Hi all,
I've been re-visiting the crossover on one of my old 3-way builds as I didn't feel I was getting the best out of it. I was working from a combination of boxsim for simulation and physical measurements/prototypes as I modified the filter network.
I was measuring close (~30mm) to the midrange, deliberately ignoring baffle step for the time being.
I was happy with the response (red line, attached image)
Out of curiosity more than anything else I took some measurements from about a metre away. Note that this is in a garage with lots of nearby surfaces that are all hard and parallel, however the measurement was gated for the first obvious reflection.
The new measurements showed an odd extension to frequency response that would need me to crossover with the tweeter at about 6.5k rather than the original aim of around 4k.
Any ideas on how/why this is happening and if I need to worry about this?
Measuring outside isn't currently an option due to living in britain, home of rain.
I should say that I can repeat the measurements from multiple near/far positions with the same results.
I'll be putting carpet up on the walls soon enough to help dampen down the reflections in the garage.
I've been re-visiting the crossover on one of my old 3-way builds as I didn't feel I was getting the best out of it. I was working from a combination of boxsim for simulation and physical measurements/prototypes as I modified the filter network.
I was measuring close (~30mm) to the midrange, deliberately ignoring baffle step for the time being.
I was happy with the response (red line, attached image)
Out of curiosity more than anything else I took some measurements from about a metre away. Note that this is in a garage with lots of nearby surfaces that are all hard and parallel, however the measurement was gated for the first obvious reflection.
The new measurements showed an odd extension to frequency response that would need me to crossover with the tweeter at about 6.5k rather than the original aim of around 4k.
Any ideas on how/why this is happening and if I need to worry about this?
Measuring outside isn't currently an option due to living in britain, home of rain.
I should say that I can repeat the measurements from multiple near/far positions with the same results.
I'll be putting carpet up on the walls soon enough to help dampen down the reflections in the garage.
Attachments
The second measurement at 1 m is more representative of the upper part, i.e. it shows the LP slope of the DUT correctly. The frequencies above (very apprx) 500 Hz can be taken as accurately captured even if taken in a closed and cramped space (and you used gating).
The nearfield - 30 mm in your case - will be accurate in the lower range, the upper range will be marred by cancelations/resonances from the cone/dust cap and is limited by the cone diameter.
So you can trust the aprx 1 m measurement (in the upper 2/3 rds).
The nearfield - 30 mm in your case - will be accurate in the lower range, the upper range will be marred by cancelations/resonances from the cone/dust cap and is limited by the cone diameter.
So you can trust the aprx 1 m measurement (in the upper 2/3 rds).
Last edited:
As said by Draki, in order to measure correctly the high frequencies, you need to measure far field. What is far field however depends on the baffle geometry (largely the width) and/or the driver diameter, so maybe you can measure at less than 1m and be far field anyway (this will help in having a longer gate).
Measuring very close (some 5mm) you can measure the low frequencies, only because you can safely ignore all reflections (because they are very low in level in respect to the direct sound), and use a very long gate. You need to correct for baffle step however.
Your measure at 30mm doesn't fall in the far field or very close category so it is pretty useless.
IMHO I wouldn't worry in adding carpet on the walls. The gate depends on the geometry (path difference between the direct and reflected sound), if the reflected sound is still there (the carpet can't cancel 100% the reflected sound), your gate doesn't change.
Ralf
Measuring very close (some 5mm) you can measure the low frequencies, only because you can safely ignore all reflections (because they are very low in level in respect to the direct sound), and use a very long gate. You need to correct for baffle step however.
Your measure at 30mm doesn't fall in the far field or very close category so it is pretty useless.
IMHO I wouldn't worry in adding carpet on the walls. The gate depends on the geometry (path difference between the direct and reflected sound), if the reflected sound is still there (the carpet can't cancel 100% the reflected sound), your gate doesn't change.
Ralf
Thanks for the explanations guys.
To clarify, am I correct understanding that how the driver measures at around 30mm is NOT the actual output of the driver; to get the proper measurement I need to be 'far field'?
I'm struggling a little to understand the intuition, I would have thought that if I had refections and phase cancellation with the near field response it would have been at certain resonances and created a more 'wavey' response in the graph, which is actually what I see in the far field response.
it's a 3" scan speak dome on a 34cm W baffle
To clarify, am I correct understanding that how the driver measures at around 30mm is NOT the actual output of the driver; to get the proper measurement I need to be 'far field'?
I'm struggling a little to understand the intuition, I would have thought that if I had refections and phase cancellation with the near field response it would have been at certain resonances and created a more 'wavey' response in the graph, which is actually what I see in the far field response.
it's a 3" scan speak dome on a 34cm W baffle
The nearfield measurement tells you what the driver (and only the driver) does at low frequencies. The nearfield measurement is usually no good at frequencies > 300 Hz or so.
The gated farfield measurement tells you what the driver and baffle do. The farfield measurement is usually no good at frequencies < 300 Hz or so.
For x-over development you'll also have to look at off-axis response. If possible, determine the power response in order to understand how much power is coming out of your speaker at a given frequency. The power response is a much better indicator of the sound balance perceived by listeners in a room.
The gated farfield measurement tells you what the driver and baffle do. The farfield measurement is usually no good at frequencies < 300 Hz or so.
For x-over development you'll also have to look at off-axis response. If possible, determine the power response in order to understand how much power is coming out of your speaker at a given frequency. The power response is a much better indicator of the sound balance perceived by listeners in a room.
Since this is 75 mm dome mid, presumably the HP xover point is above 600/700 Hz - it probably is, as visible from the measurement.
It is on a 34 cm wide baffle, so you can measure at 70/80 cm and still be far enough. "Far" being at least 2 times the baffle width. Use gating.
You can measure this way the tweeter also for blending with the mid.
It is on a 34 cm wide baffle, so you can measure at 70/80 cm and still be far enough. "Far" being at least 2 times the baffle width. Use gating.
You can measure this way the tweeter also for blending with the mid.
Hi mbrennwa, Merry Christmas.
Thanks for the message. When you talk about power response, do you mean beyond measuring the frequency response? In general terms I'm trying to hit LR4 slopes generated in arta then combining the output... Crossover design is much more on a driver by driver basis in that way than an overarall speaker approach.
Thanks for the message. When you talk about power response, do you mean beyond measuring the frequency response? In general terms I'm trying to hit LR4 slopes generated in arta then combining the output... Crossover design is much more on a driver by driver basis in that way than an overarall speaker approach.
Power response is more than a single measurement of a (on axis) SPL curve. This is a good explanation: https://www.diyaudio.com/forums/multi-way/327594-source-monkey-box-86.html#post5777272
If you want to dig deeper, this may be useful: GitHub - mbrennwa/mat_lspr: m-file to calculate power response of a loudspeaker from sound-pressure measurements arranged along two orbits around the speaker.
If you want to dig deeper, this may be useful: GitHub - mbrennwa/mat_lspr: m-file to calculate power response of a loudspeaker from sound-pressure measurements arranged along two orbits around the speaker.