Passive high pass filter - Marchand XM446XLR-A


So, I am about to set up a new pair of monitor speakers and fill in the lower frequencies with two subs.  The monitors are Aerial LR5s and the subs will be two Aerial SW12s with internal 400 wpc amps.  The LR5s weigh 105 pounds each and play flat to just a bit above 40Hz (-6dB at 35Hz).  The Aerial subs are fast, powerful and musical and should integrate well with the LR5s.  My pre and power amps are solid state, fully balanced and the power amps provide 300/600 wpc into 8/4 ohms.  I will be running separate ICs from the preamp to my power amps and a second set from the preamp to the input on the subs.

My first question is whether I would benefit from using a high pass filter between the preamp and amplifiers to limit the low frequencies going to the LR5s so the amps are not trying to drive them full range.  

I didn't find a DIY design that I was comfortable with but I did run across several posts by people using a Marchand passive high pass filter.  The Marchand XM446XLR-A is fully balanced and has a high pass corner frequency of 80Hz (i.e., the frequency that is down by 3dB).  The Marchand has a standard slope of 24dB/octave but can be made for other slopes.

My second question is whether anyone out there has experience with the Marchand passive high pass filter and is it sufficiently transparent or should I expect to hear unwanted artifacts with the filter in my system?  

Finally, is there a higher quality alternative or should I simply play with the low pass on the sub and forget about filtering the signal to the main speakers?
http://www.marchandelec.com/xm46.html
mitch2
drop me a note if you'd like to experiment - i have a marchand that didn't leave when i sold my subs.
The Marchand crossovers are very well regarded in the hardcore DIY community. The maker can supply the owner replaceable "cards" to insert in the x/o, providing just about any x/o frequency and slope one desires. And most are available with balanced XLR jacks, unlike the similarly-priced First Watt B4, another fine unit. The B4, unlike the Marchands, contains no Opamps, achieving filtering via all discrete parts. The B4 also needs no extra-cost filter cards, having internal dip-switches to set frequency and slope characteristics.
Thanks guys for the responses so far.  bdp24, I believe you are talking about Marchand's active crossovers but I am not interested in putting another active component in the signal chain.  Thank you for referencing the First Watt B4.  Even though it is an active device, it was still interesting to read the manuals by Pass.  It would be interesting ot me if he would offer a passive kit that the user could stuff with their favorite caps and resistors, and to provide some technical information on component compatibility and matching subs to the main speakers.

I have not yet read anything saying the Marchand passive high pass filter is exceptional, or does nothing unwanted to the signal.  Instead, what I have mostly read is that some prefer the Marchand high pass to doing nothing because they believe reducing the demand for the main amplifier to drive low frequencies improves clarity in the upper frequency ranges.  The flavor of most of the posts seems to be that one way or the other is a lesser of two evils.

I am really interested in hearing from folks with a high end and resolving system who are using monitor type main speakers mated with subs, and the method they have found to best cross over the subs to the mains.

One advantage I have is that my subs and main speakers come from the same manufacturer and are made to work together.  There is an option to use a crossover within the subs to high pass the main signal so that is something else I can at least try, particularly if I position the subs near the main speakers so the second IC from the subs to the mains would be short.

Now that I have finally received the main speakers, second sub and stands for everything, I will contact the manufacturer to verify their recommendation for the best way to integrate the high and low frequencies.  Would still appreciate hearing about success stories.

Ah, okay, it's passive you want. The purist way to high-pass filter the signal going to your power amp and speakers, to relieve them of reproducing bass frequencies (a worthwhile objective imo), one that has been used since the DIY 1950's, is to install a single capacitor (the value of which is determined by 1- your power amp's input impedance, and 2- the desired x/o frequency) on the inside of your power amp, on it's input jacks. This will create only a 1st-order, 6dB/octave filter. That is the way to incur the least damage to the signal, both by the use of but a single part---a capacitor, and by removing the need for an extra inter-connect---no x/o box necessary. The formula for determining the capacitor value required for your application can be found all over the 'net.

What I have heard from other Magneplanar Tympani owners on the Planar Speaker Asylum, many of whom use Marchand active x/o's in their systems, is that the active Marchands are actually better sounding than the passive ones. How can that be, you ask? Aren't passive parts more transparent than an active circuit? Look at the parts inside those passive Marchands---the parts necessary to provide a higher-than-1st-order filter. Nasty, ugly sounding things, worse than the parts in an active circuit.

Speaking of ugly sounding parts---all Marchands contain Opamps (except the tube model, one presumes). The First Watt B4 is completely discrete---all resistors and capacitors, no Opamps, no IC's. The B4 provides 1st-2nd-3rd-4th order (6-12-18-24 dB/octave) filters from 25Hz to 3200 Hz, in 25 Hz increments. Incredible! Unfortunately, it is available unbalanced/RCA jacks only, so may be unsuitable for your system.

Thanks bdp24, I have calculated the capacitor size needed but everything I have read seems to be for single-ended signals.  If I want to go the single capacitor route with my balanced amp, do you know what I would do differently...i.e., do I put a series cap on both of the differential signal legs?  Is the calculation different for a balanced signal?
Great question Mitch, one I need the answer to as well! I've done it on a single-ended amp (my Bedini 25/25, for use with Quad 57's and subs), but not a balanced amp. Hopefully someone with more technical knowledge than I can help us out. I have a Music Reference RM-200 Mk.2 waiting for the answer! It has balanced/XLR inputs only. I'll be shooting Roger Modjeski an email, so I'll let you know what he says.
Gentlemen, yes, you would need a capacitor on each of the two signal lines in the balanced signal pair. Preferably the two capacitors should have a reasonably tight +/- tolerance on their value (for example 5%), so that they match reasonably closely, although I doubt that the tolerance is particularly critical.

Usually, although not always, the specified input impedance of a balanced input reflects the sum of the input impedances between each of the two signal lines and ground. That appears to be the case for the RM-200 Mk2, since the spec is 30K balanced and 15K unbalanced, the unbalanced spec presumably reflecting the impedance that would be seen if a single-ended signal were provided to the XLR connector. Stereophile measured the actual balanced impedance as 29K, so you would calculate the capacitor value based on a 14.5K input impedance.

Regarding Mitch2’s Clayton amps, you would want to verify that the specified 100K input impedance reflects the sum of the input impedances of the two legs. If so, the value of each of the two capacitors would be calculated based on a 50K input impedance.

Good luck. Best regards,
-- Al

Great, thanks Al---Eric.
Thanks much Al!
The Claytons only have balanced inputs so I am confident the 100K input impedance is for the balanced condition.
Good point on matching the caps.  I plan to try something relatively high quality like Jupiter copper film caps.
Partsconnexion will match them for $1 per pair.
I plan to try 0.022uf caps for a cut-off frequency of about 72 Hz and then start by low-passing the subs at 80 Hz.
The next cap size up (0.047uf) gives a cut-off frequency that is a little low at 34Hz but may still be an option.
If I wanted something in-between, could I series the 0.022 with a 0.01 cap?
Caps are thankfully 25% off this month, which is good since I need four of them.
Tim (Mitch2), to be sure it’s clear, I was saying that in most cases manufacturers specify balanced input impedances as the sum of the input impedances of the two balanced signal legs. I have seen a few exceptions, however, in which the specification corresponds to the input impedance of each leg. The fact that an unbalanced input is not provided has no relevance to that.

If the 100K spec on your amp is defined in the usual manner, you would want to base your calculation for each of the two capacitors that are required (per channel) on a 50K input impedance, as I indicated earlier. The 0.022 uf cap would then result in 144 Hz, not 72 Hz. And the 0.047 uf cap would result in 68 Hz, not 34 Hz.

In the much less common situation where the input impedance is defined on a per leg basis, meaning that each leg is 100K, your calculations would be correct.
If I wanted something in-between, could I series the 0.022 with a 0.01 cap?
No, you would want to parallel them. The values of capacitors in parallel add. The value of two capacitors in series corresponds to their product (multiplication) divided by their sum, which will always be less than the lower of the two individual values. It’s the opposite of what happens when resistors or inductors are connected in series or in parallel.

Paralleling two resistors or two inductors or two capacitors results in a lower impedance than either of the two paralleled parts would present individually, at a given frequency. Since the impedance of a capacitor at a given frequency is inversely proportional to its capacitance, to have a lower impedance the capacitance of the paralleled capacitors must be greater than the capacitance of each of the two. Again, it’s the opposite for resistors and inductors. The impedance of a resistor is identical to its resistance (to the extent that it behaves in a theoretically ideal manner), and the impedance of an inductor is directly proportional to its inductance, rather than inversely proportional.

Best regards,
-- Al

Dang, my momma was right when she told me I shoulda finished high school.
Thanks again Al.  It seems the 0.047uf caps will be a good start.  If the cut-off is at 68Hz then that will actually be perfect and if it is at 34Hz then it still relieves the amplifier from trying to drive the lowest frequencies.  Another thing I could do is to try some less expensive caps first to determine the correct value, and then later move up to the caps I want to use.

I also appreciate the analysis and technical information because one of my favorite things about this site is learning about stuff I don't fully understand.

In summary, there doesn't seem to be anything close to a universal answer to this situation.  I have decided not to pursue a Marchand filter at this time because I don't want something in the path that may negatively affect the signal going to my main speakers.  The subs do have a (reportedly) high quality on-board high pass filter so I can give that a try.  The single cap in-line interests me and I may give that a try.  If it works well, I could actually wire them inside of my preamp, or make a pair of balanced cables with the capacitors in-line.  From what I have read, it doesn't seem to matter where the caps are placed as long as they are between the preamp output and the amplifier input.  Some put them inside the input of the amplifiers.

Thanks for everyone for the comments and help.
Sounds like a plan, Tim!  Best of luck.

-- Al
 
Post removed 
For owners of Music Reference RM-200 power amps, a bit of clarification: Al's input impedance figure for the amp is correct for only the original incarnation of the amp---30kHz balanced, 15k per leg. In the Mk.2 (current) version of the amp, RM-200 designer Roger Modjeski increased the figure to 60kHz, 30k per leg. That's better for use with some tube pre-amps, which often have somewhat higher output impedances than most solid state pre's.
Thanks Eric (bdp24). I’m wondering, though, if the change you referred to from 15K/30K to 30K/60K might have been introduced sometime after the Mk.II was initially marketed.

The Stereophile review I referred to indicates in numerous places that it is a review of the Mk.II version, and describes various changes and improvements that were made relative to the original version. The photo of the amp on the first page of the review also shows "RM-200 Mk.II" as being marked on its panel. As I mentioned, that review indicates a measured balanced input impedance of 29K (JA’s measurements of balanced input impedances always represent the sum of the two legs), and also indicates a spec of 15K ("one leg driven")/30K (balanced).

Also, Roger’s website still indicates 15K/30K, although he has not updated the amp's description to say "Mk.II."

Best regards,
-- Al

Hmm Al, I'll have to get it straight from Roger. I got the impedance figure from my Mk.2 owners manual.