of the dynamic changes in driver behavior. I'm not convinced that it makes a whole lot of difference, but the immunity from driver effects, dynamic and otherwise, that the intervening amplifier provides is one of the arguments for an active design.
I am sure that this has been mentioned before on here but I feel that Linkwitz's explanation of why active is better makes perfect sense..
The amplifier has more control over the driver and you can implement different types of crossovers for the best possible sound.
That said, I also feel like having to have 2,3, or even 4 channels of amplification for each speaker gets to be a little bit ridiculous and sends the power bill through the roof.
The amplifier has more control over the driver and you can implement different types of crossovers for the best possible sound.
That said, I also feel like having to have 2,3, or even 4 channels of amplification for each speaker gets to be a little bit ridiculous and sends the power bill through the roof.
Not an issue if you use appropriately sized amplifiers. Four 60 Watt amps (on the same power supply) don't draw any more idle current that a single 500 (which is what it takes to match their performance).
Any dynamic of driver that is not linearly dependent to drive signal is non-linear distortion.
How do XO components working within their linear regions correct a non-linear behavior?
Regards,
Andrew
For one, all dynamic drivers change their T/S parameters with level (and temp & barometric pressure), T/S are not scalars but curves -- no driver has completely horizontal curves
dave
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.