diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Analog Line Level (http://www.diyaudio.com/forums/analog-line-level/)
-   -   Audiophile active crossover (http://www.diyaudio.com/forums/analog-line-level/230591-audiophile-active-crossover.html)

argonrepublic 22nd February 2013 02:01 AM

Audiophile active crossover
 
My amplifiers are near complete. I am using Dynaudio D28/2 and Dynaudio MW160. Crossover frequency around 2k. Currently I am using a Phoenix gold MX3i automotive 3 way crossover. I have been told that it is a good quality unit and is a candidate for upgrades. To me it sounds very good but I can not compare it with anything. I am contemplating some 2 way crossover PCBs to mount in my enclosures. I seen them on ebay and am unsure of the quality. I will also have to find another place to separate my subwoofer. I could buy a rackmount type of crossover but am unsure of which to choose. Many of them are music store units. In tems of sound quality, I have always been left wanting from any playback item that has come from a music store. I have also considered scrapping any resale value of my phoenix gold and mounting it in a rack with a nice clean 12Volts, audio grade caps and good quality ganged pots instead of independent left and right level controls. I am appreciative of any suggestions.
Thank you.

Jay 22nd February 2013 03:54 AM

Poor the Dynaudios to have such a crossover. The opamps are of LF353/NE5532 quality.

prairieboy 23rd February 2013 10:51 PM

Hmmm ... and yet Doug Self (who is certainly an accomplished and respected designer) thinks they are more than adequate to design a preamp, with phone stage, using them. Perhaps if there is a problem with that unit, it isn't the fault of the op-amps?

Jay 24th February 2013 01:22 AM

Quote:

Originally Posted by prairieboy (Post 3383429)
Hmmm ... and yet Doug Self (who is certainly an accomplished and respected designer) thinks they are more than adequate to design a preamp, with phone stage, using them. Perhaps if there is a problem with that unit, it isn't the fault of the op-amps?

As for listening, many have been in disagreement with respected designers.

As for opamp fault, in active crossover, the opamp is not only one or two, but many, and it matters.

As for the unit, the circuit used in the said machine is not better than ESP active crossover. At Linkwitz site if I'm not mistaken, it was mentioned that without a delay network an active crossover is pretty unusable.

If there is a reference opamp-based active crossover, I think it is the one used by Linkwitz' speakers. Note that the crossover has been part of certain speaker design.

I haven't heard the Orion or the newer speaker, but I think the strength of the speaker is not the opamp. The speaker should still has the opamp sound quality weaknesses but the strength is a top design that is almost impossible to be achieved with single amplifier or passive crossover.

argonrepublic 24th February 2013 02:20 AM

So then the first thing would be to swap in some OPA134?

Jay 24th February 2013 03:50 AM

Quote:

Originally Posted by argonrepublic (Post 3383580)
So then the first thing would be to swap in some OPA134?

Theoretically NE5532 is sufficient for the job. I don't think OPA134 will give quite an improvement, if any. I have many opamps and I have the crossover but I didn't do anything except using better power supply which I have on hand.

Well, sure you can improve anything. I just think that "audiophile" label doesn't fit well with this kind of circuit.

Calvin 24th February 2013 04:31 AM

Hi,

The requirements for phono are very different to those in an active crossover.
Phono is one of the few areas in audio where high gain is required at the same with filtering functionality. Another would be DC-servoes. Both's amplitude response is quite similar to the openloop amplitude response of OPamps.
A active crossover On the other hand may be constructed the most from buffers utilizing a gain of just 1. You may search the web for 'unity-gain Sallen-Key' filters.
In most textbooks you'll find the typical triangular OPamp icon for the active part in the filter, sometimes a '1' added to it. You can easily replace this OPamp by a discrete buffer made as simple as from 2 transistors, either bipolar or JFET. I prefer JFETs, because of circuit simplicity and sound. The transistors would be configured as emitter- or source follower, loaded by a constant current source. Nothing complicated, nothing expensive but much more musical and true. Such a simple buffer may not be totally transparent THD wise, but the added THD is low and THD distribution is optimal. Cascoding JFETs, thereby increasing the parts number count to 4 JFETs per buffer would allow for a THD transparent circuit structure.
A second important point is to reduce the number count of active stages.
If you compare active to passive crossovers almost all active filters are just that ... filters, sheepishly following textbook ideals. Passive filters on the other hand include equalizing functionality, which is what any speaker requires.
Sometimes one can find additional EQs in active filters, but simply in form of additional circuitry. Nothing hinders You to deviate from textbook formulas and to design the filter response as required by the speaker. If for example your Midrange driver requires a slight boost at the lower bandwidth limit for a linear amplitude response, a butterworth filter with a Q of 0.7, or even worse a Linkwitz-Riley filter with a Q of 0.5, would certainly be the wrong choice. A non-standard filter with the amplitude response and Q taylored to the requirements would be alot more convenient, elegant and lower on parts number count. And the best for last .... it'd sound better too.

jauu
Calvin

CharlieLaub 24th February 2013 09:05 PM

Quote:

Originally Posted by Jay (Post 3383546)
At Linkwitz site if I'm not mistaken, it was mentioned that without a delay network an active crossover is pretty unusable.

One can not say this is a true statement for all cases. Sometimes you can do completely without delay compensation. As an example, how many passive crossovers include delay stages???

What you want is for the phase response of each of the two drivers that you are pairing with a crossover to be "tracking closely" around the crossover point. You can state the design goal as something like "the relative phase angle between the two drivers when connected to the crossover network should stay within 45 degrees whenever one driver is within 40dB of the other." There is more than one way to attain this goal.

Since filters have a (non-constant) phase response, one can use one or two extra orders of HP filter on the tweeter to better align the phase response with the mid/woofer. This approach works for both active and passive filters. The asymmetry of slopes is of no concern. As long as the phase tracks relatively well and you can get the combined amplitude around the crossover point to be relatively flat (e.g. by adjusting the Q and corner frequency of filter stages) the result should be more than satisfactory.


-Charlie

vacuphile 24th February 2013 09:58 PM

I like the topic of an audiophile active crossover a lot, because that is my present puzzle.

An issue that will certainly be more important than the difference between a NE5532 or an OPA2134 is a correct choice of filter slopes. It is too easily assumed that with an active crossover, driver-in-enclosure parameters drop out of the equation. Nothing less is true.


Every driver in a speaker will have its typical roll of points on the low and high ends. These natural roll offs will have the typical phase shifts of a minimal phase system. In other words, on the low end drivers normally have 12dB/octave high pass slope, and that causes 180 degrees phase rotation around the -3dB point. Now, there are only two ways to go. Either you stay well away (2 octaves) of this -3dB point, or you incorporate it into your filter design. This is where it starts. Of course you need to do measurements for this, but the capability to do so has become democratized to a high degree.

Jay 25th February 2013 12:28 AM

Quote:

Originally Posted by CharlieLaub (Post 3384531)
how many passive crossovers include delay stages???

As many as good active crossover available :D Just kidding.

Quote:

Originally Posted by CharlieLaub (Post 3384531)
What you want is for the phase response of each of the two drivers that you are pairing with a crossover to be "tracking closely" around the crossover point. You can state the design goal as something like "the relative phase angle between the two drivers when connected to the crossover network should stay within 45 degrees whenever one driver is within 40dB of the other." There is more than one way to attain this goal.

Since filters have a (non-constant) phase response, one can use one or two extra orders of HP filter on the tweeter to better align the phase response with the mid/woofer. This approach works for both active and passive filters. The asymmetry of slopes is of no concern. As long as the phase tracks relatively well and you can get the combined amplitude around the crossover point to be relatively flat (e.g. by adjusting the Q and corner frequency of filter stages) the result should be more than satisfactory.

Active is not as flexible as passive crossover. Most often (like the one discussed in this thread) you only have pots that determine the roll-off frequencies. The slope order is fixed. You may have a boost at around 45Hz (often also in fixed dB). What kind of drivers blending that can be achieved with such tools?

With passive, solutions are usually integrated (there are many ways to achieve the goal). For example, you don't need to create a separated BSC circuit to compensate for baffle loss. So is with phase. Different case with active crossover. With active you have to ADD a dedicated bandpass filter to alter the phase. But you cannot cross at any frequency you want and get a perfect phase tracking.

Phase is imo the major issue with crossover design, and with active design this is a really really terrible issue.


Quote:

Originally Posted by CharlieLaub (Post 3384531)
the result should be more than satisfactory.

May be not so for audiophiles.


All times are GMT. The time now is 10:16 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