50 and 90Hz dips are room modes, don't use any more than 2-3dB boost because it won't really help but stresses woofer and amp. Left and Right speaker will have different response anyway.... So, don't pay too much attention to those now. Focus on what happens above 300Hz!
Your response looks like it uses 300 or 500ms right window gating- it is ok for room modes, but for measuring at 1-2m 4-12ms will give more adequate info (less reflections from the room). Did you measure all drivers individually and use overlay to see how acoustic responses match? How did you set delays?
Nearfield responses of woofer(s) and port would be nice to see.
What drivers are playing now, is it 2, 2.5 or 3-way?
Looks like you don't have baffle step eq at all (below 600Hz). Safest is to set 1st order - 3-6dB lowpass shelf for mid (or woofer in 2-way). Then attenuate tweeter channel respectively.
1kHz bump comes most likely from baffle step, you must eq it for all overlapping drivers (peak correction with suitable Q and -3dB gain). 6kHz bump must be flattened too.
Also remember to take off-axis measurements at least at 15, 30, 45 and 60deg too! Some verticals too.
Your response looks like it uses 300 or 500ms right window gating- it is ok for room modes, but for measuring at 1-2m 4-12ms will give more adequate info (less reflections from the room). Did you measure all drivers individually and use overlay to see how acoustic responses match? How did you set delays?
Nearfield responses of woofer(s) and port would be nice to see.
What drivers are playing now, is it 2, 2.5 or 3-way?
Looks like you don't have baffle step eq at all (below 600Hz). Safest is to set 1st order - 3-6dB lowpass shelf for mid (or woofer in 2-way). Then attenuate tweeter channel respectively.
1kHz bump comes most likely from baffle step, you must eq it for all overlapping drivers (peak correction with suitable Q and -3dB gain). 6kHz bump must be flattened too.
Also remember to take off-axis measurements at least at 15, 30, 45 and 60deg too! Some verticals too.
Last edited:
Вы можете показать мне свои идеи на рисунке, дать мне размеры, и я сделаю для вас их визуализацию.
Nahimic, via google translator is saying "You can show me your ideas in a drawing, give me the dimensions, and I will make a visualization of them for you."
Just out of interest regarding your 1KHz bump. It was there in your first measurement's I think. Speakers or something in the room, cabinet, driver frames resonating. Possibly this is why you cannot remove it vias the DSP?
As Juhazi has said reducing the window gate time should allow you to home in on things, and the port measurement maybe useful too as often 1KHz resonance makes itself known there, ideally aim for it to be 20 dB down at least. At the end of the day, its a learning exercise, these are some of the factors that come into play.
Time for more measurements, on and off axis.
Just out of interest regarding your 1KHz bump. It was there in your first measurement's I think. Speakers or something in the room, cabinet, driver frames resonating. Possibly this is why you cannot remove it vias the DSP?
As Juhazi has said reducing the window gate time should allow you to home in on things, and the port measurement maybe useful too as often 1KHz resonance makes itself known there, ideally aim for it to be 20 dB down at least. At the end of the day, its a learning exercise, these are some of the factors that come into play.
Time for more measurements, on and off axis.
@Juhazi
Re delays. I did not set any as I do not understand yet what to set and why.
The drivers are 2-way, each woofer connected to own amp but filters are identical.
I did not apply any baffle step correction (consciously). As I do not know yet how to calculate it. So far I was just trying to flatten the response using crossover, boost/cut and shelving. Very first try.
Gating was default/auto by REW, I did not choose any.
I will measure the individual drivers and port. With and without crossover applied.
Re delays. I did not set any as I do not understand yet what to set and why.
The drivers are 2-way, each woofer connected to own amp but filters are identical.
I did not apply any baffle step correction (consciously). As I do not know yet how to calculate it. So far I was just trying to flatten the response using crossover, boost/cut and shelving. Very first try.
Gating was default/auto by REW, I did not choose any.
I will measure the individual drivers and port. With and without crossover applied.
Last edited:
Here is what I get for individual drivers and the port (with 2 woofers). 5cm distance, other drivers muted in HFD. No filters or gain settings except 500Hz highpass on tweeter. 1/3 smoothing
Surprised how different the response from woofer 2 is!
Surprised how different the response from woofer 2 is!
From here I set
I get this:
- tweeter: gain -3dB. Highpass LR4 at 2k. Power limiter 15W
- woofers (identical): lowpass LR4 at 2k. Power limiter 50W
I get this:
Last edited:
So, definitely room modes. Or is it interference between the two woofers (or both)?
There is also a huge dip at 2k which I (previously, in the earlier sweep) removed by overlapping the tweeter with woofers, rightly or wrongly.
I will next try to narrow the window to remove reflections.
There is also a huge dip at 2k which I (previously, in the earlier sweep) removed by overlapping the tweeter with woofers, rightly or wrongly.
I will next try to narrow the window to remove reflections.
Last edited:
Question:You are welcome! Hope it is still relatively up to date, I know he'd have been updated a couple of times since I wrote that guide. Glad if it helps!
This is how I have integrated the amps in my build. Didn't think about fore protection though.. may have been a good idea..
Post in thread 'Small active 3-way floor stander with a punch'
How is the shape of the waveguide determined?
I would also like to make one, but how do you design it? It is not an arbitrary shape, is it?
Interference of two woofers happens only above 600Hz in vertical, in your case
Lowest modes have such long wavelength that movement (of the microphone!) should be 1-2 meters... Moving the source does't change it much.
Looks like port is too long (peak at 30Hz) and measurement has collected sound from tweeter too because it runs so high above tuning
Upper woofer (1) runs without any eq or xo
Lower woofer (2) measurement shows Z wiggle 800-1100Hz - could be floor reflection
Individual responses have not enough eq, so acoustic xo is not same as electrical
diyaudio has many good threads about wavguides, start with #augerpro
and his homepage https://www.somasonus.net/waveguides
Lowest modes have such long wavelength that movement (of the microphone!) should be 1-2 meters... Moving the source does't change it much.
Looks like port is too long (peak at 30Hz) and measurement has collected sound from tweeter too because it runs so high above tuning
Upper woofer (1) runs without any eq or xo
Lower woofer (2) measurement shows Z wiggle 800-1100Hz - could be floor reflection
Individual responses have not enough eq, so acoustic xo is not same as electrical
diyaudio has many good threads about wavguides, start with #augerpro
and his homepage https://www.somasonus.net/waveguides
Last edited:
Interesting responses, with woofer 1 looking remarkably flat up to 8khz and approx 10 dB more output, its not making too much sense to me at the moment.
The measurements are of individual drivers with the microphone on axis a few cm away, all other channels muted?
Woofer 2 I would expect to be similar amplitude and a reasonable match when measured close up.
At 1 meter distance and a short 4mS time gate you will see possibly differences caused by the drivers baffle position differences in their heights relative to the floor. Also interesting is that it does seem to have a 1Khz bump, and then its frequency response tails off in a different manner. I wonder has the suspension hardened, maybe it is getting old . Do they have identical part No's and date stamp, what happens too the measurement's if you swap drivers?
Tweeter looks OK, and the combined plot, I assume measured on tweeter axis shows how the drivers and Xover combine. Just flip the tweeter phase as a learning exercise and see how it may change the response.
I think in the basic workflows users have pointed out the idea of equalising each of the three responses in this case to get a flat response for ideally an octave outside of the desired passband and then apply your desired filter slope.
The measurements are of individual drivers with the microphone on axis a few cm away, all other channels muted?
Woofer 2 I would expect to be similar amplitude and a reasonable match when measured close up.
At 1 meter distance and a short 4mS time gate you will see possibly differences caused by the drivers baffle position differences in their heights relative to the floor. Also interesting is that it does seem to have a 1Khz bump, and then its frequency response tails off in a different manner. I wonder has the suspension hardened, maybe it is getting old . Do they have identical part No's and date stamp, what happens too the measurement's if you swap drivers?
Tweeter looks OK, and the combined plot, I assume measured on tweeter axis shows how the drivers and Xover combine. Just flip the tweeter phase as a learning exercise and see how it may change the response.
I think in the basic workflows users have pointed out the idea of equalising each of the three responses in this case to get a flat response for ideally an octave outside of the desired passband and then apply your desired filter slope.
I was also puzzled by the difference in two woofers. Worth re checking everything and re measure to eliminate any errors.
Sorry I just got here. I have a couple of thoughts. The dual woofers improve sensitivity, and a shared cabinet works against doing a 2.5 way design.
Also, one of the big benefits of the DSP capabilities in the active design is the ability to easily do optimized LR4 or higher order filters. This has the dual benefit of improving off axis response, and letting the tweeter potentially play lower than it would otherwise.
Measure the driver response on axis and at 30 or 45 degrees horizontally before committing to a particular crossover topology. I wrote a little about these design choices here in a blog post.
With a little luck, some gentle EQ and digital delay and you may find you can significantly improve upon the passive crossover by using a lower crossover point with higher slopes that gives you improved off-axis listening and a wider sweet spot.
Also, one of the big benefits of the DSP capabilities in the active design is the ability to easily do optimized LR4 or higher order filters. This has the dual benefit of improving off axis response, and letting the tweeter potentially play lower than it would otherwise.
Measure the driver response on axis and at 30 or 45 degrees horizontally before committing to a particular crossover topology. I wrote a little about these design choices here in a blog post.
With a little luck, some gentle EQ and digital delay and you may find you can significantly improve upon the passive crossover by using a lower crossover point with higher slopes that gives you improved off-axis listening and a wider sweet spot.
Last edited:
Another long session learning to measure and HFD, trying to take the above advice into account and I think the results are much better.
- the two woofers measure similar now, so I must have made a mistake before. so this mystery is solved
- the two LF dips are now proven as room modes - they change completely when the mic is moved (not the speakers!) - thanks to Juhazi!
Last edited:
It can be interesting to change smoothing to 1/24 octave, just to better see any potential resonances.
I find the ability to control the response so easily and accurately using dsp absolutely amazing! When you know what you are doing that is.
I have now added a delay to the tweeter based on the relative position to the mid depth of the woofers in the baffle plane.
Some painting work tomorrow then installation of amp into the second speaker and setting up as a pair.
I have now added a delay to the tweeter based on the relative position to the mid depth of the woofers in the baffle plane.
Some painting work tomorrow then installation of amp into the second speaker and setting up as a pair.
I had poor success attempting to physically measure any distances, or even using interferometry in the case of using Hypex. The delay I thought I should use and what ended up being right were not even close.
A better solution for me came from examining the impulse response for the initial delay setting for a rough delay estimate followed by inverting the tweeter and carefully nudging the delay until I achieved the minimum null. Helps to add excess delay at first so you can clearly differentiate the two different signals. Measure the delta, subtract your excess delay and then nudge.
Also, using higher order (LR4) electroacoustic targets really helped optimize the off axis response.
Last edited: