Rerurn to Romy the Cat's Site

Audio Discussions
Topic: Thought process behind crossover estimating, testing and execution

Page 1 of 1 (7 items)


Posted by Romy the Cat on 09-24-2010
fiogf49gjkf0d

I was reading the Stefano Bertoncello's post about his new multi-amping again options.

http://twogoodears.blogspot.com/2010/09/multi-amping-again.html

.. and I feel that it would be a good idea to brief summarize multi-amping crossovering. Being myself multi-amping advocate and seeing a lot of multi-amping advantages I still feel that multi-amping crossovering is not well understood.

Since the multi-amping crossovering need to be implemented before amplification, or what we call at line-level then there are 2 main options: use active devise and use passive devises.  I would like to stress the difference the difference between active and passive crossovering.

Active crossovering imply use of active gain elements (tube or solid state) and the crossover curve is written around the aliments with active gain. This has a lot of advantages: might allow regulation of gain for each crossover channel, eliminate any input and output impedances variations, might even permit to wary the order or Q of the filter, stabilize the filter with respect of load. However, the active crossovering also has just one minor disadvantages - it put a collection of active stages into you signal path. Do you remember how much you suffered until you found a preamp that is neutral and transparent enough? There are plenty of preamps that cost $10K-$40K and that claim some sort of neutrality. The requirements to active crossovers are absolutely identical as the requirements to preamps. I would say that the requirements to active crossovers even higher than the requirements to preamps. Some people use passive inductive preamps or one stage buffer preamps and fell that only they can do “transparency”. However, active crossovers almost always imply more than one active stage – to far from minimalistic approach of a simple buffer.

For any active crossover, or a phonostage at this mater, it always was good test to disconnect the filtering part, to set the unit in unity gain to perform the “insertion” or “bypass” test.

http://www.romythecat.com/Forums/ShowPost.aspx?PostID=2589

Only if a unit demonstrated transparency being “inserted” then the filter might be introduced (it might be a bit complicated if filets are written in feedback). My experience with line level electronics suggests that it is VERY difficult, practically imposable to have line level active devise with gain to be “inserted” and to be absolutely transparent. Still, many people run active crossover with multiple active gain stages and do not realize that their active crossover is also acting a very pure sounding line level preamps. This way I have a stand against active crossovers and feel that additional active stages might be used only if a crossover needs to be higher than second order.

Now we enter the second option of line-level crossovering – the passive filters. There are two types of line-level passive crossovering: out-of-devise and in-devise. Out-of-devise implies that a filter located between line-level components.  This is very effective filters that do minimal damans to sound but they practically always are very fixed and in many ways they are tunes to the output and input impedance of line-level units. The out-of-devise passive filters have a disadvantage that they are environmentally dependant and in some ways even change of interconnects might hugely affect the performance of the filter. To cure the problem of environmental dependency exist the “in-devise passive crossovering”.

In-device passive crossovering are the same passive filters but they are imbedded into line level or power –level electronics. Since Multi-amping implies multiple amplifiers then it always there is a huge amount of locations where passive crossovering filters might be implemented. The sonic damage of in-devise passive filter is absolutely lessen of any topology and in very many cases the in-devise passive filter in fact make sound even better. A very simple example – you need to implement a high pass filter. In case of out-of-device passive crossovering you run a capacitor in series with your power amp. However, your power amp has 2-3 stages with capacitive couponing and by you might use one of the couponing caps as your filter. This way you not only eliminate one extra capacitor from a signal path but will be able to use the remaining capacitor of much less value, which most likely would lead to a better quality of capacitors.  Sure, the examples might be more complicated but the key is in following: the in-devise passive crossovering is a crossovering topology that permits an active a devise to perform it’s main duty while using the active stages of the devise to write the necessary filters. By definition the in-device passive crossovering imply no damage to signal as in-devise passive crossovering is always subtractive.

I hope the folks who use multi-amping review their playback configurations under the light of this article and see how and if they can take own crossovering a bit further in the “interesting realms”…

Rgs, Romy the Cat

Posted by scooter on 09-25-2010
fiogf49gjkf0d
Very helpful primer. This might be a good place to elaborate a bit on approaching the task of crossover selection and testing for each driver, especially with respect to points, filter selection, steepness and other parameters one should consider. From a practical point of view, does it make sense to estimate crossover parameters for each driver, then temporarily use something like the Behringer DCX (or some computer-based crossover software and appropriate multichannel card) for testing and optimization, before execution? Some color and motivations behind the thought process and potential stumbling blocks would be useful. 

Posted by Romy the Cat on 09-25-2010
fiogf49gjkf0d
Of course most of the people do prototyping the type of crossover they need for a given channel using active crossovers.  As the crossover point, slop and in some cases Q are found then the same data is applied to a given topology of line level in-device filter. Still, my experience indicates that it is very frequently there is need to fine-adjust a bit the in-device filter. The in-device, or imbedded filter, are always cleaner and they also make a channel more transperent then with use of active crossovers.

The caT

Posted by skushino on 06-11-2012
fiogf49gjkf0d
I'm working on a project that may be of interest to others, and thought posting to this old thread was a better way than beginning a new post.

I use speaker level passive crossovers, between multi-amps (not DSET) and drivers.  My desire is to move the filters from speaker level to line level, and to move them inside the amp. The current speaker level crossovers are first order, caps or coils, depending on application, plus a voltage divider to attenuate the MF.  There seem to be two methods to apply filters inside the power amp, 1) locate the filters at the amp input, before the gain stages, and 2) locate the filters in between gain stages.  If I understand correctly, the advantage of option 1 is simplicity (at least for me) - I can calculate filter component values based on the measured or published input impedance of the power amp.  The advantage of option 2 is the opportunity to use less components by combining coupling and filter capacitor values.

Is this understanding correct?

For an example of option 1, if I wanted to add a 70hz high pass filter in the ML2 with a 41k input impedance, then a cap of .0568uf is all that's required, correct?

If I want the filter located in-between gain stages, for the benefit of reducing component count, I have no idea how to proceed.  Wouldn't this require a circuit schematic?

For those who have experience, am I on the right track, or over complicating?  I like the idea of going with option 2 in the interest of having one less cap in the signal path, but I only feel confident working with option 1 now.

Thanks for any clarifying replies.



Posted by Romy the Cat on 06-11-2012
fiogf49gjkf0d
Skushino, from what you say I presume you use two pair of ML2, right? Yes, what you say is correct but if I were you then I would not introduce an addition caps to signal path. In my view the best scenario for high path would be to minimize the coupling cap between the stages. Figure out the stage impedance and calculate the new cap that would roll the whole amp at 70hz. It is very simple to attach it to the existing cap – juts clip one lead of the current cap and parallel it with the new smaller cap. For the low path you might go for coil but it will be large and you need to shield it. It is much better to go for cap to ground, sort of subtraction filter. An alternative would be to run the filter in the ML2’s feedback. Lamm run global feedback to the cathode of the input stage. If you contra-filter the feedback then you can get the same desirable result.

Posted by Jorge on 06-12-2012
fiogf49gjkf0d
I think ML amps have an input capacitor put there to isolate from the other components, I have a close friend who has multiamped for a long time with ML 20.6 monos,  he had once 6 monos on his system,  he now uses only one pair with the input cap reduced to work from 80 hz up and a sub amp.

Lamm run global feedback ? really?

Posted by Romy the Cat on 06-12-2012
fiogf49gjkf0d
I do not think that Lamm uses input capacitor, come on, this is not am amp from Radio Shack!  Anyhow, I would not use input capacitor under any circumstances, would it be for filtration or for any other reasons. Yes, ML2 used global feedback.

Page 1 of 1 (7 items)