Marchand XM9 analog electronic crossover (short of) review

[Sokel]

This is short of a review for the very old Marchand XM9 2-way crossover boards.
It's a 30 years old design,probably discontinued and I got the unpopulated boards from a friend that kept them in a drawer with no use of them.

Board looks like this:

I used some different components,as LME 49720 op-amps which I already had instead of the shown OPA2134PA,PRP low ppm resistors as they set the x-over point (240Hz),etc.
Boards were cleaned and heated so to get the moist of all these years off and working with it was a joy as it's fairly high quality PCB.

I have to admit that starting this review was sure enough the my old interface was good enough (<10dB THD+N as needed) for such an old design,but...
So see it as a fun thread as a better interface or measuring rig seems to be needed for this.
Measuring an x-over has it's oddities but I tried to get a good picture of it despite of that.

What I wanted to compare more was the low's performance as it seems to be a challenge for both analog and digital gear as we go lower,and some of them are really bad at near 20Hz or so.

Test conditions
E-MU 0204 I/O (ADC+DAC) at 96kHz for all measurements
All connections are unbalanced.
Multitone Analyzer for all measurements and REW cross-check for sanity.
Both Low and Mid-High ways measured separately,x-over set at 240Hz and all the measurements are with this filter in place.

Classic SINAD at 1kHz :



It lands nicely at 95.4dB with fairly low noise.

SINAD at 80Hz,100Hz and a bonus 20Hz (danger zone) one:




That's where the good news are coming,both low distortion and noise even down to 20Hz.

Frequency response both low and high



Slopes are visible following the filter.

Crosstalk low and high:



Dual mono so it would be good any way.

Multitone32 for both lows and highs:



CCIF 18.6kHz/19.5kHz IMD test:



Nice and clean all the way down.

Dynamic Range:



Up to 105dB or so.

THD+N vs Level (also broken down to its elements) for both lows (80Hz) and highs (1kHz) :



Really nice and linear

.... and a cross-check with REW:



Nothing different.

I admit it was fun both soldering it and measuring it.Nice (all analog) performance is really nice,and clears CD quality at all tests.
I understand that there are ready to use ones with the same boards (or updated ones) at the second hand market coming straight from Marchand,also balanced ones,etc for fair prices.

Nonetheless,I see very limited use cases these days for such as this,digital ones are way more handy and easy to play with and good ones has very good performance high and about the same with this down low.
The advantage of Marchand though is set and forget,is of enough quality to last a century without hiccups,accidental full-blast incidents,etc.

Hope you enjoyed it!

 

[kmanusa]

Sorry for the late response but thanks for the excellent review!! I bought one of these about 20 years ago and have used it to integrate passive subs with main speakers such as Magnepan MG-III, Quad 988, and currently Martin Logan CLS. I have wondered about the performance of the Marchand gear given Amir's largely negative review of the XM-44 back in 2020. But from your test results, the XM-9 appears to be a solid performer

Obviously, DSP-based units can be more versatile and better performing than this analog state-variable design but require some time and effort to optimize. Other than the need to use the Marchand "frequency modules" to set crossover points, the unit is easy to use and has allowed me to make a decent integration of the CLS with Kinergetics SW-800 passive subs without much effort and fiddling. I much prefer to use the XM-9 to the Kinergetics electronic crossover which applies a substantial low frequency boost to the subs. In my experience with using the SW-800 in several different rooms, this low frequency boost is too extreme and largely unnecessary.

A while back I acquired a dbx 223xs crossover (before Amir's review) and it is interesting to compare the dbx with the Marchand since they both use Linkwitz Riley 24 dB/octave slopes. I like the dbx front panel control of crossover point and wide range of level controls for both low pass and high pass channels. The Marchand's single ended I/O is a bit better suited to my mostly singled ended system components-- as compared with the balanced I/O of the dbx-- with no apparent increase in noise. I also like the use of IC sockets in the Marchand that have allowed me to easily swap out the original LM353 ICs with OPA2134s. Finally, simple mods for high input impedance and baffle step compensation are shown in the XM-9 manual along with instructions on how to swap in OPA2134, AD712JN, or OP275GP ICs for the LM353s.

 

[Mr. Widget]

This is short of a review for the very old Marchand XM9 2-way crossover boards.
It's a 30 years old design, probably discontinued and I got the unpopulated boards from a friend that kept them in a drawer with no use of them.

Just saw this today... took me way back.
Out of curiosity, I checked. Marchand has an active website and these boards are listed as still available at $150 ea.

True that digital potentially offers much more control and if done well it will offer even higher performance, but it is nice to see affordable quality analog still alive and well.

 

[Sokel]

Sorry for the late response but thanks for the excellent review!! I bought one of these about 20 years ago and have used it to integrate passive subs with main speakers such as Magnepan MG-III, Quad 988, and currently Martin Logan CLS. I have wondered about the performance of the Marchand gear given Amir's largely negative review of the XM-44 back in 2020. But from your test results, the XM-9 appears to be a solid performer

Obviously, DSP-based units can be more versatile and better performing than this analog state-variable design but require some time and effort to optimize. Other than the need to use the Marchand "frequency modules" to set crossover points, the unit is easy to use and has allowed me to make a decent integration of the CLS with Kinergetics SW-800 passive subs without much effort and fiddling. I much prefer to use the XM-9 to the Kinergetics electronic crossover which applies a substantial low frequency boost to the subs. In my experience with using the SW-800 in several different rooms, this low frequency boost is too extreme and largely unnecessary.

A while back I acquired a dbx 226xs crossover (before Amir's review) and it is interesting to compare the dbx with the Marchand since they both use Linkwitz Riley 24 dB/octave slopes. I like the dbx front panel control of crossover point and wide range of level controls for both low pass and high pass channels. The Marchand's single ended I/O is a bit better suited to my mostly singled ended system components-- as compared with the balanced I/O of the dbx-- with no apparent increase in noise. I also like the use of IC sockets in the Marchand that have allowed me to easily swap out the original LM353 ICs with OPA2134s. Finally, simple mods for high input impedance and baffle step compensation are shown in the XM-9 manual along with instructions on how to swap in OPA2134, AD712JN, or OP275GP ICs for the LM353s.

Just saw this today... took me way back.
Out of curiosity, I checked. Marchand has an active website and these boards are listed as still available at $150 ea.

True that digital potentially offers much more control and if done well it will offer even higher performance, but it is nice to see affordable quality analog still alive and well.

One can tell the quality of the board and the skill of the designer only by populating the boards.

Every bit of performance and every combination of the available options which are no few for a board like that:
slopes (that needs cutting some -marked- traces and some changes)
x-over frequency (actually 4 resistors of the same value on a IC socket)
BSC (addition of a simple RC)
balanced or SE inputs (both are standard and available just with a cut of a trace) and balanced outputs (with the addition of some components)
stereo or summed to mono bass
high input impedance (that requires some changed resistrors,etc) from the standard 25kOhm to 100kOhm (yes,that kind of immunity at the cost of 2dB noise)

Only a seasoned,real world engineer can squeeze so much options and performance at such a small board with through-hole components which by itself is of very high quality,a joy to work with and at a very decent price as I read above.

I expect even nicer performance with a +-15 switching quality PSU as of the meanwell shorts or even better with shunt regulators,etc. instead of my silly DIY +-15 LM78/7915 which powered it for this (short of) review.

Comparison with the shorts of dbx 226xs can show plus and minus with the main disadvantage of the pro interface being the (easy but) highly inaccurate pots.
Marchand can be populated with multiturn pots so the agreement between channels,etc can be at the sub-mV range.
Cheaper ones are just horrible i have some measurements of one of them here .

I'll say it again,the ease of using something like this has no equal,peace of mind as well.
The ONLY thing one has to make sure is to have the freq module tightly put in place or else some serious DC can output but that's trivial.

 

[kmanusa]

Agreed, the pots on the dbx are disappointing especially those used for setting the crossover points. Thanks Sokel for your insightful analysis of the Marchand!

 

[Trdat]

I have moved onto digital Tri-amped DSP, however for the simplicity of integrating subs and of course for the beginner nothing beats a analogue crossover, mine a XM66 and a phase aligned software DSP to tie it in all together. Although, now there are the minidsp options.

 

[Sokel]

In case any of you wonder how it measures with the stock op-amps OPA2134 (I found some at the depths of the famous drawer) that's how it looks like (directly comparable with the above 80Hz one) :

Couple of dB worst THD,a dB worst SINAD .
Nothing to pay attention too,perfectly fine with those too.

 

[Rewind]

Nice review, in contrast to Amirm’s review of the XM44 where he neglected to mention that it was a kit faulty assembled by an amateur.

 

[Sokel]

Nice review, in contrast to Amirm’s review of the XM44 where he neglected to mention that it was a kit faulty assembled by an amateur.

If I was to guess, something went really wrong with the input and grounding scheme at XM44.

It seems easy to assemble one but it's not and I wouldn't advise it without the right tools to verify the build.
Keeping mains hygiene failed as well, despite the effort of short of isolating the PS.

I was really surprised to see it performing so bad but now that I know about the assembly makes scene.

The one of this thread simply maxed out my ancient silly interface's performance with ease, in reality is better than what the charts show.

 

[Rewind]

If I was to guess, something went really wrong with the input and grounding scheme at XM44.

It seems easy to assemble one but it's not and I wouldn't advise it without the right tools to verify the build.
Keeping mains hygiene failed as well, despite the effort of short of isolating the PS.

I was really surprised to see it performing so bad but now that I know about the assembly makes scene.

The one of this thread simply maxed out my ancient silly interface's performance with ease, in reality is better than what the charts show.

It is shame, because for years the review persuaded me not to buy a Marchand analog crossover.

 

[Sokel]

It is shame, because for years the review persuaded me not to buy a Marchand analog crossover.

Crossovers, no matter analog or digital are a pain to measure against their intended use.
That is, with filters in place, optimal gain and levels, FR (specially low) , etc.

They're also tricky to use as an overall gain-staging scheme has to be designed right from the start, the machine alone involves a bunch of different levels and their sums (also adjustable independently) at the crossover points.

At least this one keeps things sane with high (25 kOhm) or very high (100 kOhm) input impedance, low (less than 100 Ohm if I remember right) output impedance, high voltage output on demand and so on.

As far as I have searched when I measured this, Marchand is solid and with good reputation at US, so...

 

[Rewind]

Crossovers, no matter analog or digital are a pain to measure against their intended use.
That is, with filters in place, optimal gain and levels, FR (specially low) , etc.

They're also tricky to use as an overall gain-staging scheme has to be designed right from the start, the machine alone involves a bunch of different levels and their sums (also adjustable independently) at the crossover points.

At least this one keeps things sane with high (25 kOhm) or very high (100 kOhm) input impedance, low (less than 100 Ohm if I remember right) output impedance, high voltage output at demand and so on.

As far as I have searched when I measured this, Marchand is solid and with good reputation at US, so...

I can buy an older XM-9 KIT - uh-oh - locally in Norway. It is to be used with very sensitive horns, so I might notice its age, and might benefit from a newer model. But my amps are what ASR would describe as 2nd harmonic distortion generators.
The lack of delay, compared to digital models like the newer Minidsp, is also an issue. Vituixcad simulates large cancellations if the physical time alignment between speaker drivers is off. Above 500Hz everything is physically time aligned, but the woofers take too much space.
Maybe I will just use a nice DAC for mids and highs, and convert everything back to digital below 500Hz, to take care of business with digital delays.

 

[Sokel]

I can buy an older XM-9 KIT - uh-oh - locally in Norway. It is to be used with very sensitive horns, so I might notice its age, and might benefit from a newer model. But my amps are what ASR would describe as 2nd harmonic distortion generators.
The lack of delay, compared to digital models like the newer Minidsp, is also an issue. Vituixcad simumlates large cancellations if the physical time alignment between speaker drivers is off. Above 500Hz everything is time aligned, but the woofers take too much space.
Maybe I will just use a nice DAC for mids and highs, and convert everything back to digital below 500Hz, to take care of business with digital delays.

If the boards have been populated with the usual top-notch quality components as usual, all it needs is a nice bath with contact spay (without oil! ) , leave them suck overnight, dry them with hot air (not too hot! ), clean them and that's it.

Delays is an issue, yes. You could involve some analog all-pass filter to it, if delay is small and you know exactly what you want, I would mail Marchand about it.
But your other scheme can work fine as well.

 

[Rewind]

Alpass has a to be very precise, and only works in a narrow frequency band. But my idea with analog above 500Hz, and digital bass, does not work, since the bass will be delayed because of the digital processing, and will never be able to startnplaying before the mid and treble horns, which start playing earlier since the are closer to the wall.

It is either or.
1.Full analog with possibility with a nice DAC
or
2. Full retard and enjoy the digital shimmer of a DSP on all channels.

Maybe I will get away with it, without allpass or delay! Not likely.

 

[Rewind]

According to Marchand, in my ”hybrid” horn speaker with the woofers 60cm offset of the horn speaker drivers, allpass will not work. At 500Hz with 2ms delay, it will work great, but to continue to delay up to 20kHz will sound muddy.

So going full digital. I see only the Minidsp Flex as a viable option with a modern DAC chip. T.racks 408 FIR has one with the older/ancient PCM1794a. Not sure which one will sound best.
I wish there was a hifi tape delay with 1-2 ms delay, that did not degrade the sound.

 

[john61ct]

Maybe not as pristine, but AudioControl active non-DSP give the "xover freq module" approach, let you select LR4 or gentler, include anti-rumble HPF for subs

I realize also old tech and tainted by association with the car audio world :-0

but in that context seem pretty solid rep for clean SQ and apparently good at avoiding phase issues?

 

[Rewind]

Maybe not as pristine, but AudioControl active non-DSP give the "xover freq module" approach, let you select LR4 or gentler, include anti-rumble HPF for subs

I realize also old tech and tainted by association with the car audio world :-0

but in that context seem pretty solid rep for clean SQ and apparently good at avoiding phase issues?

Does it have delay?

 

[john61ct]

not as far as I know, certainly analog only which is what I want to start with

If that aspect turns out to be significant and unsolvable by placement, that is the point I look at DSP

 

[Rewind]

not as far as I know, certainly analog only which is what I want to start with

If that aspect turns out to be significant and unsolvable by placement, that is the point I look at DSP

I won’t know how signifcant until I A/B with less than 10 second pause.

 

[john61ct]

Or REW gives big delay #s