What Kind of Crossover is This?

[watchnerd]

I'm looking at pictures of the stock (first) and DIY-modified (second) crossover for an ATC SCM19 (version 1, I think).

I see 2 inductors and 2 capacitors per board, which leads me to think 2nd order.

But there is another extra cap that does I don't know what.

And some mysterious white blocks (resistors?) on the stock board, and green tubes and pink tubes (other resistors?) on the DIY board.

Can any of you analog EE types reverse engineer these circuits?

 

[DonH56]

The white, green, and pink things are power resistors.

Hard to reverse engineer without being able to see all the wires (hidden on the PCB at top, and some are covered by the inductor foil on the bottom picture). Looking at the bottom picture, best I can tell with a quick look, it looks like the input is the YEL wire on the lower right, and the black is ground. There is a series-L then shunt C with another series-RC shunt to the red wire top left that goes to the woofer. The tweeter goes to a series C then a shunt L to ground, then resistor pad to the tweeter. It is not uncommon to have to pad (attenuate, reduce) the signal to the tweeter to get the output response smooth.

My picture is probably not quite correct or complete but looks reasonable for the pictures provided.

HTH - Don

 

[watchnerd]

The white, green, and pink things are power resistors.

Hard to reverse engineer without being able to see all the wires (hidden on the PCB at top, and some are covered by the inductor foil on the bottom picture). Looking at the bottom picture, best I can tell with a quick look, it looks like the input is the YEL wire on the lower right, and the black is ground. There is a series-L then shunt C with another series-RC shunt to the red wire top left that goes to the woofer. The tweeter goes to a series C then a shunt L to ground, then resistor pad to the tweeter. It is not uncommon to have to pad (attenuate, reduce) the signal to the tweeter to get the output response smooth.

My picture is probably not quite correct or complete but looks reasonable for the pictures provided.

HTH - Don

View attachment 8875

Is that a Zobel network across the woofer?

 

[DonH56]

Yes, to help compensate for the woofer's large inductance. Similar circuits are used extensively for matching devices in the RF world as well as for snubbing the voltage spike generated when switching a relay/solenoid coil on and off. I rarely use "Zobel" for this circuit and usually call it a snubber circuit.

Zobel was a Bell Labs guy who developed a class of filter networks based on image theory IIRC. Long before my time, back in the 1920's or 1930's, I think. A Zobel network to me is a bridged-T network, a very clever matching network often used in HF circuits today to "tune out" the capacitance of the big ESD diodes at the input and output pads of HF chips.

Edit -- Found a reference: https://en.wikipedia.org/wiki/Zobel_network and specifically https://en.wikipedia.org/wiki/Zobel_network#Bridged_T_implementation

 

[watchnerd]

Yes, to help compensate for the woofer's large inductance.

I know the answer to this question is probably "it depends", but is such compensation necessary or just a precaution? If the driving amp has low output impedance and high damping factor, is inductance such an issue?

 

[Blumlein 88]

The zobel will help the crossover work at the desired frequencies and slopes. They aren't so much for the amp as to make the crossover work closer to textbook fashion.

 

[watchnerd]

The zobel will help the crossover work at the desired frequencies and slopes. They aren't so much for the amp as to make the crossover work closer to textbook fashion.

Ah, thank you!

So if one wanted to rip out the passive crossover and go externally active DSP crossover / amp, this can be ignored, right?

 

[Blumlein 88]

Ah, thank you!

So if one wanted to rip out the passive crossover and go externally active DSP crossover / amp, this can be ignored, right?

Yes, unless you were using an amp with a high output impedance, then a Zobel would change what the amps sees, but not effect the output of the driver. So it would be useless with DSP and amps directly connected to the driver.

 

[DonH56]

I know the answer to this question is probably "it depends", but is such compensation necessary or just a precaution? If the driving amp has low output impedance and high damping factor, is inductance such an issue?

Answered above by @Blumlein 88 . It acts as a snubber, but also provides impedance-matching to the crossover to present a more constant load to the crossover and thus better control the transfer function. Net result is smoother frequency response.

If the amplifier is directly attached you probably do not need the Zobel but it can make it easier on the amplifier. With a SS amplifier it probably doesn't help much unless the amplifier has stability issues or high'ish output impedance; it may benefit a tube amplifier more due to its higher output impedance. In the past I would sometimes retain the RC in an active design to help reduce charge kick-back from the driver and help the amplifier's stability (sometimes the charge kick can look like wideband impulses).

Note that passive crossovers tend to buffer and reduce impedance variation from drivers and so a passive speaker may actually present an easier load than an active design with amplifiers connected directly to the drivers.

 

[watchnerd]

Answered above by @Blumlein 88 . It acts as a snubber, but also provides impedance-matching to the crossover to present a more constant load to the crossover and thus better control the transfer function. Net result is smoother frequency response.

If the amplifier is directly attached you probably do not need the Zobel but it can make it easier on the amplifier. With a SS amplifier it probably doesn't help much unless the amplifier has stability issues or high'ish output impedance; it may benefit a tube amplifier more due to its higher output impedance. In the past I would sometimes retain the RC in an active design to help reduce charge kick-back from the driver and help the amplifier's stability (sometimes the charge kick can look like wideband impulses).

Note that passive crossovers tend to buffer and reduce impedance variation from drivers and so a passive speaker may actually present an easier load than an active design with amplifiers connected directly to the drivers.

The amp in question, in my case, is a Devialet 400 Pro, dual mono, with an output impedance of .001 ohm.

 

[watchnerd]

An interesting contrast to the relatively simple ATC SCM19 crossover is the one below for the JBL 4367. Interestingly, both are two-ways, but the 4367 is so complex / sophisticated it's on two boards, has battery-coupled capacitors, and clearly has 'a lot going on':

 

[March Audio]

Ah, thank you!

So if one wanted to rip out the passive crossover and go externally active DSP crossover / amp, this can be ignored, right?

As already said, yes. My dsp xo speakers are directly connected to the amps with no adverse effects.

 

[watchnerd]

As already said, yes. My dsp xo speakers are directly connected to the amps with no adverse effects.

What's your recommendation for the best way to measure driver sensitivity to set the levels correctly?

 

[March Audio]

I use an Earthworks M23 measurement mic, but the usb umik mic from minidsp is also works well. Used with REW and Acourate software. Also remember that in room driver measurements are problematic at low frequencies. A good idea to make them in an open space such as your garden.

Excuse me if youv already know this, but remember you will probably need further correction and not just the crossover filtering. Passive xos often have frequency response corrections built in to them, not least of which the baffle step correction.

Can you remind me what you are using to do the xo filtering and its capabilities?

 

[watchnerd]

I use an Earthworks M23 measurement mic, but the usb umik mic from minidsp is also works well. Used with REW and Acourate software. Also remember that in room driver measurements are problematic at low frequencies. A good idea to make them in an open space such as your garden.

Excuse me if youv already know this, but remember you will probably need further correction and not just the crossover filtering. Passive xos often have frequency response corrections built in to them, not least of which the baffle step correction.

Can you remind me what you are using to do the xo filtering and its capabilities?

I have a umik that came with my miniDSP. I assume you're measuring with the drivers mounted in the box/baffle?

What test signal are you using?

 

[watchnerd]

The white, green, and pink things are power resistors.

Hard to reverse engineer without being able to see all the wires (hidden on the PCB at top, and some are covered by the inductor foil on the bottom picture). Looking at the bottom picture, best I can tell with a quick look, it looks like the input is the YEL wire on the lower right, and the black is ground. There is a series-L then shunt C with another series-RC shunt to the red wire top left that goes to the woofer. The tweeter goes to a series C then a shunt L to ground, then resistor pad to the tweeter. It is not uncommon to have to pad (attenuate, reduce) the signal to the tweeter to get the output response smooth.

My picture is probably not quite correct or complete but looks reasonable for the pictures provided.

HTH - Don

View attachment 8875

Below is the schematic for the older, closely related SCM20 (same mid/woofer, but with a Vifa tweeter, instead). I think your photo-derived schematic is pretty close!

 

[March Audio]

I have a umik that came with my miniDSP. I assume you're measuring with the drivers mounted in the box/baffle?

What test signal are you using?

Its swept sine. I take it you havent played with REW? Its free. Yes the speakers are in the box.

https://www.roomeqwizard.com

 

[watchnerd]

Its swept sine. I take it you havent played with REW? Its free. Yes the speakers are in the box.

I haven't used REW. I've used the horrible software that miniDSP provides to configure the box, Roon convolution filters, Pro Tools, and Matlab.

 

[watchnerd]

Its swept sine. I take it you havent played with REW? Its free. Yes the speakers are in the box.

https://www.roomeqwizard.com

To elaborate more on what I said:

Once you make your equation in Matlab and have the library defined, it's easy to implement store in a .yaml file, then implement using:

https://github.com/CPJKU/madmom
https://github.com/LewisJEllis/awesome-lua

You can them implement in alsamixer via its plugin library.

 

[March Audio]

I haven't used REW. I've used the horrible software that miniDSP provides to configure the box, Roon convolution filters, Pro Tools, and Matlab.

ReW is great. Download it and start playing. The umik works directly with it. Make sure you load the calibration file that comes with the mic.

I have played with a lot of diffetent methodolgies before settling on a solution. Including minidsp, but Acourate has been by far the best.