Have been working on a diy 6-24/FR EQ combo and reading and watching videos on xovers, reading the Douglas Self book on xo design, etc., but having some intrusive thoughts, lol. I don't have any formal background in electronics, so may be missing the obvious so thought I'd ask here.
In my limited research, I never ran across the anyone asking if a passive crossover circuit board could be used in an active stage (for a commercial speaker it was designed for and integrated and sold with, of course) . I'm assuming there might be buffering required at inputs and outputs to account for impedance differences, or need to have a strong drive due to insertion losses or larger components, perhaps more noise, idk, but I'm wondering if it's theoretically possible.
If so, I'm assuming there must be caveats that keep doing something like or a scaled version this off the table as a recommended approach. (other than opportunity loss for other active xo benefits like custom/complex filtering via circuits or dsp). Perhaps the level of effort to convert vs just starting anew?
My lay thinking is that small line level AC signals should act identical to large amplified AC signals, challenges for each extreme aside?
Would something like this be theoretically possible?
if so, what are the downsides?
Mighty there be some potential for transferring inductors or other very specific/custom/proprietary components to the active stage when transitioning from passive to active?
In my limited research, I never ran across the anyone asking if a passive crossover circuit board could be used in an active stage (for a commercial speaker it was designed for and integrated and sold with, of course) . I'm assuming there might be buffering required at inputs and outputs to account for impedance differences, or need to have a strong drive due to insertion losses or larger components, perhaps more noise, idk, but I'm wondering if it's theoretically possible.
If so, I'm assuming there must be caveats that keep doing something like or a scaled version this off the table as a recommended approach. (other than opportunity loss for other active xo benefits like custom/complex filtering via circuits or dsp). Perhaps the level of effort to convert vs just starting anew?
My lay thinking is that small line level AC signals should act identical to large amplified AC signals, challenges for each extreme aside?
Would something like this be theoretically possible?
if so, what are the downsides?
Mighty there be some potential for transferring inductors or other very specific/custom/proprietary components to the active stage when transitioning from passive to active?
Last edited:
The LC crossover filters normally used for loudspeakers have impractically low impedances, but when you scale up the impedance level a few hundred times, you get an LC filter that can very well be used for line level signals, especially with a few simple buffers. Most designers won't like the inductors, though.
Thanks MarcelvdG. I'm curious, Are the inductors problematic because of the potential EMI, saturation or something else/multiple things?
When you scale up the impedance, you also scale up the inductance, so air core inductors become impractical. (Air core inductors and any other inductor with an open magnetic circuit can also have issues with hum pick-up when they are in the neighbourhood of supply transformers.) With ferrite or iron cores, you have to watch out for distortion, although it can be made arbitrarily small when you use a large enough core with a sufficiently long air gap. Another issue is the costs, NP0 capacitors tend to be cheaper than inductors.
I've used big gapped potcore inductors in the reconstruction filter of my valve DAC and it works fine in my opinion. It wasn't cheap, though. For another hobby DAC, I intend to make a mixed filter, mostly active RC, but with an inductor in the first stage.
I've used big gapped potcore inductors in the reconstruction filter of my valve DAC and it works fine in my opinion. It wasn't cheap, though. For another hobby DAC, I intend to make a mixed filter, mostly active RC, but with an inductor in the first stage.
Harrison Labs FMODS are a basic example of what you seem to be asking about.
There is interaction with the input impedance of the amp (as you suggested) that affects the cross point (just like you'd have in a speaker level crossover with different driver impedances), so it's not a one size fits all solution that can use passive circuitry only and give precise results for any amplifier it might be connected to.
Here's a list of frequency shifts with various input impedances:
https://www.hlabs.com/products/crossovers/index_files/Page446.htm
Typical analog line level crossovers also basically do what you are asking about, but with active elements used as well. There are still passive circuits around them doing the filtering, though they tend to avoid using inductors, as mentioned above.
Some examples from Linkwitz:
https://www.linkwitzlab.com/filters.htm
There is interaction with the input impedance of the amp (as you suggested) that affects the cross point (just like you'd have in a speaker level crossover with different driver impedances), so it's not a one size fits all solution that can use passive circuitry only and give precise results for any amplifier it might be connected to.
Here's a list of frequency shifts with various input impedances:
https://www.hlabs.com/products/crossovers/index_files/Page446.htm
Typical analog line level crossovers also basically do what you are asking about, but with active elements used as well. There are still passive circuits around them doing the filtering, though they tend to avoid using inductors, as mentioned above.
Some examples from Linkwitz:
https://www.linkwitzlab.com/filters.htm
Last edited:
Thanks mattstat,
Yep, I'm aware of pllxo and harrison, but was just wondering about the reuse (with necessary scaling) of xo's themselves or components pre-existing in a cabinet, before conversion to bi-amp/active and perhaps preserving the designers original EQ filters, etc. More of an intellectual rather than practical curiosity.
Yep, I've been reading SL and Rod Elliot's pages as well. Great info!
Yep, I'm aware of pllxo and harrison, but was just wondering about the reuse (with necessary scaling) of xo's themselves or components pre-existing in a cabinet, before conversion to bi-amp/active and perhaps preserving the designers original EQ filters, etc. More of an intellectual rather than practical curiosity.
Yep, I've been reading SL and Rod Elliot's pages as well. Great info!
Last edited:
That DAC and tube DAC sounds like neat projects MarcelvdG!
Was your use of inductors in those a design choice vs RC? I'm assuming there's some benefit to doing so? Something akin to Pultec/Manley eqs that have particular characteristics?
Was your use of inductors in those a design choice vs RC? I'm assuming there's some benefit to doing so? Something akin to Pultec/Manley eqs that have particular characteristics?
Marchand line level crossovers use inductors.
https://www.marchandelec.com/xm46.html
Similar inductors are used in graphic equalizers from 70ties and 80ties. Very high level inductors are typically replaced with opamp gyrators.
https://www.marchandelec.com/xm46.html
Similar inductors are used in graphic equalizers from 70ties and 80ties. Very high level inductors are typically replaced with opamp gyrators.
Last edited:
When I designed my valve DAC, my target was to make all critical analogue and mixed-signal parts with valve technology or with technology that was available in the valve age. I only used semiconductors in the digital part and in supporting circuits such as the power supply. Building an active filter with valves would probably be more expensive than a passive filter with potcores and it would certainly draw much more power.
Regarding the other DAC, a technical advantage of passive filters over active ones is that they have no problem with steep edges / with signals with a lot of high-frequency content. Active circuits relying on negative feedback have the problem that the feedback gets less effective with increasing frequency. I therefore like to partly filter the output signal of a single-bit DAC passively before it reaches the first active circuit, if possible. I already do that in this DAC with an RC section, but I would like to see if I can improve it by also using an inductor (or actually four inductors, as it is a balanced stereo circuit).
I'm no expert, but I think it's theoretically possible.
Preamp -> input buffer -> speaker level crossover -> voltage divider -> power amp -> speaker driver
A low power amplifier is used as an input buffer to feed the speaker level crossover, and a voltage divider (LOC) is used to reduce the signal that feeds the power amplifier to drive the speaker.
The disadvantage is that the cost of the input buffer (low power amplifier) may be higher than the cost of a properly designed line level crossover component.
Since passive crossovers in speakers are designed to match the power distribution of all filters, you'll also lose the advantage of using a lower power amplifier to power the tweeter and mid-range.
Keep in mind that passive XOs are tuned for every specific driver's frequency-dependent impedance. You would need to recreate (at least some of) this impedance characteristics with further rather big and expensive passive parts.
It's possible to do that but the question is: why would you?
This is rather common. Think about a satellite subwoofer system. You are basically doing just what you are asking about. Troels Gravesen has a number of hybrid speakers that have passive tweeter/mid crossover with 2-way plate amplifiers. No reason not to really. In those cases you get the benefits of DSP for the entire speaker, exctra dynamic headroom thanks to the woofer getting it's own amp, and tweeter/mid amp not having bass frequencies. The only real "drawback" in those cases is that doing 4th order filters in DSP land is super easy so I'd be very tempted to go full active.... but I could see say if I wanted to do a 4-way with a 3-way plate amp a hybrid approach would be dandy.
Sure. The filter design is independent of the method with my crossovers.
Better to find your own way to the response target than exchange filter for filter..?
I missed the part where you were talking about converting a speaker. I'm of two minds. Lots of expensive commercial speakers have crossovers we might find objectionable. They don't have the right slope, or are crossed in ways we would really not do if time and money were no object. Point is it's always a good idea to do a full analysis of the original design first. Understand what you like an don't like about it.
For instance, I have seen an amazing rework of a B&W 802 speaker that brings down the tweeter crossover in order to improve off axis-response but at the cost of a dozen new crossover parts.
After you have completely understood the manufacturer's choices and outcomes then you start to ask "do I want to keep this as is?" If so, you want to replicate the voltage transfer function. Your analysis here will serve by giving you a curve to match in your active crossover.
One area that you might want to improve going to active/DSP land is increasing the filter order to get better vertical dispersion. Another might be taking advantage of the EQ features to alter where you want to cross a particular driver based on off-axis response. It is really hard for an engineering mind to keep the pasisve crossover as the reference when we so want to improve things. 🙂
For instance, I have seen an amazing rework of a B&W 802 speaker that brings down the tweeter crossover in order to improve off axis-response but at the cost of a dozen new crossover parts.
After you have completely understood the manufacturer's choices and outcomes then you start to ask "do I want to keep this as is?" If so, you want to replicate the voltage transfer function. Your analysis here will serve by giving you a curve to match in your active crossover.
One area that you might want to improve going to active/DSP land is increasing the filter order to get better vertical dispersion. Another might be taking advantage of the EQ features to alter where you want to cross a particular driver based on off-axis response. It is really hard for an engineering mind to keep the pasisve crossover as the reference when we so want to improve things. 🙂
As has been pointed out, Marchand produces an example of an LC PLLXO.
The biggest issue with these is trying to find inductors of large enuff values to work into high impedance loads typically seen non power amplifiers. Marchand uses a 5k termination to ameriolate this, but thatmeans a punishing load for the preamp… will need to have very low Rout and likely high current capability.
dave
The biggest issue with these is trying to find inductors of large enuff values to work into high impedance loads typically seen non power amplifiers. Marchand uses a 5k termination to ameriolate this, but thatmeans a punishing load for the preamp… will need to have very low Rout and likely high current capability.
dave
Inductors for speaker crossovers tend to be easier to design/make than inductors for high impedance circuits. Some like to re-purpose ballasts and power transformers as inductors and this can be a good way to find the component you need for an uncommon circuit. Especially with high impedance circuits it's necessary to measure them for self resonance as some need to be rejected as not covering the needed band.