• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Passive to Active DIY 3-way

DeruDog

Member
Forum Donor
Joined
May 13, 2020
Messages
68
Likes
65
Location
Philadelphia, PA, USA
Two questions here: is it worth it/should I? How to do it?

I am building a triplet of 3-way passive speakers (Philharmonitors). I was considering amping them with Buckeye 3-ch Purifi, then remembered that I always wanted to do another active system; I made one before for car audio but never for home. So perhaps I go with 6-ch Hypex amplifier and 2 MiniDSP 2x4HDs. I would tune with UMIK microphone and REW, though I don't have an anechoic chamber or anything. 9-ch Hypex and 3 MiniDSP would break the budget by a bit, though I could do 2 fully-active LR channels and a fully passive Center.

Second question, assuming the above is wise, how do I do it? Two parts to this: the analog and the digital. For the passive crossovers, do I just skip the woofer crossover and do that w/ DSP, then wire the other DSP/amplifier to the mid/tweet circuit? Or is it more complicated than that?

For the digital side, I imagine there are already threads about how to tune an active 2-way system with MiniDSP. Is that the way to go?

I am currently using a Denon AVR X4300H to amp all most of my speakers. I intend to use the pre-outs from that to my LCR.

edit: With the above 6-ch amp and DSPs I was thinking to biamp the speakers with active high pass on the woofer and active low-pass on the mid/tweeter circuit. I could also reverse engineer the woofer crossover and set that up in DSP (I think) then just bypass that Xover altogether. If it makes no sense to monkey with DSP until I have better measuring equipment, I could alternatively bi-amp with the passive crossovers until I have the bandwidth to learn more accurate measurement techniques. Thanks for your thoughts.
 
Last edited:

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,513
Likes
7,014
Location
Stow, Ohio USA
Two questions here: is it worth it/should I? How to do it?

I am building a triplet of 3-way passive speakers (Philharmonitors). I was considering amping them with Buckeye 3-ch Purifi, then remembered that I always wanted to do another active system; I made one before for car audio but never for home. So perhaps I go with 6-ch Hypex amplifier and 2 MiniDSP 2x4HDs. I would tune with UMIK microphone and REW, though I don't have an anechoic chamber or anything. 9-ch Hypex and 3 MiniDSP would break the budget by a bit, though I could do 2 fully-active LR channels and a fully passive Center.

Second question, assuming the above is wise, how do I do it? Two parts to this: the analog and the digital. For the passive crossovers, do I just skip the woofer crossover and do that w/ DSP, then wire the other DSP/amplifier to the mid/tweet circuit? Or is it more complicated than that?

For the digital side, I imagine there are already threads about how to tune an active 2-way system with MiniDSP. Is that the way to go?

I am currently using a Denon AVR X4300H to amp all most of my speakers. I intend to use the pre-outs from that to my LCR.

Yes, active crossovers would likely improve most any speaker. To yield significant improvements, it is likely an entire redesign though and will take equipment and time to do so. Notably, you need to be able to do semi-dual channel and quasi-anechoic measurements (better mic, turntable and possibly an audio interface and/or mic preamp).

Active crossovers like a minidsp HD allow much more design flexibility than a designer would normally be able to consider with a passive components. If you do attempt, would take steps to protect the tweeter (at least an inline cap).

Hope this helps a bit!

Rick
 

maxxevv

Major Contributor
Joined
Apr 12, 2018
Messages
1,872
Likes
1,964
There have been 2 threads previously on such an idea.

One was a very thorough discourse on using a Okto 8 DAC to digitally control a pair of 3 way speakers with 2 spare channels used for controlling other outputs.

 
OP
D

DeruDog

Member
Forum Donor
Joined
May 13, 2020
Messages
68
Likes
65
Location
Philadelphia, PA, USA
Yes, active crossovers would likely improve most any speaker. To yield significant improvements, it is likely an entire redesign though and will take equipment and time to do so. Notably, you need to be able to do semi-dual channel and quasi-anechoic measurements (better mic, turntable and possibly an audio interface and/or mic preamp).

Active crossovers like a minidsp HD allow much more design flexibility than a designer would normally be able to consider with a passive components. If you do attempt, would take steps to protect the tweeter (at least an inline cap).

Hope this helps a bit!

Rick
Thanks Rick, that does help. Investing in a mic preamp, software for semi-dual and quasi-anechoic measurements, building a testing station, and then learning how to make it all work is a project for another year, I think. I had hoped that even basic tuning with a UMIC and REW would yield advantages. If that's not going to get me anywhere, I will probably stick with the passive route for now until I have more time to fiddle with a new speaker design.

I was wondering if having a digital high pass for the woofer and low-pass on the mid would yield a noticeable improvement over the all passive route and it sounds like it will not. Another year I may be able to take on the learning for proper tuning.

Thanks again for the input.
Dan
 

polmuaddib

Senior Member
Forum Donor
Joined
Mar 1, 2020
Messages
479
Likes
852
I am also thinking of upgrading my passive three way speakers by converting them to 3way active.
The easiest route, I believe is with hypex plate amps. They also have active crossover and DSP inside. So, a complete package. Software is free to download and try. Seems easy enough.
Of course, measurements mic is a must, but do we really need quasi anechoic measurement if we are building speakers for our use? Because we are then basically creating a FR that is tailored, room corrected.
The point is, it shouldn’t be hard.
Of course, I still haven’t managed to find the time to do it myself…
 
Last edited:

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,513
Likes
7,014
Location
Stow, Ohio USA
I am also thinking of upgrading my passive three way speakers by converting them to 3way active.
The easiest route, I believe is with hypex plate amps. They also have active crossover and DSP inside. So, a complete package. Software is free to download and try. Seems easy enough.
Of course, measurements mic is a must, but do we really need quasi anechoic measurement if we are building speakers for our use? Because we are then basically creating a FR that is tailored, room corrected.
The point is, it shouldn’t be hard.
Of course, I still haven’t managed to find the time to do it myself…

Quasi-anechoic measurements will actually simplify the effort but you could just do some off-axis measurements rather than full spins. Full spins without automation are more time consuming. If you have a dedicated space and can leave equipment fixed in place, it also makes for more consistent measurements and simplifies the effort. Certainly a number of decent DIY speakers have been done without much more than on-axis measurements, but as measurement tools have become more affordable, there is less guesswork to know you are getting a good result.

If the effort is solely for your own use ofc you can do whatever you prefer. As a technical forum, ASR has certain standards and if your effort is meant to share, following them helps. Otherwise, have at it if you are just marching to beat of your own drum. ;) Since time is a factor, I think you will find that following more common processes keeps the work simpler. It reduces the potential influences that affect your design inputs and, if you need help, will allow others to better understand them too.

If you have not done a speaker design before, you are likely underestimating the complexity and time it will take to do. Suggest you check out Linkwitz’s writings. Notably, this one: https://linkwitzlab.com/Fitz/considerations.htm
 
OP
D

DeruDog

Member
Forum Donor
Joined
May 13, 2020
Messages
68
Likes
65
Location
Philadelphia, PA, USA
Quasi-anechoic measurements will actually simplify the effort but you could just do some off-axis measurements rather than full spins. Full spins without automation are more time consuming. If you have a dedicated space and can leave equipment fixed in place, it also makes for more consistent measurements and simplifies the effort. Certainly a number of decent DIY speakers have been done without much more than on-axis measurements, but as measurement tools have become more affordable, there is less guesswork to know you are getting a good result.

If the effort is solely for your own use ofc you can do whatever you prefer. As a technical forum, ASR has certain standards and if your effort is meant to share, following them helps. Otherwise, have at it if you are just marching to beat of your own drum. ;) Since time is a factor, I think you will find that following more common processes keeps the work simpler. It reduces the potential influences that affect your design inputs and, if you need help, will allow others to better understand them too.

If you have not done a speaker design before, you are likely underestimating the complexity and time it will take to do. Suggest you check out Linkwitz’s writings. Notably, this one: https://linkwitzlab.com/Fitz/considerations.htm
I think polmuaddib, like I am, is not planning to design a speaker per se. He would take speakers that exist, presumably with a good spinorama, and modify them with an electronic crossover. I would picture that if one used the same baffle, enclosure, and drivers and then measured their drivers for a flat on-axis response, that the spinorama would be very similar and perhaps better due to the advantages of active crossovers.

In other words, off-axis is primarily affected by cabinet, baffle, and drivers. Frequency response by crossovers. Is this correct?
 

Plcamp

Addicted to Fun and Learning
Joined
Jul 6, 2020
Messages
860
Likes
1,315
Location
Ottawa
In other words, off-axis is primarily affected by cabinet, baffle, and drivers. Frequency response by crossovers. Is this correct?
DSP can enable topologies, driver delays, fir filters and all pass filters that would impact all of the above.
 

OWC

Active Member
Joined
Apr 14, 2019
Messages
204
Likes
154
In other words, off-axis is primarily affected by cabinet, baffle, and drivers. Frequency response by crossovers. Is this correct?
No, that's not correct.
They all contribute.

It depends what your goal is.
If you just simply want to transform the passive crossovers to active one, I would recommend just measuring the crossover frequency response incl impedance.
Or better, draw a schematic of it and simulate it (with the individual speaker impedances)
In combination with a before and after on-axis and 30 degrees off-axis measurement.

If the goal is to design a whole new crossover, a lot more is needed.
Minimum is to measure near-field, incl burst decay (type of waterfall diagram) and ideally individual distortion, far-field on-axis from 0 to 45 degrees. With a so called stitching method a full response can be created.

Audio Precision had a very nice guideline/tutorial for this;
AppNote: Loudspeaker Electroacoustic Measurements
 

Attachments

  • AudioPrecision-AppNote-LoudspeakerEA Measurements.pdf
    1.9 MB · Views: 146
OP
D

DeruDog

Member
Forum Donor
Joined
May 13, 2020
Messages
68
Likes
65
Location
Philadelphia, PA, USA
No, that's not correct.
They all contribute.

It depends what your goal is.
If you just simply want to transform the passive crossovers to active one, I would recommend just measuring the crossover frequency response incl impedance.
Or better, draw a schematic of it and simulate it (with the individual speaker impedances)
In combination with a before and after on-axis and 30 degrees off-axis measurement.

If the goal is to design a whole new crossover, a lot more is needed.
Minimum is to measure near-field, incl burst decay (type of waterfall diagram) and ideally individual distortion, far-field on-axis from 0 to 45 degrees. With a so called stitching method a full response can be created.

Audio Precision had a very nice guideline/tutorial for this;
AppNote: Loudspeaker Electroacoustic Measurements
Thanks!

This is similar to how I was expecting to go about it: near field measurement of passive crossovers followed by duplication with the active crossover. I have not found an impedence simulator that is intuitive enough for me though I will keep looking. In my case, with bi-amping a 3-way speaker, I would duplicate the woofer crossover actively and actively filter the bass from the mid-tweeter before sending it to that passive network. Then I would look at the measurements of the full-passive speaker and ensure that the levels were correct between the woofer driver and the mid-tweet to even response.

Actually the first step would be simply biamping the passive speaker as-is, since I don't have the time to do the active crossovers right now. Too many projects!

The Philharmonitor is so flat and good off axis (at least in the horizontal field) I don't want to mess with the sonic design ... though if I note any peaks in the measurements I might try to knock those down a bit.
 

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,513
Likes
7,014
Location
Stow, Ohio USA
I think polmuaddib, like I am, is not planning to design a speaker per se. He would take speakers that exist, presumably with a good spinorama, and modify them with an electronic crossover. I would picture that if one used the same baffle, enclosure, and drivers and then measured their drivers for a flat on-axis response, that the spinorama would be very similar and perhaps better due to the advantages of active crossovers.

In other words, off-axis is primarily affected by cabinet, baffle, and drivers. Frequency response by crossovers. Is this correct?

I understand and in your case, probably would not do. Have you asked @Dennis Murphy ?

With ample amplifier power, I have not heard dramatic changes just swapping active for passive. Some if this depends on crossover frequencies and how big and the type of inductors that are in the filter for the woofer. Harder to estimate without more details. If you want, PM with crossover details. I will respect Dennis’s IP.
 

D!sco

Senior Member
Joined
Jun 16, 2021
Messages
483
Likes
381
Been wondering this for a while, seems like the right thread to ask. Is there a way to simulate the transfer functions of analog crossovers in digital, and use them? I was contemplating this for two reasons: An active system can simulate a passive crossover without a bucket of swappable parts. An active system can use the existing analog crossover as a starting point.
 

OWC

Active Member
Joined
Apr 14, 2019
Messages
204
Likes
154
I understand and in your case, probably would not do. Have you asked @Dennis Murphy ?

With ample amplifier power, I have not heard dramatic changes just swapping active for passive. Some if this depends on crossover frequencies and how big and the type of inductors that are in the filter for the woofer. Harder to estimate without more details. If you want, PM with crossover details. I will respect Dennis’s IP.
Fact is that a passive filter is just always inferior for multiple reasons, the crossover itself is highly depending on the impedance of the speakers.
Since the inductance changes with output (Klippel Le(x) ), as well as power compression, this means the filter will shift as well.
With series crossover typologies, this effect is even worse. This is measurable btw.

Besides the inductors can also be non-linear and sometimes the capacitors (although very often effect of capacitors are negligible).
In general it's a lot harder to make a proper lowpass/highpass with bafflestep compensation filter compared to an active system.

With active filters higher slopes are not an issue, as well as very easy to implement parametric EQ's, delays, FIR filters and such, without getting strange peaks and such, because of the interaction between the components vs the impedance.
Little hard to explain, but people who have designed passive filters know immediately what I am talking about.
Passive components have a quite an high tolerance as well.

In the end it can mostly just be a very fiddly process, while an active design (can also be analog active btw), is a piece of cake.
Also power is being wasted in passive filters, for home-hifi use not much of a deal, but for PA and sound-reinforcement applications not practical anymore. In both cases it's often just a matter of costs as well.
I have seen passive filters that will exceed the price of a VERY good amplifier

There is only one advantage with a passive filter for the high-section (tweeter/horn) and that is because of the often implemented attention circuit (resistor divider, also called a L-pad) the amplifier noise also goes down, at the cost of using more amplifier output.
So basically you're improving the SNR of the amplifier in a way.

Since we have to compensate for the bafflestep that is at least -6dB, but often tweeters also have an higher sensitivity
So practically around -12dB or so, which is very significant in noise performance.
Fact is that one also needs 10^(12/10) = 16 times more output power. (being wasted in the resistors)

An hybrid approach is also possible of course.

Home hifi is pretty much the only area were passive filters are still being used, although that is rapidly chancing for the last few years (which was predictable). Some people still seem to swear by it seems.
From a logic point of view it doesn't make a lot of sense.
I guess people just like their own amplifiers to much.
 
Last edited:

OWC

Active Member
Joined
Apr 14, 2019
Messages
204
Likes
154
Been wondering this for a while, seems like the right thread to ask. Is there a way to simulate the transfer functions of analog crossovers in digital, and use them? I was contemplating this for two reasons: An active system can simulate a passive crossover without a bucket of swappable parts. An active system can use the existing analog crossover as a starting point.
Yes that is perfectly possible, even in free programs like VituixCAD, LTSpice there are a bunch other ones.

Although I wouldn't really directly translate a passive design to something active, for the fact that with an active filter one can just have a much better crossover control overall.

If you mean translating from active to passive, that is also possible.
But when calculating and simulating filters, you will already know the values from there.
Unless you mean just to use for comparison for different filter variations?
That also happens very often.
 

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,513
Likes
7,014
Location
Stow, Ohio USA
Fact is that a passive filter is just always inferior for multiple reasons, the crossover itself is highly depending on the impedance of the speakers.
Since the inductance changes with output (Klippel Le(x) ), as well as power compression, this means the filter will shift as well.
With series crossover typologies, this effect is even worse. This is measurable btw.

Besides the inductors can also be non-linear and sometimes the capacitors (although very often effect of capacitors are negligible).
In general it's a lot harder to make a proper lowpass/highpass with bafflestep compensation filter compared to an active system.

With active filters higher slopes are not an issue, as well as very easy to implement parametric EQ's, delays, FIR filters and such, without getting strange peaks and such, because of the interaction between the components vs the impedance.
Little hard to explain, but people who have designed passive filters know immediately what I am talking about.
Passive components have a quite an high tolerance as well.

In the end it can mostly just be a very fiddly process, while an active design (can also be analog active btw), is a piece of cake.
Also power is being wasted in passive filters, for home-hifi use not much of a deal, but for PA and sound-reinforcement applications not practical anymore. In both cases it's often just a matter of costs as well.
I have seen passive filters that will exceed the price of a VERY good amplifier

There is only one advantage with a passive filter for the high-section (tweeter/horn) and that is because of the often implemented attention circuit (resistor divider, also called a L-pad) the amplifier noise also goes down, at the cost of using more amplifier output.
So basically you're improving the SNR of the amplifier in a way.

Since we have to compensate for the bafflestep that is at least -6dB, but often tweeters also have an higher sensitivity
So practically around -12dB or so, which is very significant in noise performance.
Fact is that one also needs 10^(12/10) = 16 times more output power. (being wasted in the resistors)

An hybrid approach is also possible of course.

Home hifi is pretty much the only area were passive filters are still being used, although that is rapidly chancing for the last few years (which was predictable). Some people still seem to swear by it seems.
From a logic point of view it doesn't make a lot of sense.
I guess people just like their own amplifiers to much.

Thanks, yes this has all been discussed before here and on Rod Elliott’s site. See my notes earlier about the potential to improve the design. However, an amateur taking an experienced designer’s passive crossover may have benefits but does not guarantee he gets a great improvement.
 
Last edited:

OWC

Active Member
Joined
Apr 14, 2019
Messages
204
Likes
154
Thanks, yes this has all been discussed before here and on Rod Elliott’s site. See my notes earlier about the potential to improve the design. However, an amateur taking an experienced designer’s passive crossover and may have benefits but does not guarantee he gets a great improvement.
Oh yeah sorry, it's a little hard to keep up sometimes what has been said and what not :)

On that last note, there are always ways to just improve but also keep the original filter.
To name just one simple example;
Passive crossovers don't always nicely follow a 2nd order slope (only after -20dB or so), active filters always do.
Another one would be just some peaks that are very difficult to correct with a passive filter.

But otherwise, everyone can share their measurements here, plenty of people with heaps of experience around to help with the new crossover design.
 

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,513
Likes
7,014
Location
Stow, Ohio USA
Oh yeah sorry, it's a little hard to keep up sometimes what has been said and what not :)

On that last note, there are always ways to just improve but also keep the original filter.
To name just one simple example;
Passive crossovers don't always nicely follow a 2nd order slope (only after -20dB or so), active filters always do.
Another one would be just some peaks that are very difficult to correct with a passive filter.

But otherwise, everyone can share their measurements here, plenty of people with heaps of experience around to help with the new crossover design.

That is up to the OP. :)
 
Top Bottom