DIY Waveguide loudspeaker kit

[gedlee]

Originally Posted by DougSmith

Thanks, Earl... it would be great to see some examples, but that is certainly helpful.

-Doug

Yes, I was waiting to describe this until I had examples that show these points. I have those examples, I just am short on time.

[Pallas]

Originally Posted by gedlee

I'm almost ready to disclose my new ideas in testing subs. I kept thinking that there had to be some way to get something meaningful. To me, frequency response of a sub is pretty much irrelavent. Once in the room its response changes and then when I EQ it it changes again. By the time I'm done there is no relationship between the FR that I started with and what it is now. What matters is: can the sub handle all this manipulation, because if it can't then I'm sunk.

So what I need to know is when does it "fart". Thats what I call it because, and the test bears this out, what usually happens is the sub is fine then all of a sudden its crap! And then it sounds like its farting. I have to know when this happens.

I do a sweep from about 10 Hz to 200 Hz and plot the spectrum in a spectrograph. I do this starting at 94 dB nearfield and raise the signal 3 dB each sweep. At some point, sure enough, every sub goes to hell. BUT, this happens at a huge variation in output levels. Sometimes - most often - when there is a plate amp it clips before the speaker shows much problems. This tends to happen at a very low SPL level. When I ran my subs with an external amp I got almost 16 dB more OBF (Output Before Fart).

Aren't you, in essence, measuring two different things here: real-world linear volume displacement of the subwoofer (which is to say, in a 4th order BP the driver/airspring/port) and real-world amp output before clipping?

Or is it more complex than that?

I think it would be interesting for you to confirm by comparing your OBF measurements (using a known high-power amp) to Klippel xmax measurements of the drivers.

[gedlee]

Originally Posted by Pallas

Aren't you, in essence, measuring two different things here: real-world linear volume displacement of the subwoofer (which is to say, in a 4th order BP the driver/airspring/port) and real-world amp output before clipping?

Or is it more complex than that?

I think it would be interesting for you to confirm by comparing your OBF measurements (using a known high-power amp) to Klippel xmax measurements of the drivers.

Correct, its those two things that I am measureing. When the sub has a plate amp installed, thats all that I CAN measure. And the combination is the most important thing - sometimes it can be seperated and sometimes not. Because seperating them is not always possible, comparisons can only be made on the unseperated results.

Kliipel only measures the driver, while my test is measuring the system. There is no reason to believe that the two things would be the same.

[Pallas]

Originally Posted by gedlee

Kliipel only measures the driver, while my test is measuring the system. There is no reason to believe that the two things would be the same.

Leaving out the amp, I think one may want to see if there's a correlation between linear xmax as measured by something like a Klippel or DUMAX. Assuming enclosures roughly similar in size and tuning, and of course drivers of similar surface area, the big variable would intuitively seem to be linear throw. (True, that is a lot of assumptions to make.)

[markus76]

If Earl's method is working it would be very welcome because it would tell you if a particular subwoofer in a particular setup for a particular application has adequate performance.

This doesn't replace good engineering and established measurement methods. It just helps the user to get the performance he's looking for. I'm curious about it.

Best, Markus

[gedlee]

Originally Posted by Pallas

Leaving out the amp, I think one may want to see if there's a correlation between linear xmax as measured by something like a Klippel or DUMAX.

This correlation or lack of, has been thrown arround a lot. It would be very nice to know, but it would require measureing a lot of different drivers and subs and I'm only making one. My guess is that there would be little correlation between the numbers and the perception.

[PB2]

Earl, what you are suggesting as your "fart test" is not so new. The speaker reviews in High Fidelity back in the 70s perhaps even the 60s would increase the bass test level until 10% distortion or buzzing was reached. Keele, in his reviews for Audio, tested for displacement limited max SPL using a shaped burst and he also increased level until 10% distortion or buzzing was encountered. He used a bridged Crown amp and some speakers would handle very short term bursts of up to 10KW - not at low bass frequencies.

It is clear that some drivers are usable well beyond the linear Xmax, depends on the driver and the nature of the non-linearities.

Pete Basel

[gedlee]

Originally Posted by PB2

Earl, what you are suggesting as your "fart test" is not so new.

Pete Basel

I think that the "new" part would be to look at the spectrum as the signal is swept. This is quite a bit different than measuring THD or a subjective "buzz". But of course if they are all "good" tests then they should all find the same results.

[pjpoes]

Wow it's been a long time since I looked back here or what has been posted. I'm glad to see that Dr. Geddes is starting to put out information on his sub testing method. This is a little bit more advanced version of what I've been doing, and very close to what I've been trying to do with the new software I've been using.

Anyway, another thing I found that worked (at least sometimes) was that I could also figure out what aspect of the port was overloading by taking a series of measurements around the port and looking at the spectrograph. Basically (as long as the mic isn't being overloaded) close measurements at different angles about the port can, to a point, differentiate if the overload is from turbulence around the outside of the port, or outright core turbulence. I mean, you can calculate this too, but I'm not so sure the correlation is there. When the mic is placed at a strong oblique angle to the port mouth, if I see a lot of higher order spectrum noise, that is less prevelant fully onaxis, that seems to jive with it being mostly turbulance from edges. If on the other hand it basically looks the same nomatter where and what angle the mic is placed, and is also higher order, it seems to be core overload as well. I've not done extensive experiments to see how accurate this is.

The way I validated the method was that I figured if the noise was higher order and thus higher in frequency, that it would probably be a bit more directional and should change with angle. If the source was the edge of the tube, such as from a lack of flaring, then the point at which the angle is most "off-axis" to the source should be the middle of the tube. If there is no core turbulance, that should be the lowest amount of extra noise. I happened to have a speaker that I had experimented with a very low tuning, basically a series of 3" drivers tuned to 35hz instead of the more optimum 135hz. In order to get a .15 cubic foot box tuned to 35hz I ended up using a 1" port, and because this was such an intentionally off design, it ended up making overloading really easy. In fact, if I vary the number of drivers operating, I can vary if the driver or port would overload first, and all of this happens at a low enough volume to make testing good. This was actually designed as a on wall flat panel tv speaker. I had calculations of what level of flow would cause edge turbulence, which would cause core turbulence, and what should cause the speaker to overload. I then cross-validated this to the measurements, and low and behold it seemed to all work ok. Where I became less certain was at bigger and better designed setups. My ported sub will reach it's limits from the port before the amp or speaker. The speaker is capable of 38mm one way xmax, in excess of 4000 watts rms, and at those levels would be well in excess of 130db's in my house, I simply can not test that. Cant test it outside because wind screws up the turbulance measurements. Also, at those levels, most of my mics, placed up close, are overloading. I have one mic capable of over 190db's, but its for spl measurements and doesn't have a flat response.

[gedlee]

Matt

Nice post.