Driver bandwidth and determining crossover frequency: Questions

[dlneubec]

According to Dickason, the general rule of thumb (of course there are exceptions) for the lower crossover limit for midrange and tweeters is to be at least 2x the FS of the driver, 3-4x is even better.

He also suggests that the upper level crossover for woofers and midranges should be determined by the drivers horizontal polar response, in which case he suggests that the polar response at 45º off axis should be down somewhere between 3db and 6db. The reasoning is that this is a measure of on axis beaming. As frequency increases and wavelengths get shorted than the driver dia., it will start to beam and consequently its off axis polar response begins to drop off.

I’m working on a 3way design that is omnidirectional that has up and downfiring mid and woofers. The measurements I’m taking on the listening axis therefore represent readings taken on a driver’s horizontal response 90º off axis if it were mounted vertically, in a typical front firing configuration. I can see how this is very valuable information in that it really does reflect what part of the FR of the driver is omnidirectional in nature, since the more directional frequencies are going to drop off significantly at 90º off axis, as Dickason describes.

What I’m curious about is how I might be able to apply his standards for crossover points for 45º horizontally off axis to my 90º off axis situation? It would seem to simple to suggest that since 90º is 2x 45º that I could simply double his figures and look for crossover points where the drivers are -6db to -12db down.

My second question is in regards to how to determine crossover slope limitations in my application. There are cone breakup modes and distortion measurements often available for drivers on axis. These help designers determine what slopes to consider once you have determined a crossover point (or vice-versa if you are choosing the slope first).

My question is say that cone breakup and distortion limits your usable bandwidth on axis to 3 octaves. Is that 3 octaves still a good number to use when your application is 90º off axis or does a whole new set of distortion and breakup modes need to be considered?

Let’s say you're using Dickasons rule of thumb for -3 to -6db at 45º axis for determining crossover points and slope. With the RS52 you determine that you want to be -24db at 500hz and -24db at 6000hz as measured on axis. If your application, like mine, is 90º off axis, will the -24db at 500hz and 6000hz still apply, or is there a whole new set of rules? Does distortion increase off axis? Do breakup modes increase off axis?

Here is my real life example (see plot below). These are FR at 0, 45, 60 and 90º off axis for the RS52. In the omnidirectional application I’m using it in, my on axis response, since the driver is up firing, is like the red line on this graph. How do I determine the best crossover points and slopes in this application? Further, let’s say my CTC distance is a minimum of 4.75” to the tweeter. What upper crossover frequency and slope should I be considering?

I have the same issues to address with the crossover from the RS225 to the RS52. In the second attached plot, the green line is the FR of a single RS225 firing up or a 90º off axis measurement. The black line is the 90º off axis for the RS52 for reference. Actually it is asomewhat boosted FR because of the box that sits above it about 1" that would house the tweeter. For just the RS52 without the extra boost, look at the red line in the first plot, which is probably 4db or so down from the black line shown here. There will be two RS225’s, so you can add 6db to the green plot for comparison with the RS52. BTW, these were taken at 1m, 28” high, 4ms gated window.

Any thoughts on crossover points, slopes, etc.?

[joecarrow]

Dan,

The whole concept of "pistonic motion" is that the driver moves as a solid object- it does not flop around or "break up". Metal cone drivers tend to be very pistonic over their operating range, and that's why we like them.

If a driver is not exhibiting cone breakup on axis, that is to say it is still behaving pistonically, then it will not have any breakup off axis.

Regarding distortion- distortion is a cone movement that is other than what is intended. Distortion is (as far as I know, always) of a higher frequency than the input signal, and is therefore more directional. Any high frequency distortion should show up on the on-axis measurement. One of the benefits of having an up-firing woofer is that sometimes the high frequency distortion and breakup beams up toward the ceiling, and not toward the listener. It sounds extraordinarily weird if there's an operating ceiling fan above it.

[dlneubec]

Anyone else willing to offer some insight?

[TacoD]

1) No it isn't a linear relationship or physical law, just some guidelines. Ideally you could determine the upper frequency limit until roll of you can measure on the listening spot when the driver is pointing upwards. So measure the SPL response in your intended application. I do not expect any deviations below this frequency when going off-axis (compare with MBL radialstrahler). Your RS52 has full output til 2 kHz, you have to correct the response (lower SPL in mid range ) when you want to cross at say 4 kHz. Unfortunate you are wasting a lot of direct energy by pointing upwards, this will result in a lower sensitivity for the midrange.

Still, it is very difficult to predict the lobing and phase tracking errors due to your setup when you add a second source (e.g. the tweeter). The general "xo-lobing" experiences are not 1to1 applicable. I would aim at steep filter slopes to overcome too much overlap and hence phase problems of the drivers.

2) Cone break-up is caused by the intrinsic material parameters. Off axis this could be perceived differently (magnitude of the peak in FR), but the reference frame (or listening angle) does not influence the material behaviour, it's just another way to look at the same phenomenon.

3) Picking XO positions is a bit tricky in this situation, in your case I would fill in baffle step with the woofers so cross in the 300-500 Hz range (when the RS58 permits), also as the low frequencies are already omnipolar I'm not sure if the 90 degrees woofer orientation will be an improvement. The cross to the tweeter as low as the vifa allows, therefore 3 kHz should do, but in that case you have to pad down the mid.

Unfortunate speaker design is an iterative process, there isn't such a thing as a standard recipe to succes. And in your non-standard application it is even more difficult to predict the outcome. Even for more experienced DIYers.

[dlneubec]

Thanks, TacoD, that is very helpful.

Since the drivers are time aligned (at least on axis) doesn't that potentially help with phase issues?

Which slope types improve power response, shallow or steep, even or odd order?

I plan to do my measurements on the listening axis, which is 90º off the normal driver axis. That's how I did the original omni project design and that worked out quite well.

I will most likely have a baffle or diffuser/deflector above the mid, which in my tests so far, boosts the 90º off axis output by around 4-6 db. This comes close to offsetting the off axis losses, so the driver is not that much less efficient. In the first graph below, you can see the normal on axis RS52 response (purple). The 90º off axis response is red. However, in the second graph, you can see the 90º off axis response (this time black) and the same response with a baffle set 1" above it. The boost is spl is similar to the loss in the first graph.

With a minimum CTC distance of 4.75 from the mid to the tweeter, doesn't that suggest a lower crossover point than 3000hz, where the wavelength is about 4.5"?

[dlneubec]

1) 3) Picking XO positions is a bit tricky in this situation, in your case I would fill in baffle step with the woofers so cross in the 300-500 Hz range (when the RS58 permits), also as the low frequencies are already omnipolar I'm not sure if the 90 degrees woofer orientation will be an improvement. The cross to the tweeter as low as the vifa allows, therefore 3 kHz should do, but in that case you have to pad down the mid.

TacoD,

I did not notice your reference to "filling in the baffle step". In this application, with omnipolar woofers and mid, I don't believe there will be any need to apply any baffle step adjustments. This is due to the fact that the woofers and mid are all playing in 4pi space, so the effect of baffle size vs. wavelength do not apply as they normally would. At least this is what John K. explained to me in some advice he gave me on my first omni project.

Would you disagree with this?