A five-steps guide in great sounding Dipole loudspeakers
This is a simplified step-by-step instruction in measuring dipole loudspeakers. Dipole loudspeakers are unique. The rear radiation influence the front and this presents challenges.
Assumptions:
Blue = measurement step
Orange = configuration step
Step 1 - Decide baffle width
It is a compromise between SPL and polar response and the general rule of thumb is baffle width should be narrower than 2x diameter of the midrange. Design the speakers so that the ear/listening height is between Midrange and Tweeters.
Step 2 - Measure the Midrange
Mount the Midrange to the baffle, take it outside, lift it up so the midrange is >= 2m above the ground, and:
Measure from 1m at Midrange axis
This will provide a graph which shows the frequency, height, and Q of dipole peak. Compensate this using a notch filter. q=1.4 and a=-4db is a good start. Notch frequency f is really hard to predict.
The lowpass of the midrange will depend on excursion demand. Typically 100-200hz is good. Set this as lowpass frequency of midrange xo.
Measure off axis from 0 deg to 90 deg, in 10deg increments.
This will provide graph which shows where the dipole behaviour ends. Typically between 1.5khz to 2khz. This will set the M-T crossover point.
If the crossover point is too low for the tweeter (normally because the baffle width is too wide), then the dipole would not sound natural. Change the baffle.
Example: Measurements
showing excellent dipole response up to 2k
after which its polar response varies.
Both of the above midrange measurements cannot be reliably done indoors even with gating etc. Measurements for tweeters and woofers can be done indoors as shown below.
Step 3 - Measure the Tweeters
Mount the tweeters and take the speakers inside if needed (but outside is always better). Mount the rear tweeter and wire in reverse if you are using back-to-back domes. Set the tweeter high-pass xo as per value found from step 2.
Measure from 1m at listening axis (between M and T) and play pink noise or use impulse. Set tweeter level so that it is reasonably flat. Don't worry if it sounds too bright at this stage.
Reverse tweeter polarity and find the deepest null by setting delay values or all-pass filter of the tweeter. This is to time-align both drivers. Once found, reverse back the tweeter. Measurement should be perfectly flat now. Note that with LR2, the tweeter should be reversed.
Step 4 - Measure the Woofers
Mount the woofer and measure nearfield (1cm from cone). If it's W or H-frame measure from the opening plane. Observe the shape of the frequency response and decide the Fs and Q. From there obtain values for Linkwitz transform target F and Q. F=20hz with Q=0.5 is good depending on woofer's excursion capability.
Set the Linkwitz Transform using values above and measure again to confirm. Set also the +6b/oct dipole rolloff compensation starting from xo point down to target F.
Example: Dipole woofer measurement before (black) and
after (blue) LT filter is applied, showing F=20hz with Q=0.5.
Linkwitz transform is explained here.
Measure from 1m from listening axis between Midrange and Tweeter with pink noise. Reverse the woofer and find the deepest null by setting Woofer's delay. Reverse back and the Woofer and Midrange is now time-aligned.
Step 5 - Listen and adjust levels
Listen to the completed speakers in stereo. Observe bass level and set accordingly. Listen to tweeter level with various program material (acoustic recording). If simple dome tweeters are used, most likely it will be too bright due to uneven dispersion. Reduce tweeter level about -2 or -3db (room dependent), or optionally set a global EQ using Linkwitz DSS filter (Shelving highpass with F=1.5khz, Q=1, attenuation= -1db).
This part of the design does require subjective observations, but do not try to compare with sounds from other loudspeakers. Use live acoustic (non-amplified) performance as reference and memory from speech, talking, etc. is useful.
Note that smooth dispersion at tweeter region is a pinnacle of Dipole design. When they are smooth, a flat setting would be correct. Refer to "NaO Note" or my own "S16/MS".
Summary
Dipoles can be methodically built just like any other loudspeakers using tools widely available to amateurs/DIYers. The critical section is the midrange which requires freespace/outdoor measurements when anaechoic chamber is not available.
Note that measurements at listening position is not necessary. While it is intuitive to do so, one cannot correct deviations in time domain with amplitude correction.
to be continued ...
Advanced topics:
What then from here
Electrical vs. acoustic response
Difficulties with passive dipole loudspeakers
Fixing the achiless heel of typical dipole setup
Room analysis
Updates:
28/5/14 Minor editorial changes
This is a simplified step-by-step instruction in measuring dipole loudspeakers. Dipole loudspeakers are unique. The rear radiation influence the front and this presents challenges.
Assumptions:
- 3-way with active crossovers and Linkwitz-Riley filter using MiniDSP as active crossover
- You understand basic crossover, filters (LT, Notch, Shelves), able to configure MiniDSP
- You have and understand basic acoustic measurement tools
Blue = measurement step
Orange = configuration step
Step 1 - Decide baffle width
It is a compromise between SPL and polar response and the general rule of thumb is baffle width should be narrower than 2x diameter of the midrange. Design the speakers so that the ear/listening height is between Midrange and Tweeters.
Step 2 - Measure the Midrange
Mount the Midrange to the baffle, take it outside, lift it up so the midrange is >= 2m above the ground, and:
Measure from 1m at Midrange axis
This will provide a graph which shows the frequency, height, and Q of dipole peak. Compensate this using a notch filter. q=1.4 and a=-4db is a good start. Notch frequency f is really hard to predict.
Observe where the dipole rolloff starts, and compensate using +6db/oct lowpass shelving filter.
The lowpass of the midrange will depend on excursion demand. Typically 100-200hz is good. Set this as lowpass frequency of midrange xo.
Measure off axis from 0 deg to 90 deg, in 10deg increments.
This will provide graph which shows where the dipole behaviour ends. Typically between 1.5khz to 2khz. This will set the M-T crossover point.
If the crossover point is too low for the tweeter (normally because the baffle width is too wide), then the dipole would not sound natural. Change the baffle.
Example: Measurements
showing excellent dipole response up to 2k
after which its polar response varies.
Both of the above midrange measurements cannot be reliably done indoors even with gating etc. Measurements for tweeters and woofers can be done indoors as shown below.
Step 3 - Measure the Tweeters
Mount the tweeters and take the speakers inside if needed (but outside is always better). Mount the rear tweeter and wire in reverse if you are using back-to-back domes. Set the tweeter high-pass xo as per value found from step 2.
Measure from 1m at listening axis (between M and T) and play pink noise or use impulse. Set tweeter level so that it is reasonably flat. Don't worry if it sounds too bright at this stage.
Reverse tweeter polarity and find the deepest null by setting delay values or all-pass filter of the tweeter. This is to time-align both drivers. Once found, reverse back the tweeter. Measurement should be perfectly flat now. Note that with LR2, the tweeter should be reversed.
Step 4 - Measure the Woofers
Mount the woofer and measure nearfield (1cm from cone). If it's W or H-frame measure from the opening plane. Observe the shape of the frequency response and decide the Fs and Q. From there obtain values for Linkwitz transform target F and Q. F=20hz with Q=0.5 is good depending on woofer's excursion capability.
Set the Linkwitz Transform using values above and measure again to confirm. Set also the +6b/oct dipole rolloff compensation starting from xo point down to target F.
Example: Dipole woofer measurement before (black) and
after (blue) LT filter is applied, showing F=20hz with Q=0.5.
Linkwitz transform is explained here.
Measure from 1m from listening axis between Midrange and Tweeter with pink noise. Reverse the woofer and find the deepest null by setting Woofer's delay. Reverse back and the Woofer and Midrange is now time-aligned.
Step 5 - Listen and adjust levels
Listen to the completed speakers in stereo. Observe bass level and set accordingly. Listen to tweeter level with various program material (acoustic recording). If simple dome tweeters are used, most likely it will be too bright due to uneven dispersion. Reduce tweeter level about -2 or -3db (room dependent), or optionally set a global EQ using Linkwitz DSS filter (Shelving highpass with F=1.5khz, Q=1, attenuation= -1db).
This part of the design does require subjective observations, but do not try to compare with sounds from other loudspeakers. Use live acoustic (non-amplified) performance as reference and memory from speech, talking, etc. is useful.
Note that smooth dispersion at tweeter region is a pinnacle of Dipole design. When they are smooth, a flat setting would be correct. Refer to "NaO Note" or my own "S16/MS".
Summary
Dipoles can be methodically built just like any other loudspeakers using tools widely available to amateurs/DIYers. The critical section is the midrange which requires freespace/outdoor measurements when anaechoic chamber is not available.
Note that measurements at listening position is not necessary. While it is intuitive to do so, one cannot correct deviations in time domain with amplitude correction.
to be continued ...
Advanced topics:
What then from here
Electrical vs. acoustic response
Difficulties with passive dipole loudspeakers
Fixing the achiless heel of typical dipole setup
Room analysis
Updates:
28/5/14 Minor editorial changes
3 comments:
Hey Gain, the bass can be EQed for the listening position. Other than that, nice write up!
Dan
Hi,
Thks for this write I gain valuable info.
I still need to get my bass response right I wrote a thread on DIYaudio from inputs on how to improve the in-room bass response. Please red it any inputs is appreciated.
regards,
ttan98
Hi,
Thks for this write I gain valuable info.
I still need to get my bass response right I wrote a thread on DIYaudio from inputs on how to improve the in-room bass response. Please red it any inputs is appreciated.
regards,
ttan98
Post a Comment