Passive Line Level Crossovers and Alephs (from a Neophyte)

[GeorgeBoles]

Dear All,

I am going to build an Aleph 30 and an Aleph 60 (made up as mono-blocs), which i am going to use bi-amped with a BOSOZ pre-amp. (Lots of building to do .) I am going to try to keep the connections between the BOSOZ and the Alephs balanced, and put the Alephs as close as possible to the speakers.

Where would be the ideal point in an Aleph design amplifier, esp the Aleph 30 and 60, to insert an input from

a) a passive line level crossover;
b) an active line level crossover?

Regarding the passive line level crossover:

I had thought that for the passive line level crossover, inputing the signal just to the far side of the Gate (221R) Resistors of the input differential pair would be the way to go, i.e. on the side further from the Gate.

I thought I would graft the input resistor network from an Aleph 2 or 5 onto the front of my passive line level crossover to "decode" the balanced signal into unbalanced form. Does this make sense, or is life much more complicated than this?

I also thought that the lower impedance of the Aleph 2 or 5 front end rather than a 30 or 60 (to be grafted onto the front of my passive line level crossover) would be preferable because of the lower impedance (higher signal voltage) which would be fed from the XO to the differential pair. Is this the right thing to do?

(Why passive? My preference is to try passive first in order to keep with the "simple" philosophy of the Alephs, rather than to add hundreds more transistors between signal and speakers.)


Regarding an active line level crossover:

Many of the active line level crossovers handle balanced inputs themselves, and with the "amplification" of the signal provided by the "active" bit, that the output of the crossover could be fed to the unbalanced inputs of the Alephs. Is this correct?

I have yet to finalize my thoughts regarding LC or RC designs, and the actual passive filter topology. The LC idea is much better documented in normal HIGH level speaker crossovers. It appears to me that the concepts of RC and LC designs are similar except that RC has much higher signal loss, but I have not investigated this too much as yet ... I can work on this when I am not actively soldering up my Aleph Clones.

I shall now recite my current mantra:

"Thank-you Nelson Pass for your generosity and magnanimity... " (Repeat until asleep at night)

Regards,
George.

PS Note: It is probably obvious from the above letter, but I am a bit of a neo-phyte with regards to this stuff, and actually "designing" anything electronic is a bit of a gamble ... hence the request for help.

 

[Hattori Hanzo]

A good reading on passive line level crossovers is here: http://www.t-linespeakers.org/tech/filters/passiveHLxo.html

Basically the crossover goes between the BOSOZ and the Alephs. A first order high-pass filter is already there. It is the BOSOZ output cap and resistor together with the Aleph input resistance. Now it depends what you want. For a start I suggest you resize the coupling cap on the BOSOZ to lift up the corner frequency. Depending on the output and input resistance you most probably end up with a much smaller cap (that now can be of the highest quality without selling your house).

For the low pass filter it's basically the same although R's and C's are switched and you again need (a second or the original) coupling cap on the drain of the BOSOZ gain resistor. You can cascade two RC 1st order filters to get a 2nd order filter. Beyond 2nd order I would rather go for an active filter. To retain the balanced connection between the BOSOZ and the Aleph simply build a filter for each leg of the four 'legs' of your BOSOZ output.

I have done a 1st order passive high-pass filter between my Aleph 1.7 and my balanced Zen V2 and a 3rd order active low-pass between the Aleph and my LCAudio Millenium power-amp in the above way. Works great and I will never go back...

 

[audionut]

I use a passive line level XO between before a BOSOZ and a set of aleph 3 amps. (I have to drive long lines to the amps) The XO's are LC in nature. The L is 500mh...which is hard to find in an inductor. I like the LC arrangement because the transfer function looks more like what you see in the text books vs RC filters. The Problem is Damping out the resonant peak in an LC XO. This means a damping resistor, which means a bigger load for the DAC source in my case.

Years ago I used a Marchand Active crossover, and discovered that passive if far superior to active.

I think the sound is unbelievable, Nelson's simple circuits are the best...Make the passive crossover so that you can plug in the Cs, R's and L's. Then you can experiment with different crossover points, damping, line levels and etc.

 

[Bricolo]

You can easilly do it with a passive XO, if you don't want more than a 2nd order filter.

Since you have a balanced connection, placing a RC network at the + AND - input will form a 2nd order filter.

 

[till]

something like this? http://www.diyaudio.com/forums/showthread.php?s=&threadid=42437

 

[GeorgeBoles]

With thanks

Thank-you all for the responses:

Hattori, I was going to TRY to keep the BOXOZ and Alephs in a versatile form, so that I can use them as "normal" pre-amp and amp with any other speakers which may come my way at some stage in the future. To implement what you describe, it actually renders the equipment useful only for my one set of multi-amped speakers. I guess I could institute your suggestions using another set of inputs into the Aleph which would be specifically for the filtered signal. Would that work? Or am I making things too complicated again?

Hattori and Bricolo, I infer something slightly different between what each of you wrote. Hattori, you say that if I put in a filter at each of the balanced "legs" of my system, I keep it all "balanced", but Bricolo, you go one step further and say that doing so increases the order of the overall filtration, i.e. if I use a simple RC first order filter at each "leg" of the balanced signal I also change the filter configuration from first order to second order. Would either/both of you be able to expand or clarify for me please.

Bricolo, if I put a second order filter on the plus AND the minus signal lines, I would then have a 4th order filter (along the lines of the above paragraph.) I had imagined that the problem with such high order filters is usually signal loss due to impedance. (e.g. 2 cascaded 2nd order Butterworth filters). Hence your reservation about high order passive filtration. Is this as much of a problem if the filters are "spread across" each of the signal lines in a balanced system?

Everyone,

If I implement passive line level XO, in light of the general increase of impedances in the system, should I use a lower input impedance in the Alephs (a 23K design or less such as on aleph 5 or 2, compared with the approx 47K design of the Aleph 30 and 60), or doesn't it matter much ?

Many thanks for all your patience so far.

Regards,
George.

 

[Hattori Hanzo]

I guess I could institute your suggestions using another set of inputs into the Aleph which would be specifically for the filtered signal. Would that work? Or am I making things too complicated again?

Yes you can simply add another set of (unfiltered) outputs (BOSOZ) or inputs (Aleph). If you take my example from the previous reply you could either keep the original output network and use it as an additional unfiltered output or you simply parallel the small and the original decoupling cap and add a switch to the big one so that the cutoff frequency is switchable between e.g. 3Hz and 2500Hz.

Hattori, you say that if I put in a filter at each of the balanced "legs" of my system, I keep it all "balanced", but Bricolo, you go one step further and say that doing so increases the order of the overall filtration, i.e. if I use a simple RC first order filter at each "leg" of the balanced signal I also change the filter configuration from first order to second order.

I don't have a precise and clear explanation but I'm pretty sure that the overall filter characteristics don't change. Of course I'm looking at a balanced, symmetical (-, GND, +) connection and not just at the two phases. Maybe that's the difference.

If I implement passive line level XO, in light of the general increase of impedances in the system, should I use a lower input impedance in the Alephs (a 23K design or less such as on aleph 5 or 2, compared with the approx 47K design of the Aleph 30 and 60), or doesn't it matter much ?

It doesn't matter but keep in mind that the input impedance is part of your filter network and you cannot change it without changing other parts/values.

 

[Bricolo]

That's easy,

Imagine each input of the amp as a buffer. Then, think about the signal path:
it is: source+ -> BOSOZ+ -> insode the BOSOZ -> BOSOZ- -> source-

So, it's like having source->filter->buffer->filter->buffer

2 cascaded 1st (or 2nd) order filter, this results in a 2,d (4th) order filter.

Of course, if you want to go 4th order, it's more difficult, because the 2nd RC cell must be of high impedance compared to the 1st, instead the 1st won't see the load it expects, and the frequency will change.



I hope this answears your question

 

[Hattori Hanzo]

Imagine each input of the amp as a buffer. Then, think about the signal path:
it is: source+ -> BOSOZ+ -> insode the BOSOZ -> BOSOZ- -> source-

So, it's like having source->filter->buffer->filter->buffer

I agree if you only look at the two phases. But you miss the ground path and the fact that the signal swings between - and GND and + and GND and not just between - and +.