Ground-plane measurement method

[NyxOne]

Hi Everyone,

I'm currently reading "Testing loudspeaker" by Joseph D'appolito and he seems to use the "ground plane" measurement method alot.

I was wondering why nobody use that method ? It seems a bit easier than the "150 pound 4' x 8' infinite baffle" method ? Don't you think ?

Is it because of the large surface requirement ? Is there any drawback of using this method ?

Thanks,
NyxOne

[Paul W]

I've had good success with GP measurements when I want increased resolution at low frequencies. It can be tricky at HF because of surface texture, not being able to position tweeter close enough to the plane, etc. As an example, here is a 2x2 array of RS-180's at 1M measured GP.




This photo does not show a GP measurement, but it is good GP test environment...a large flat pad away from other boundaries. When interested in HF, I place a 4x8 sheet of hardboard in front of the speaker...or just elevate the whole rig as pictured.

[NyxOne]

I've had good success with GP measurements when I want increased resolution at low frequencies. It can be tricky at HF because of surface texture, not being able to position tweeter close enough to the plane, etc. As an example, here is a 2x2 array of RS-180's at 1M measured GP.

Paul,

Humm ... now that I think of it "most" of the time D'Appolito talks about GP measurement it's for the woofer. I've also seen subwoofer tested that way! So if I get the point right then GP measurement is good as long as you have a really flat floor without bump/cracks and is also better suited for LF measurement as HF measurement tend to be harder to take ?

[...] When interested in HF, I place a 4x8 sheet of hardboard in front of the speaker...or just elevate the whole rig as pictured.

I'm not sure I get this one ? What do you mean when you say "I place a 4x8 sheet of hardboard in front of the speaker" ? This doesn't make sens to me since that would cause a lot of reflections ? Hum ... could you give me further details ?

[ ... ] or just elevate the whole rig as pictured.

About this picture, won't the box cause alot of diffraction when compared to a much larger baffle ? I mean, the HF measurement won't probably have floor/ceiling bounce but you won't be able to know what the real HF looks like ? It will probably make it harder when it comes times to simulate a crossover! Am i wrong ?

BTW, forgive me if some of my questions feels a bit dumb but I really want to take measurement but I'm having trouble finding was is the simplest way to have reliable data. I don't think i can do it door since all my floor/wall and ceiling are all to close together.

Thanks, for the info.
NyxOne

[Paul W]

Yes, you want a flat floor for LF GP measurements but you also need no other surfaces nearby (like walls or ceiling).

The sheet of (1/4") hardboard I mentioned is laid flat on the ground between the mic and speaker to ensure the pebble surface of the driveway doesn't interfere with HF...there is diffraction at the edges of the sheet but the impact is lower than the pebble surface directly between the mic and speaker. Yes, after noticing erratic results at HF, I use GP for only LF.

The picture was only intended to show the flat surface with no nearby reflecting walls. In the photo, I was comparing 10" drivers rather than working on a crossover so not concerned about HF diffraction.

With crossover work your purpose, the most practical setup is elevating the speakers with as much separation from boundaries as practical. Use the actual boxes and baffles for the finished speakers...outdoors if at all possible. Accurate measurements in the actual boxes will help ensure your crossover simulations reflect actual results.

Real world, the LF is determined by the combination of speaker and listening room so baffle step compensation (for me) requires a bit of guesstimation and adjustment.

[mazurek]

I incorporate this into my "body of evidence" method. Grad school taught me to be skeptical of individual studies, rather I should collect as much corroborating evidence as possible. I feel like a good way of understanding what is really happening with low frequencies is to combine:
-in box impedence testing (along with predicted FR from T-S parameters),
-extreme near field testing,
-ground plane testing
-placing microphone very near wall for omni frequencies (don't trust unless it agrees with body of evidence)
-measuring response farther away with speakers away from boundaries, placing foam on ground between speaker and microphone to try and smooth out issues relating to test geometry,
-performing modeling on what it should be doing
-measurements many places in space
-manually adjust delays to see whether phase matching in crossover is as expected
-your ears

I guess the most important thing when you are designing a test is to make sure you know what information you actually need for a given design element. Some imperfect testing is useful as well in that it will tell you how the environment is interacting with your design, but it sure helps to be able to know what is due to the environment as see through measurement noisiness.

[NyxOne]

Yes, you want a flat floor for LF GP measurements but you also need no other surfaces nearby (like walls or ceiling).

The sheet of (1/4") hardboard I mentioned is laid flat on the ground between the mic and speaker to ensure the pebble surface of the driveway doesn't interfere with HF...there is diffraction at the edges of the sheet but the impact is lower than the pebble surface directly between the mic and speaker. Yes, after noticing erratic results at HF, I use GP for only LF.

Hummm ... i must admit that i find it surprising that you can see the effect of pebbles in your HF measurement. This is really interesting! I guess the best way to do GP measurement would be on a concrete floor outside.

The picture was only intended to show the flat surface with no nearby reflecting walls. In the photo, I was comparing 10" drivers rather than working on a crossover so not concerned about HF diffraction.

Ok, that clarify things!

With crossover work your purpose, the most practical setup is elevating the speakers with as much separation from boundaries as practical. Use the actual boxes and baffles for the finished speakers...outdoors if at all possible. Accurate measurements in the actual boxes will help ensure your crossover simulations reflect actual results.

I've taken good note of this part and I'll take it in account the next time I try to take measurement.

Real world, the LF is determined by the combination of speaker and listening room so baffle step compensation (for me) requires a bit of guesstimation and adjustment.

And this is why and when "voicing" comes into play ... i think!

Thanks Paul for all the details,
Chuck

[NyxOne]

I incorporate this into my "body of evidence" method. Grad school taught me to be skeptical of individual studies, rather I should collect as much corroborating evidence as possible. I feel like a good way of understanding what is really happening with low frequencies is to combine:
-in box impedence testing (along with predicted FR from T-S parameters),
-extreme near field testing,
-ground plane testing
-placing microphone very near wall for omni frequencies (don't trust unless it agrees with body of evidence)
-measuring response farther away with speakers away from boundaries, placing foam on ground between speaker and microphone to try and smooth out issues relating to test geometry,
-performing modeling on what it should be doing
-measurements many places in space
-manually adjust delays to see whether phase matching in crossover is as expected
-your ears

I guess the most important thing when you are designing a test is to make sure you know what information you actually need for a given design element. Some imperfect testing is useful as well in that it will tell you how the environment is interacting with your design, but it sure helps to be able to know what is due to the environment as see through measurement noisiness.

Hi mazurek,

Your technique certainly seems bullet proof but unfortunately such rigorous work demand a lot of time ... and I don't really have a lot of free time. I could quit my job but then I wouldn't have money to buy drivers and crossover parts. Seriously, the idea is more to have data that is accurate enough to work it rather than having a perfect set of data, at least that's what I want!

Thanks for the input,
Chuck

[Dennis H]

-measuring response farther away with speakers away from boundaries, placing foam on ground between speaker and microphone to try and smooth out issues relating to test geometry,

I like that one for an outdoor measurement. Speaker sitting at its normal height, mic at normal listening distance and normal ear height, foam or fiberglass to absorb HF reflections off the ground.

The trouble with jacking a speaker way high in the air outdoors to try to get an anechoic measurement is you always get bass reinforcement from the floor that doesn't show up with the speaker up high. But you don't want the HF floor reflections to be a distraction in your XO design so some absorbing material takes care of that.

[mazurek]

Dennis H:

I agree that you have to be careful not to throw out information by measuring in too unrealistic an environment. Random zig-zags in frequency response are not too important imo, but broader, more spacially uniform trends such as baffle step and boundary reinforcement can be important.

NyxOne:

I hope you find the right space, time, and care to take good measurements. That is good for you that you are reading up and planning. My list gets that long because I measure in big rooms and not warehouses/outdoors, and because sometimes my ears tell me I'm not done, and in my quest for engineering perfection/frustration I keep making up measurements until they explain what I think I'm hearing : )