[GeorgeAb]

This is a build thread for the Nelson Pass designed bi-amp (tri-amp in my case) crossover. 

Nelson Pass provided a schematic and later boards with matched field effect transistors (FETs) and bias resistors were provided for the bi-amp xo at the DIY Audio Store. https://diyaudiostore.com/collections/crossovers/products/diy-biamp-6-24-crossover In theory this should sound really good. No feedback, no op-amps, just FETs to maintain signal level. Capacitors and resistors are used to set the crossover point with multiple stages to get steeper slopes of 6/12/18/24 dB/octave. The variable resistors provide the ability to change the crossover points from say 200Hz to 450Hz for example. Nelson describes here https://www.firstwatt.com/pdf/art_diy%20biamp_6-24_%20crossover.pdf



Why: Some folks, me being one, really like the tight control and accurate phasing in the bass and midrange of having an amplifier directly connected to a driver. No inductors, capacitors, and resistors to push through. In my experience (YMMV) even using a digital solution like a Behringer active xo, DBX234 or Minidsp will garner improvement over a passive xo, but as much effort and expense as most have put into sorting our digital front end hard to justify an A/D converter and multiple D/A converters after the DAC.

Caveat: Going with active crossover is no small endeavor. Adding amplifiers, de-integrating your system you are increasing your likelihood of ground loops (hum), expense, and hassle. Getting the crossover points and slopes right for your speaker is no small endeavour. Without tools such as Room EQ Wizard (REW) or having an idea of how your speaker is crossed over it would be difficult if not impossible. I have already been down that road and had a programmable XO Behringer to figure out optimal crossover points and slopes.

Few Analog Active XO Commercially Available: This is kind of a continuation of "why". Not a lot of choice out there. The Pass Labs XVR1 is no longer in production. When in production the unit cost $5K for two channels. Considering you need two for a three channel system it would be $10K. Few come up on the used market and asking prices generally range from $3,500 to $4,500. Phil Marchand of Marchand Electronics makes a nice active XO. The XM-44 for a three channel system with balanced input and output is $1,550. I have one and I really like it. It works well, no quirks, its good kit. It uses operational amplifiers which require feedback, so as good as it is, this DIY solution held promise.

Update: A commercial available discrete active crossover is available from Marchand Electronics. It is manufactured on a custom basis.

When these boards became available, I jumped in ordered the DIY 6-24 crossover boards and started planning. I needed two boards for a three channel system. A highpass section would feed a lowpass section crossed over at a higher frequency to create a bandpass filter section. The schematic uses variable resistors and electrolytic capacitors on input and output sections. I did not want to use variable resistors so to attenuate sections I used a resistor ladder to step down the input in 1dB increments a maximum of -6dB. I used plug ins for the resistors so I could vary the crossover frequency if desired. With that said, I know what works with my speakers. I chose to use polypropylene capacitors throughout signal path.

To determine your crossover frequencies rather than modeling in math cad or some complex program that has a steep learning curve. Mike Rothacher's provided a filter application to determine your values here: http://doublesecretlabs.com/apps/passxo

Listening impressions: I am over the moon with this DIY active XO! What I am hearing is its excellence at what it doesn’t do, no sonic signature or smearing that can be experienced even with well implemented op amps. It performs its function divided audio frequencies to amps without adding or subtracting information from signal. This is high end kit. Just really happy with this active xo.

DIY is usually wonky. There are normally issues that a commercial product will resolve. Things like balanced input so you get common mode rejection, noise that is on both lines is not passed on; only differences between the input and common is passed. For me to get rid of hum from input lines, needed to not use my unshielded interconnects along with an isolation transformer (ISO-Max Jensen). You get a pop of transient if you turn the xo on after your amps are on or off before amps are off.  Nelson addressed this with a circuit that uses relay to shunt output to ground until after a few second delay from power on, and at turn off when voltage starts to decrease output is shunted to ground. In my case this just delayed the transient that would initiate protection circuitry. Not a big deal, I will just turn on amps after turning on xo. When I get around to it, I will engineer a delay for 12V trigger voltage to turn on and off amp. This is something to point out that can be worked through. Just trying to make the point of DIY is a tad wonky. It also gives you an appreciation for manufactures who work through issues before a product comes to market.

Not cheap. The boards were $100, but when all was done likely have well over $700 into this. Then when you add in Iso-Max transformer that can run $250 well you get the idea.  Not that far off from a commercial viable product like the Marchand. The performance per $ however is through the roof... again hi-end kit, discrete components, ... no passive crossover.

Skills. Soldering is fairly straightforward, but it would be nice to have some background in electronics. For example the schematic shows 50K variable resistor for attenuating the input. So 50K input impedence (Z); no, its in parallel so 25K for 2 channel and 16K for three channel. Likely too stiff as 47K input Z is pretty standard. So use 150K if a 3 channel or 100K for 2 channel. Did not see this going in, so had to replace 7 resistors on each of 6 rotary switches, which is a number of hours of re-work. 

Tools and Time to put together. I probably have 60 hours into this, more if you include chasing down ground loops and other issues. Drilling out 7/8” holes for xlr connectors and Dremel for fitment, or making a .5X.75 square hole with drill bits in 4mm thick aluminum for the on off switch takes time with a Dremel and finally hand filing. Definitely need some tools to get this done along with soldering equipment, visor for magnification, etc. Expect to huddled over your kitchen table or work bench for days. Props to the spouse. She came home, I have a full blown electronics lab on our kitchen table! Computer using the signal generator that comes with REW and preamp for amplification to input signal to test attenuation of bandpass filters. She did’t say a word. I’m a lucky guy!

Would I do it again? Definitely. This is in my wheelhouse, I have the skills. As high end an xo as I would like is not commercially available. This is high end, so for me totally worth it. 








 

[mgalusha]

Very nice. I've build several of the Marchand products, they are very nice but I too like the idea of discrete fets and no feedback.

[FullRangeMan]

How it sound in a two ways speaker?

[brj]

Very nice!  And I can't believe that I missed this crossover option entirely!

Can you comment more on how it compares to the Marchand?

I bought a used XVR1 nine years ago (though I either got it for a steal, or prices have gone up since then) and while I love it's flexibility, the noise floor is a bit higher than I'd prefer.  Possibly due to all of the ZTX450 diodes.  (I have plots galore from when I was exploring the effects of Q, etc..)  I've looked at the XM44-2FB for my 2-way system, but would want to have it modded it out of the gate (ahem... @mgalusha!  ), and I haven't had time to dig into it enough to figure out if it might actually improve on the XVR1.

I keep toying with the ideal of a fully differential, discrete, active analog crossover, fixed at the slopes and crossover points upon which I've settled, but the time to figure out the details is beyond what I have available these days.  (I handle the driver correction digitally via an impulse response filter convolved in HQPlayer, so at least my crossover needs are simplified in that it *only* has to handle the frequency splitting between drivers.)  This has definitely made my rainy day reading list - thanks for posting!

[kd4ylq]

GeorgeAB (or any others)
                did you have the chance to evaluate the performance of the active/passive xover designed & built by Dalhquist at any point?

[GeorgeAb]

@brj WRT to comparison I am sticking with my guns and stating that the Marchand is the best available active XO currently commercially available. Super appreciative of Phil and his still providing an active XO, and out of respect do not want to be construed as derogatory to his product. The Marchand was the strongest link and yet the weakest link in my audio chain. The strongest link in providing the ability to control subs, low bass, and mids without anything between the drivers and the amplifiers and the weakest in there are a whole bunch of op amps with resistor capacitor combinations in their feedback loop. It is a solid performer and likely better than any digital solution, I know it is better than the Behringer. The First Watt discrete DIY XO resolves this issue and the audio performance is not night and day, but clearly like the Hippocratic oath does no harm. It is very good, everything comes thru, no smearing around the edges. Hope this addresses your query. 

Great feedback from an XVR1 owner! I always wondered how Nelson was able to get all those capacitor combinations switched in and out of the circuitry. When I look at pictures I see dual in-line packages (DIPs) so I was curious if it used op-amps or was discrete. It sure looks good. Yes, it listed for $4,995 when in production without the external supply.   

@kd4ylg Until you pointed out the Dahlquest was not aware of its existence. It looks like an op-amp with fixed capacitance and varying resistance to change the crossover points. The big issue I see is it limited to a highest frequency of 400 Hz for the high pass. Most high pass on a two way system is going to be substantially higher than that. Therefore, pretty limited in its application. For reference the DIY First Watt can be changed to go from 1.5 Hz on the low pass to -1 dB at 200 KHz on the top so will work in any application.

[Jon L]

I've looked at the XM44-2FB for my 2-way system

XM44-2Discrete by drjlo2, on Flickr

After figuring out the best crossover point/slope for my speakers with Bryston crossover and dBX dsp crossover, I custom ordered the XM44 Discrete version from Marchand.  I requested deletion of volume pots for purest signal path, and for those looking for the cleanest, purest active crossover and do not want to DIY, I highly recommend this approach 

[lokie]

Cool! Thanks for posting.
Would you mind a picture of the back? and maybe even the innards?
XO point etc.., price?

[kmmd]

@brj Great feedback from an XVR1 owner! I always wondered how Nelson was able to get all those capacitor combinations switched in and out of the circuitry. When I look at pictures I see dual in-line packages (DIPs) so I was curious if it used op-amps or was discrete. It sure looks good. Yes, it listed for $4,995 when in production without the external supply.

I had an XVR-1 to bi-amp my Maggie 3.6Rs.  It is discrete without op amps.  They come up for sale about once a year for around $4 - 4.5K.  I sold mine last year to a nice gent close to San Diego for his Apogees.  Are you gents sure that it sold without the external power supply?  I’m not sure how that’s possible, since there’s no room for a power supply in one box.  As for noise I never had any, but it was plugged into my Shunyata Everest power conditioner with Omega XC power cord. 

Photos of my XVR-1 below.  It is more transparent than my modded Marchand XM-44.  Anyone interested in my Marchand?  It’s been sitting around for years.

[Tyson]

With the Pass crossover, how do you address things like baffle step compensation?

[GeorgeAb]

XM44-2Discrete by drjlo2, on Flickr

After figuring out the best crossover point/slope for my speakers with Bryston crossover and dBX dsp crossover, I custom ordered the XM44 Discrete version from Marchand.  I requested deletion of volume pots for purest signal path, and for those looking for the cleanest, purest active crossover and do not want to DIY, I highly recommend this approach 

I did not know a discrete option was available. Good to know there is a commercial option available. I see discrete listed on the website, but I do not recall seeing this a few years ago when I got my XM44. I wonder if this is more recent that Phil is providing discrete option. Man, that looks like nice kit. Phil's pricing has always been fair, curious on cost?

[uncola]

How does the sublime acoustic k321 rate vs these options?

[planet10]

:^)

Mine is starting to be built now. I just need 2-way.

dave

[planet10]

sublime acoustic k321

It is OpAmp based and locks you into 4th order LR. That would be a deal breaker for me (more the latter than the former).

dave

[Jon L]

Cool! Thanks for posting.
Would you mind a picture of the back? and maybe even the innards?
XO point etc.., price?

XM44-2DiscreteB by drjlo2, on Flickr

Opting not to have the 4 volume controls installed, my price was lower.  Single-ended only (Balanced is optional) also lowers price, so my price was actually not much higher than the op-amp version.  You order the unit with specific crossover point and slope you want, but later, you can order additional plug-in cards if you need to change them.

Unless going all-out with Vishay naked resistor in shunt mode, etc, I hear clear degradation in transparency from volume controls, so I use EVS Ultimate attenuators on the higher-gain amp input.  I also use dBX Driverack to measure my room and figure out the compensation needed first, then I use Equalizer APO when using the Marchand x-over.  Room correction, unfortunately, can break or make a system IME.

I believe Marchand only recently started offering these custom discrete units, which have a long lead time but worth the wait 

[brj]

If you know the attenuation needed, I suspect that you could just calculate the needed resistor and add it to the signal path of the appropriate channel.  (If you're switching amps on a semi-regular basis, then an adjustable solution like the in-line attenuators would clearly provide the needed flexibility.  In my case, my amps channels are all identical, but my tweeter is much more sensitive than the woofer of my 2-ways.  Since I know this delta exactly, having the appropriate knock-down resistor on the tweeter side of the crossover would likely be best.  Though making it removable might be smart just in case I do need to tweak it a bit.)

With the Pass crossover, how do you address things like baffle step compensation?

I can't speak for others, but I handle everything other than the actual frequency splitting in the digital domain before the DAC.  I used Room EQ Wizard to create a 32-bit / 192 kHz impulse response filter implementing all of the driver EQ specified by the designer of my speakers.  HQ Player then convolves this IR file with the music stream in real-time as the final step performed before leaving the computer for my USB connected DAC.  So any BSC would be easily accommodated in such a setup.

[arteom]

Have ordered the essentials kit from diyaudio store. Working now to get a cart put together with all other parts needed. The only bit I'm not 100% certain on is the value for C.

I'm hoping to put this to use in a system with a pair of Tannoy LGM full ranges. Planning to put together a pair of Bucket Subs to fill in the low end for those. The Tannoys get a Pass F6 clone, the subs fed from Class D Audio amp.

[planet10]

This should be helpful with that: http://doublesecretlabs.com/apps/passxo/

dave

[arteom]

Thanks, ended up going with .15uf and .075uf, also ordered some sip sockets planning to use for C and C/2. This might come together this weekend. Need to get going on the bucket subs, have the drivers in hand, need to pickup and prep the buckets!

[arteom]

Completed the Pass 6-24 Crossover. Socketed the filter capacitors. Right now the Tannoy's are playing with the 8" drivers on my Mirage M3's. Not blending all too well, need to figure out a system to get it all tuned in. Though the end goal is not to use the Mirage, but get a pair of bucket subs built for that purpose. Still working on getting the parts together for those. Would like to avoid the blue or orange branded buckets from the big box stores. Was hoping to find something else locally, but not having the best of luck with that.