I have data!
I retook the measurements with the gap filled between the waveguide and the planar transducer first. You'll see how different they are and how much smoother. Red line is the new measurement. I will hold of conclusions until the end.
Next up I took the tweeter measurement. I first took it with the woofer in place. Then I removed the woofer and installed a perfectly smooth plate in its stead. This should have removed the rubber surround as a cause of the reflections.
Basically the same result. Not enough variation to conclusively say there is a distinct measureable difference
Now, after pondering for a few minutes I threw up the old tweeter measurement from the last round of measurements that I took. Here's what that looks like. Super choppy
If you recall, I did not take back to back measurements with the mid. There was no way for me to take a measurement and get it EXACTLY back to where it was. Too much variation for scientific purposes.
Both of my old measurements were super choppy. These newer measurements are way smoother with further gating. I'm thinking my last set of measurements was no good. Something messed them up. I do not know what. They both have issues within the 2000 hz range. I am not sure why this would be. I did change the port/stand a bit but it shouldn't have this much effect. This little burble happens on all of the axis of measurement.
Even at 60 degrees
Something is also happening to the mid. I have no idea why I have such different data. I will be running measurements again. I will keep it in its current position for measurement since I know this area will have no reflections.
The closest reflective surface to the old measurements was a pole. That pole was a good 5 feet away though and only 1.5" wide in relation to the speaker. The burble also shows up in my even older measurements. I am not sure why this is. There must have been some issue with my old measurement setup.
There is also the possibility that the mid was improved by the gasket. I will leave the gasket as I suppose it is better to not have any gap than it is to have a 1.5mm gap.
I retook the measurements with the gap filled between the waveguide and the planar transducer first. You'll see how different they are and how much smoother. Red line is the new measurement. I will hold of conclusions until the end.
Next up I took the tweeter measurement. I first took it with the woofer in place. Then I removed the woofer and installed a perfectly smooth plate in its stead. This should have removed the rubber surround as a cause of the reflections.
Basically the same result. Not enough variation to conclusively say there is a distinct measureable difference
Now, after pondering for a few minutes I threw up the old tweeter measurement from the last round of measurements that I took. Here's what that looks like. Super choppy
If you recall, I did not take back to back measurements with the mid. There was no way for me to take a measurement and get it EXACTLY back to where it was. Too much variation for scientific purposes.
Both of my old measurements were super choppy. These newer measurements are way smoother with further gating. I'm thinking my last set of measurements was no good. Something messed them up. I do not know what. They both have issues within the 2000 hz range. I am not sure why this would be. I did change the port/stand a bit but it shouldn't have this much effect. This little burble happens on all of the axis of measurement.
Even at 60 degrees
Something is also happening to the mid. I have no idea why I have such different data. I will be running measurements again. I will keep it in its current position for measurement since I know this area will have no reflections.
The closest reflective surface to the old measurements was a pole. That pole was a good 5 feet away though and only 1.5" wide in relation to the speaker. The burble also shows up in my even older measurements. I am not sure why this is. There must have been some issue with my old measurement setup.
There is also the possibility that the mid was improved by the gasket. I will leave the gasket as I suppose it is better to not have any gap than it is to have a 1.5mm gap.
I saw that impulse wiggle in your first post and thought it looked funny. It's in the GRS, but not the HiVi. I couldn't tell if it was in the woofer.
Have you considered a test measurement filling in the middle area maybe with wadding and cover it with masking tape to make a flat surface? That central hole looks like trouble to me. Roundovers are nice on the outside where you have to have a termination anyway, but smooth n flat is better.
Have you considered a test measurement filling in the middle area maybe with wadding and cover it with masking tape to make a flat surface? That central hole looks like trouble to me. Roundovers are nice on the outside where you have to have a termination anyway, but smooth n flat is better.
Have you shorted out the non-used drivers when making measurements?
If not shorted or connected to an "on" power amp, they will re-radiate (or absorb) some of the output of the driver under test.
100% did consider this. I actually modeled this. I probably have that file somewhere in my mess of saved models.
I did not like the look of it. It looked super weird with one flat spot on a mass of curves. Like a man made part in a forest of organic material. I'm nearly always on board with function over aesthetics. A rarity, I decided to go for aesthetics this time.
On every build I use separate posts for every driver. So this model has 3 sets of brass posts. One for each driver. I do this because it makes it much easier to test the units. All I do is move the alligator clip to the next set of posts. I don't have to risk accidentally moving the speaker during my testing.
In addition, this leaves all my builds open to going active should I ever drop the $1k required to do that right.
From your description it sounds like you have left the drivers not under test un-terminated, allowing them to be a passive radiator, a "drone cone" (or ribbon..).
In the example above, a single woofer was measured outdoors, with the mic placed 2 meters from the cabinet on the ground, resulting in the relatively smooth response of the green trace.
The mic and cabinet were left in the same position, a second woofer of the same type placed beside the first with no termination, resulting in the purple trace, the response now has varied +2/-3dB, the "drone cone" adding constructively "in phase" and destructively "out of phase".
The blue trace is the same condition, the only difference an alligator clip shorting out the un-driven woofer.
The second cabinet now has no influence on the response, other than a slight increase due to the larger baffle area.
Mark100 has posted examples of tweeter/midrange/and low-mid interactions with similar deviations when the adjacent drivers not under test are not terminated.
Anyway, the response of each individual driver in your enclosure will be different if the adjacent drivers are un-terminated than they will be when they are, and as can be seen in the above, the differences are not subtle.
Art
This would be a false assumption. I ensure every driver has its own personal air space.
I actually assumed the tweeter was self isolated but it was not. If I ran the woofer I could hear the tweeter move from its back wave in the cabinet. The tweeter now has its own built in, back cup.
Lesson is, never assume a tweeter back is entirely sealed.
I don't think the idea of personal air space works, as far as isolating acoustical interactions between drivers.
I strongly second Art's recommendation to short out all drivers not being driven.
Try it...it's brain dead easy to do...and if measurements don't change, well...your air space works this time...but having made grillions of multiway measurements, it's never a good bet to count on 🙂
It's electrical - you need to short the input to the drivers that you dont measure. Or have the xover, if passive, connected. By making a driver see a short or a very low impedance, it will stop the cone moving if it is being effected from outside by a sound pressure. From an neighbour drivers perspective it will "steal " energy if it has open terminals compared to if they where shorted it becomes "dead" not interfering.
//
//
Resonant behavior does not stop at 300Hz.
See post #48 & 49, and test the difference shorting makes.
Separate internal air spaces don't stop external airwaves effect on drivers.
See post #48 & 49, and test the difference shorting makes.
Separate internal air spaces don't stop external airwaves effect on drivers.
Last edited:
So what do you do? Just hook up an amplifier playing nothing straight to the other two drivers?
By "short out" I am imagining hooking up a resistor and 9v battery to the other two drivers forcing their cone out. While this might work with a woofer, I feel like is a very bad idea for a planar driver and even more so for a ribbon
An alligator clip between the terminals of the undriven drivers shorts them out? The driver can still move though.
There is no data in post #48 or #49. I think this is nonsense above 300hz. Show me otherwise.
The alligator clip shorts the motor out and provides a bucking force to external acoustic energy trying to move the cone/diaphragm.
Same way an amplifier with no signal holds cone in place. The bucking force, counter EMF, can't fully stop the cone from moving under external force, but it greatly diminishes it.
Physical shorting, or amp shorting is well know standard operating procedure for measuring active multi-way sections.
I've seen a lot of passive guys have to learn this when they move up to active.
Like Art said, resonances have no frequency boundaries. How do you think tuning forks can excite each other above 300Hz ? 😉
The resonance effect between un-shorted driver sections can be quite dramatic....I've seen it swamp diffraction on the high end of the spectrum.
Why would the measuring technique be any different from passive to active? Wouldn't the cone/diaphragms move either way? Maybe people are measuring their drivers far different from me but I don't know what that technique would be. Straight leads to each driver individually. I don't see why anyone would the other drivers hooked up to anything else if they were not actively measuring that driver
@mark100 I may try this. I will sweep a tweeter at 80dB with woofer shorted or not. Measure from 1m.
What will you say if I show the same tweeter response? Lack of measurement resolution? Improper gating? Maybe it doesn't always happen? I can imagine that maybe an aluminum cones will vibrate, but something like Nomex doesn't seem likely. It will be interesting if it does. Usually during x-over work, both drivers have a x-over connected while I measure the other driver, and when they are played together, doesn't matter. But, if the tweeter sweep doesn't match the combined sweep, and this is the reason, that would be good to know.
What will you say if I show the same tweeter response? Lack of measurement resolution? Improper gating? Maybe it doesn't always happen? I can imagine that maybe an aluminum cones will vibrate, but something like Nomex doesn't seem likely. It will be interesting if it does. Usually during x-over work, both drivers have a x-over connected while I measure the other driver, and when they are played together, doesn't matter. But, if the tweeter sweep doesn't match the combined sweep, and this is the reason, that would be good to know.
Last edited:
I'm not embarrassed by my ignorance, although maybe I should be 😉
What are you doing with the alligator clip....is it just connected to the + and - of the inactive driver terminals and nothing else?
I think the idea here is the motor becomes a generator when shorted out. If you close the circuit and then push the cone in you'll create a microscopic amount of current.
This is how the alternator in your car creates electricity to recharge your battery.
By connecting this circuit you are making a miniscule back force against moving the cone.
I'm taking measurements this afternoon. I will try shorting with a gator clip and connecting a separate, non operating but on amplifier and see if there is any difference.
I'm going to try it too. This afternoon. Let's both come back with data and see if this is true or not.
There are physics to support it. Enough to make a difference to a measurement? We shall see. I won't say anything until I have tested for myself