• 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!

Effectiveness of DSP at passive crossover

Jim Tonic

Member
Joined
Oct 20, 2022
Messages
10
Likes
4
Hi,
I recently tried to implement some room correction with a Minidsp 2x4HD, a boring transistor-amp and my passive Speakers.

The calculated correction graphs in REW looked really nice, but the measured result in my room looked completely different.

It looks like kind of trial and error ...

So my question is:

If I want some serious digital room correction, I should sell my Speakers/Amp and go for Speakers with active crossovers and integrated DSP (high, low and mid chassis are directly "powered" by the DSP)?

Have a nice day,
Jim

PS: there was already a thread about it, but was closed
 

fpitas

Master Contributor
Forum Donor
Joined
Jul 7, 2022
Messages
9,885
Likes
14,191
Location
Northern Virginia, USA
First, did you measure using time gating? Otherwise, the reflections in your room make it a hopeless house of mirrors.
 

levimax

Major Contributor
Joined
Dec 28, 2018
Messages
2,349
Likes
3,462
Location
San Diego
Hi,
I recently tried to implement some room correction with a Minidsp 2x4HD, a boring transistor-amp and my passive Speakers.

The calculated correction graphs in REW looked really nice, but the measured result in my room looked completely different.

It looks like kind of trial and error ...

So my question is:

If I want some serious digital room correction, I should sell my Speakers/Amp and go for Speakers with active crossovers and integrated DSP (high, low and mid chassis are directly "powered" by the DSP)?

Have a nice day,
Jim

PS: there was already a thread about it, but was closed
To start out I would concentrate of FR and leave some of the other fancy stuff for later. I find the MMM method (see video below) both for a "base line" to correct and then to check if your corrections worked to be much easier and faster without having to think about gating or smoothing and the like (all of which reduce information actually). The other thing to keep in mind is that below ~300Hz the room plays a huge roll and it may not be possible to correct everything with DSP (such as a null caused by destructive interference).

 

alex-z

Addicted to Fun and Learning
Joined
Feb 19, 2021
Messages
910
Likes
1,685
Location
Canada
Hi,
I recently tried to implement some room correction with a Minidsp 2x4HD, a boring transistor-amp and my passive Speakers.

The calculated correction graphs in REW looked really nice, but the measured result in my room looked completely different.

It looks like kind of trial and error ...

So my question is:

If I want some serious digital room correction, I should sell my Speakers/Amp and go for Speakers with active crossovers and integrated DSP (high, low and mid chassis are directly "powered" by the DSP)?

Have a nice day,
Jim

PS: there was already a thread about it, but was closed

If you take the corrections calculated by REW and apply them via DSP, the resulting measurement should be fairly close. Make sure the corrections actually loaded onto the miniDSP 2x4HD correctly. If you close the software too quickly, or accidentally hit "bypass", they won't apply.

DSP powered speakers are objectively better, because you can digitally time align each driver, and there is less crossover distortion + power loss. But passive speakers can still have excellent performance, don't rush to replace good gear.
 

Rednaxela

Major Contributor
Joined
Mar 30, 2022
Messages
2,055
Likes
2,676
Location
NL
If you take the corrections calculated by REW and apply them via DSP, the resulting measurement should be fairly close.
This is also my experience. In fact I’m still pretty amazed how close.

If I want some serious digital room correction, I should sell my Speakers/Amp and go for Speakers with active crossovers and integrated DSP (high, low and mid chassis are directly "powered" by the DSP)?
No. If you can’t replicate your REW prediction with a post-EQ measurement you need to fix your workflow.
 

Keith_W

Major Contributor
Joined
Jun 26, 2016
Messages
2,533
Likes
5,809
Location
Melbourne, Australia
Step 1: Correct your loudspeakers freq response. Take your speakers outside and measure them at about 1m away head on (or further away if you have large speakers). Take about 10 sweeps, discarding sweeps that look way off (wind and ambient noise can influence your sweeps). Average your sweeps, then look for frequency deviations. Generate a filter to correct those and repeat measurements until you are satisfied. If your loudspeaker has ports, especially rear ports, it might be worth repeating those steps with the ports plugged. See this thread for a fantastic description of the process.

If your speakers are too large to take outside, or if you don't have a large enough space outside, there is an alternative. Move your speaker to the middle of the room, about 1/3 distance from your rear wall (or closer if you know that your speaker has less rear radiation). Take a tape measure and place it on the floor in front of the speaker, then mark 50cm intervals in a straight line going down the room. Take gated sweeps at each 50cm mark.

The theory is that moving your microphone changes the room response, but not the loudspeaker's frequency response. Overlay all the sweeps then average it. This should be a reasonably close approximation of your loudspeaker freq response.

Step 2: Correct for room response. Take your speakers back in and put your mic in your listening position. Using your corrected filters, take 1-2 sweeps or more if there is a lot of ambient noise. Overlay your in-room response over your measurement taken outside. You will easily see what influence your room has had on your freq response. Ignore any discrepancies above the Schroder frequency (you can determine your room's Schroder frequency using this calculator, or you might be able to easily see it in your graph). As one of the above posts mention, this should be about 300Hz or so. From here, you have a few options: 1. Move your speakers until the freq response is less objectionable, 2. Install room treatment, 3. Buy a subwoofer or several subwoofers, 4. Correct frequency deviations by only cutting peaks. Do not attempt to boost freq dips below the Schroder frequency.
 
OP
J

Jim Tonic

Member
Joined
Oct 20, 2022
Messages
10
Likes
4
thanks a lot for your support, I will try it again and (double)check my workflow.

in general:

Can I get the same results with "DSP-->AMP-->passive Speaker" compared to "DSP powered speakers?

Overlay your in-room response over your measurement taken outside.
I really like that approach: it's kind of integrate the passive crossover into the target-curve-calculation.

But not to easy and for beginners like me probalby failure prone.

I think I will focus on the peaks first...
 

Keith_W

Major Contributor
Joined
Jun 26, 2016
Messages
2,533
Likes
5,809
Location
Melbourne, Australia
To answer your original question, "Can I get the same results with "DSP-->AMP-->passive Speaker" compared to "DSP powered speakers?" the answer is: no, you can't. A fully active DSP speaker has a lot of advantages compared to leaving the crossover in place. However, I fully understand that it is difficult for beginners, because I am still on the learning curve myself despite having implemented it!

My system looks like this:
- PC (plays music, does measurements, creates the crossover, applies driver and room correction)
---> 8 channel DAC (important to get an 8 channel DAC with individual volume trim for each channel!)
---> 8 channels of amplification (if your 8ch DAC does not have individual vol trim, then you need 4 integrated amps). You should preferably use identical amps because of the gain structure but I did not do this.
---> 8 drivers (2x 3 way speakers and 2 subwoofers). The speakers were originally a passive tower speaker, but I bypassed the crossover. It is simple to do, you just open it up and solder the cable to the drivers directly onto the binding post. Take note of the crossover configuration and values while you are looking so that you know what type it is, the x-over point, and slope.

The advantages are many:
- no amplifier power wasted as heat by the passive crossover so you can get away with using lower power amps
- improved damping as the amplifier directly "sees" the reactive load of the driver (back EMF)
- individual DSP control of each driver. Each driver has its own audio channel, DAC, and amp.
- you can adjust the timing of each driver so that you get a perfect impulse response.
- you can choose any target curve you like, and you can switch between them instantaneously if you use software like Hang Loose Convolver or Acourate Convolver. I have different target curves for different recordings, e.g. treble cut + bass boost, etc, and another configuration for watching movies.
- easy to change the configuration, you will be constantly tweaking it and re-measuring to see what improvements you can squeeze out
- for really advanced DSP folk, there are even more possibilities (e.g. creating a virtual double bass array, equalizing for a larger sweet spot, etc) but I have to admit that some of it goes over my head. As you learn more, and see some suggestions from the bright folk here on ASR, new possibilities emerge.

Disadvantages:
- Cost. You have to buy lots of amplification, DAC channels, measurement setup, PC, and software.
- Complexity and difficult learning curve (but this is why you have ASR)

Because of the disadvantages, I suggest you do what I did and do it in steps.
1. Acquire measurement equipment and learn to use REW. Apply DSP to your speakers with the passive crossover intact.
2. Start buying amps. The next step is to bi-amp or tri-amp your system (depending on whether your speaker's binding posts allow you to do it). This will give you some improvements but not much. You will need to construct your own cable (from DAC to 2 or more amps) as an interim step.
3. Buy a 6 or 8 channel DAC and bypass your passive crossover. Don't forget to install a protective capacitor to protect your tweeter, because you can and will blow it up if you make a mistake in your cabling or channel configuration and send bass frequencies to it!

Because of the cost and complexity it took me about 3-4 years of saving, acquiring equipment, climbing the learning curve, constructing my own subwoofers, etc. before I was finally able to get to step 3. I know it is hard. Even now I still don't fully understand what I am doing or what I see, so I completely sympathize with you.
 
Last edited:

Tangband

Major Contributor
Joined
Sep 3, 2019
Messages
2,994
Likes
2,789
Location
Sweden
Hi,
I recently tried to implement some room correction with a Minidsp 2x4HD, a boring transistor-amp and my passive Speakers.

The calculated correction graphs in REW looked really nice, but the measured result in my room looked completely different.

It looks like kind of trial and error ...

So my question is:

If I want some serious digital room correction, I should sell my Speakers/Amp and go for Speakers with active crossovers and integrated DSP (high, low and mid chassis are directly "powered" by the DSP)?

Have a nice day,
Jim

PS: there was already a thread about it, but was closed
I have tried both active dsp crossovers and combined dsp with passive crossovers and for me, the answer is very clear. I would go for an all active dsp system with dsp crossover and roomcorrection and ditch the passive crossover completely. The reason is that the sound in many cases gets better - going digital into the crossover without any A/D will take you nearer the music.

There are some threads in this forum that explain some advantages of correctly made passive crossover designs, such as theoretically lower dynamic compression and lower distortion If using passive notch filtering.

If you are one who believes that passive technique can bring a better sound in certain cases, go ahead. You really CAN combine passive and active dsp techniques but the risk is very real that you will end up with less gain in soundquality than you have wished for. You have to have great skills in analog crossover techniques to make it really good.

Its cheaper having only one poweramp with a passive crossover in the loudspeaker and an external dsp for roomcorrection. Thats the main advantage if you ask me. Its not thechnical or sonical a better solution than an all digital solution with dsp crossovers and roomcorrection, - If done right.

In this case with a passive loudspeaker and an external device like Dirac, the user very often gonna discover that the dsp roomcorrection program is not 100% transparent, thus have minor faults in A/D and D/A and analog output stages that gonna worsening the sound slightly. And the you add the fault from the analog crossover.

To minimize the problems with many A/D D/A conversions with analog stages doing room corrections and a less good passive crossover, its much better to let a minidsp flex do everything in the digital domain .
 
Last edited:

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,229
Likes
17,812
Location
Netherlands
To answer your original question, "Can I get the same results with "DSP-->AMP-->passive Speaker" compared to "DSP powered speakers?" the answer is: no, you can't.
But that wasn't the original question as far as I see it. The question was: given passive speaker A and DSP speaker B, are additional DSP corrections done on these speakers equally effective? And the answer to that is simple: yes. The additional DSP doesn't know if the correction is for a DSP speaker or a passive speaker. It will just do frequency and phase corrections, and the speaker has to follow whatever it is given to the best of its abilities. If you get equal results is a whole different matter. There are all kinds of ****** speakers, and some you can't fix. Doesn't really matter if they are passive or DSP. Best just look at the problem at hand, and deal with that the best way possible.
 

Tangband

Major Contributor
Joined
Sep 3, 2019
Messages
2,994
Likes
2,789
Location
Sweden
Step 1: Correct your loudspeakers freq response. Take your speakers outside and measure them at about 1m away head on (or further away if you have large speakers). Take about 10 sweeps, discarding sweeps that look way off (wind and ambient noise can influence your sweeps). Average your sweeps, then look for frequency deviations. Generate a filter to correct those and repeat measurements until you are satisfied. If your loudspeaker has ports, especially rear ports, it might be worth repeating those steps with the ports plugged. See this thread for a fantastic description of the process.

If your speakers are too large to take outside, or if you don't have a large enough space outside, there is an alternative. Move your speaker to the middle of the room, about 1/3 distance from your rear wall (or closer if you know that your speaker has less rear radiation). Take a tape measure and place it on the floor in front of the speaker, then mark 50cm intervals in a straight line going down the room. Take gated sweeps at each 50cm mark.

The theory is that moving your microphone changes the room response, but not the loudspeaker's frequency response. Overlay all the sweeps then average it. This should be a reasonably close approximation of your loudspeaker freq response.

Step 2: Correct for room response. Take your speakers back in and put your mic in your listening position. Using your corrected filters, take 1-2 sweeps or more if there is a lot of ambient noise. Overlay your in-room response over your measurement taken outside. You will easily see what influence your room has had on your freq response. Ignore any discrepancies above the Schroder frequency (you can determine your room's Schroder frequency using this calculator, or you might be able to easily see it in your graph). As one of the above posts mention, this should be about 300Hz or so. From here, you have a few options: 1. Move your speakers until the freq response is less objectionable, 2. Install room treatment, 3. Buy a subwoofer or several subwoofers, 4. Correct frequency deviations by only cutting peaks. Do not attempt to boost freq dips below the Schroder frequency.
Very good advice.:)
 

Keith_W

Major Contributor
Joined
Jun 26, 2016
Messages
2,533
Likes
5,809
Location
Melbourne, Australia
But that wasn't the original question as far as I see it. The question was: given passive speaker A and DSP speaker B, are additional DSP corrections done on these speakers equally effective? And the answer to that is simple: yes. The additional DSP doesn't know if the correction is for a DSP speaker or a passive speaker. It will just do frequency and phase corrections, and the speaker has to follow whatever it is given to the best of its abilities. If you get equal results is a whole different matter. There are all kinds of ****** speakers, and some you can't fix. Doesn't really matter if they are passive or DSP. Best just look at the problem at hand, and deal with that the best way possible.

I still disagree, for all the reasons I stated above. DSP into passive speakers still has the passive crossover network in place. A DSP speaker takes a digital signal, performs the crossover digitally, does multiple channel D/A conversion into multiple amps, which power the drivers directly. There is no passive crossover in a DSP speaker. The result is definitely not the same - theoretically, measurably, and audibly.
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,229
Likes
17,812
Location
Netherlands
I still disagree, for all the reasons I stated above. DSP into passive speakers still has the passive crossover network in place. A DSP speaker takes a digital signal, performs the crossover digitally, does multiple channel D/A conversion into multiple amps, which power the drivers directly. There is no passive crossover in a DSP speaker.
That's all irrelevant. This isn't about the end result. It's about dependencies, and there are none. Also, note that a great many DSP speakers do not even have digital inputs. Not everything is the same.
The result is definitely not the same - theoretically, measurably, and audibly.
The result is also not the same if you use a different DSP speaker. So whatever you put behind the first DSP, is not relevant to the question. A PEQ applied does exactly the same thing on any speaker.

Now if we look at the potential of a DSP speaker vs a passive one, that is a different issue, and you bring up a lot of valid points.
 

Keith_W

Major Contributor
Joined
Jun 26, 2016
Messages
2,533
Likes
5,809
Location
Melbourne, Australia
"Irrelevant" that there is no passive crossover in a DSP speaker? Do you think that if the measured frequency response is the same, it will sound the same?
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,229
Likes
17,812
Location
Netherlands
"Irrelevant" that there is no passive crossover in a DSP speaker?
That's not at all what I was saying
Do you think that if the measured frequency response is the same, it will sound the same?
There is more to measurements and perceived sound than just (on-axis) frequency response. "the same" does not exist with speakers, so that question is mostly moot.

But yes, this should be largely true: There was an experiment once on DIYAudio with midrange drivers of various sizes and makes. They were all bandlimited to the same band, and EQ'ed to the same response. The test panel was not able to differentiate between them. Now doing this for two different full-range speakers is basically undoable, because there are simply too many parameters. But you'll definitely get very close.
 

Tangband

Major Contributor
Joined
Sep 3, 2019
Messages
2,994
Likes
2,789
Location
Sweden
That's not at all what I was saying

There is more to measurements and perceived sound than just (on-axis) frequency response. "the same" does not exist with speakers, so that question is mostly moot.

But yes, this should be largely true: There was an experiment once on DIYAudio with midrange drivers of various sizes and makes. They were all bandlimited to the same band, and EQ'ed to the same response. The test panel was not able to differentiate between them. Now doing this for two different full-range speakers is basically undoable, because there are simply too many parameters. But you'll definitely get very close.
I dont agree on that, and I have the same opinion as Keith W on this.
 

Tangband

Major Contributor
Joined
Sep 3, 2019
Messages
2,994
Likes
2,789
Location
Sweden
"Irrelevant" that there is no passive crossover in a DSP speaker? Do you think that if the measured frequency response is the same, it will sound the same?
The active dsp crossover will make the bass sound better, more articulate and more tuneful - and there is less IMD in an active three way loudspeaker, making it sound clearer. ( If the quality is good and the implementation is correctly done.)
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,986
Location
Riverview FL
Hi,
I recently tried to implement some room correction with a Minidsp 2x4HD, a boring transistor-amp and my passive Speakers.

The calculated correction graphs in REW looked really nice, but the measured result in my room looked completely different.

It looks like kind of trial and error ...

So my question is:

If I want some serious digital room correction, I should sell my Speakers/Amp and go for Speakers with active crossovers and integrated DSP (high, low and mid chassis are directly "powered" by the DSP)?

The question was: given passive speaker A and DSP speaker B, are additional DSP corrections done on these speakers equally effective? And the answer to that is simple: yes.

I have passive and active speakers in the same room, adjacent to each other.

MartinLogan reQuest (black) and JBL LSR 308 (red).

I suppose the JBL has some internal DSP adjusting the response in its amplifier chip.

Both receive an EQ'd signal via miniDSP and AcourateDRC creating the FIR and IIR filters, with "flat" as the target..

With psychoacoustic smoothing of the graph:

The frequency response of both as measured at the listening position respond equally well to the "room correction" filters created for each speaker.

1676043977056.png
 
Top Bottom