I'm trying to figure out gating and tweeter measurements. Specifically in planar and ribbon tweeters. Dome tweeters do not seem to have the issue I am about to explain, or at least, not to the extent that ribbons and planar tweeters do.
Pertinent info: All the measurements you are about to take were taken at least 8 feet away from the nearest surface that could create a reflection.
Lets take a look a tweeter. This is the Hivi Planar RT1.3WE
https://www.parts-express.com/HiVi-...5dfNYjuxbfyS0oAO_Y0GaRLZ4JBgWhOhoCA10QAvD_BwE
Notice how it has the grill. Now, if we gate the measurement super close to the tweeter then we see the response as very smooth
Lets gate that further out now
Its already starting to get bumpy. This looks like reflective bumps, however, were this a woofer I wouldn't get any reflections until I gate out until there is reflective surface.
I will show you the woofer taken at this exact same time these measurement were taken.
Smooth even down to 230 hz
Alright, now lets gate that out to where there will be a reflection.
And there are our reflections.
I'll show another tweeter example. This is a ribbon tweeter>
GRS RT1.R
https://www.parts-express.com/GRS-RT1.R-8-Hi-Res-Neo-Ribbon-Tweeter-8-Ohm-272-202?quantity=1
This one also has a screen. It is set pretty far back into that cup so its possible its "horned" a bit. Anyways, lets take a look at its response gated really closely.
Pretty smooth down to 4khz.
Lets let it out a little bit more but still too far out to reflect off of a surface. This set of measurements were taken even further away from any reflective surface so there shouldn't be any until around 150 hz.
1 khz we have ripples. I saw this when I measured my friends studio monitors. In that case, it was from his desk. The desk was positioned around a foot away from the monitors. Once we found that out we moved the monitor and desk position so it was not a problem for his when he mixed his music. He is a musician.
I can only gate this out to 800 hz as that is as far down as I took the measurment. I will let it out far enough to get real reflections.
Now there are the reflections.
So my question is this:
Are these reflections of some kind or is it just completing the measurement as it goes or is this how measurement is aquired? As I always say, it is very possible I have a lack of knowledge here that needs filling in. I know enough to know I probably just don't fully understand.
If it is some sort of reflection then I wonder if I am trying to even out non existent response peaks when working on XOs when I should really just applying a boat load of smoothing to response.
I'm sure some of you are tiring of all my questions but I will just keep asking things I don't understand and cannot find answers for. I have a slogan that goes "the only dumb question is the one you failed to ask"
As always, thank you for reading this. Any and all responses will be read and appreciated.
Pertinent info: All the measurements you are about to take were taken at least 8 feet away from the nearest surface that could create a reflection.
Lets take a look a tweeter. This is the Hivi Planar RT1.3WE
https://www.parts-express.com/HiVi-...5dfNYjuxbfyS0oAO_Y0GaRLZ4JBgWhOhoCA10QAvD_BwE
Notice how it has the grill. Now, if we gate the measurement super close to the tweeter then we see the response as very smooth
Lets gate that further out now
Its already starting to get bumpy. This looks like reflective bumps, however, were this a woofer I wouldn't get any reflections until I gate out until there is reflective surface.
I will show you the woofer taken at this exact same time these measurement were taken.
Smooth even down to 230 hz
Alright, now lets gate that out to where there will be a reflection.
And there are our reflections.
I'll show another tweeter example. This is a ribbon tweeter>
GRS RT1.R
https://www.parts-express.com/GRS-RT1.R-8-Hi-Res-Neo-Ribbon-Tweeter-8-Ohm-272-202?quantity=1
This one also has a screen. It is set pretty far back into that cup so its possible its "horned" a bit. Anyways, lets take a look at its response gated really closely.
Pretty smooth down to 4khz.
Lets let it out a little bit more but still too far out to reflect off of a surface. This set of measurements were taken even further away from any reflective surface so there shouldn't be any until around 150 hz.
1 khz we have ripples. I saw this when I measured my friends studio monitors. In that case, it was from his desk. The desk was positioned around a foot away from the monitors. Once we found that out we moved the monitor and desk position so it was not a problem for his when he mixed his music. He is a musician.
I can only gate this out to 800 hz as that is as far down as I took the measurment. I will let it out far enough to get real reflections.
Now there are the reflections.
So my question is this:
Are these reflections of some kind or is it just completing the measurement as it goes or is this how measurement is aquired? As I always say, it is very possible I have a lack of knowledge here that needs filling in. I know enough to know I probably just don't fully understand.
If it is some sort of reflection then I wonder if I am trying to even out non existent response peaks when working on XOs when I should really just applying a boat load of smoothing to response.
I'm sure some of you are tiring of all my questions but I will just keep asking things I don't understand and cannot find answers for. I have a slogan that goes "the only dumb question is the one you failed to ask"
As always, thank you for reading this. Any and all responses will be read and appreciated.
Here are both of them mounted. There is no gap between them and the baffle. Its basically a press fit for them both.
The hivi has 1.5 inches above and 3.75" on either side.
The GRS ribbon is mounted in a teardrop as shown below.
Attachments
The GRS is reflecting off of the nearby woofer and the lip around the woofer. The HiVi is getting diffraction interference from the edges of the cabinet.
Interesting. So it is a reflection as I suspected. Short of redesigning everything...... what do you suggest I do? Apply smoothing to get my crossover in the correct place? What about that acoustic felt I see around tweeters sometimes? Does that actually do anything?
Sound absorbing material works only at frequencies where the thickness of the material is at least 1/4 the wavelength of the sound. So a 1 inch thick piece of felt on a wall will reduce reflections above frequencies that are less than 4 inches long. That means it will start working above 3000 Hz.
Below that frequency, it will just reflect as though it wasn’t there.
I suggest you try a few things with foam or felt and measure them. Also, I wonder how much of your reflection is coming from the top of the cabinet that the satellite assembly is sitting on.
If you put your microphone really close to the diaphragm and do not gate it at all, and then move your microphone further away, then when you look at the difference between the two pulses, you should be able to isolate the reflections in the second measurement.
Note that if you use a really short gate, you lose frequency resolution.
Now REW (and other log chirp measurement systems AFAIK) will only show down to a certain frequency, and the shorter your window the higher your lowest frequency shown will be. Displaying in this way gives the impression that there is a discreet frequency below which there is no data, and above which you can see clearly.
But it's not exactly that way.
In your first screen shot your window is very short, and your FR graph starts at 3 kHz. At 3k there is one data point, and the next data point isn't until 6k. Or (alternatively), at the bottom of your visible frequency range you are getting lots of smoothing.
This is the understanding I have picked up over time, I have not implemented Farina chirp stuff, I may be off on the details. But here's my point: "if we gate the measurement super close to the tweeter then we see the response as very smooth" ... well if you make your gate so short that you only start to generate data at 3k, then everything is going to look smooth. You aren't really starting to generate significant resolution within human audible bandwidth. The bottom few octaves of your visible measurement are always smooth, it's the math / process.
I kind of wish measurement software was better at indicating the available frequency resolution.
So what is a reflection? Look in the impulse response. The impulse starts as direct sound, and will be fading away, and then it pops up again with more energy, that's a reflection coming in. In your last REW screen grab, on the impulse response, you can see some energy appearing out of nowhere at something like 5.5 ms. That's a reflection. What about closer in to the main impulse? It gets a little philosophical at some point... Will a tweeter reflect / diffract off of it's own grill and mounting features? Yes, I guess, but at what point is that just the sound the driver makes? Zoom in the display to get a better look at the impulse response to identify things that are happening in a very short time period. Your black printed speaker may be generating reflections off the bumpiness of its face / front features.
As Perry says, try some foam, do a bit of experimenting. But also, don't let it bother you too much. Too much reflection / diffraction is an issue, but small features aren't audible. You are listening to direct sound + reflections all the time, there is a lot of machinery in your brain for dealing with it. I know a guy who says you can't hear finer than 1/3 octave smoothing. I'm not sure where I'd put the audibility line myself, but you get the idea.
Apologies if this response is more basic than you were looking for. Your expertise may be greater than I know.
Got it. Makes sense.
What do you mean by "satellite assembly"? Are you referring the the 3d printed enclosure or the wood cabinet?
I do plan on taking a nearfield measurement of the ribbon. I did that with the Hivi just to see. Here is the response of near field vs far field.
This is all in line with what I thought. I, as everyone, do appreciate my thought process being reinforced.
I'll be honest. I still only partially understand this. So there is a resolution factor happening here where the further I gate out the higher the resolution of the frequency response above my gate? But as it gets that resolution it is also including reflections?
"What do you mean by "satellite assembly"? Are you referring the the 3d printed enclosure or the wood cabinet?" --> Satellite assembly is the 3D printed enclosure.
The wood cabinet it's sitting on is also a source of reflection.
The wood cabinet it's sitting on is also a source of reflection.
You got it, this is the central tradeoff / difficulty in taking acoustic measurements. You want a long window to get good (dense) data about the frequencies of interest (that's FFTs for ya), but you want to be able to shorten the window to cut out reflections.
You can see in your data, FR traces only go down to a certain frequency, and the longer your window the lower that frequency goes. But it doesn't just go from 100% good to non-existent. The actual data becomes more sparse as you go down to the cutoff. So statements like...
... can be a bit misled, if you are talking about the very lowest portion of the FR shown. At the bottom of your available measurement (as dictated by your window) it will always be smooth, the data is sparse there. Lengthen your window and the lower end of your measurement will decrease, and at the same freq where your measurement previously ended more resolution will start to show more features, but these aren't necessarily reflections. You're getting more resolution at those frequencies with a longer window, so maybe that's just what the driver is doing. But they might be reflections... look at the impulse and your window to see what you are starting to include (helpful to adjust vertical limits here). And finally, when looking at the impulse at very short times it can be difficult to say what is reflection and what is direct.
Stated with fewer words: the lowest valid frequency for a particular gating is also the resolution of the measurement.
Last edited:
Ach, I apologize for writing such a lengthy letter, I hadn't time to write a shorter one. 😉
Ahh, then no. When the measurments were taken it was on a tiny little stand such that the front of the speaker overhang and no reflections could come from said stand.
Fun, so we are in the mystical area of the science now. Sounds like superconductor science.
I don't remember what you used when measuring the HiVi, but for the latest it could be reflections from the mic holder you printed.
http://www.troelsgravesen.dk/measurements.htm
http://www.troelsgravesen.dk/measurements.htm
My understanding of gate:
The gate is a window in time that is open for a certain duration - when the duration is over the gate is closed. It means that when the software that consumes the signal from the mic has the choice to listen a certain time before and after the actual measurement stimuli. What this in turn means is that the measurement data will be more or less "contaminated" with energy coming not only directly from the DUT (speaker driver) but also as reflections from nearby surfaces. Furher away surfaces, later incoming energy (copies of the direct sound from the DUT)
Frequency resolution is a poor name - it should be e.g. Mfo (lowest freq for measurement)) for a measurement. If the gate is open shorter then the time for a full sinus for a given frequency, no meaningful acquisition can be made. A gate of 1 ms naturally makes the lowest frequency measurable = 1 kHz. If you use REW and set the window to 1 ms you will see that the presentation stops at 1k. So no other "resolution" degradation than the lower cut off frequency - all data above is full "resolution" and contains no smoothing as in "Smoothing" in REW. So why does it (most often) look smoother - well you get rid of the reflections 😉 If you move the mic closer to the DUT but keep the same gate you will also see smoother traces - why? Because the relation in level between the direct sound from the driver and the reflections will increase and thus, reflections will look relatively lesser (depending on how far away the come from...)
During an open gate - all and every energy is collected to form a number of the energy level for that frequency. Longer gate, more energy.. it's like the shutter for a camera - long time, more light (but more shake ;-) - I'm not quite sure how that analogy came out - have never heard or tight about it))
Butt please correct me if I'm wrong..
//
The gate is a window in time that is open for a certain duration - when the duration is over the gate is closed. It means that when the software that consumes the signal from the mic has the choice to listen a certain time before and after the actual measurement stimuli. What this in turn means is that the measurement data will be more or less "contaminated" with energy coming not only directly from the DUT (speaker driver) but also as reflections from nearby surfaces. Furher away surfaces, later incoming energy (copies of the direct sound from the DUT)
Frequency resolution is a poor name - it should be e.g. Mfo (lowest freq for measurement)) for a measurement. If the gate is open shorter then the time for a full sinus for a given frequency, no meaningful acquisition can be made. A gate of 1 ms naturally makes the lowest frequency measurable = 1 kHz. If you use REW and set the window to 1 ms you will see that the presentation stops at 1k. So no other "resolution" degradation than the lower cut off frequency - all data above is full "resolution" and contains no smoothing as in "Smoothing" in REW. So why does it (most often) look smoother - well you get rid of the reflections 😉 If you move the mic closer to the DUT but keep the same gate you will also see smoother traces - why? Because the relation in level between the direct sound from the driver and the reflections will increase and thus, reflections will look relatively lesser (depending on how far away the come from...)
During an open gate - all and every energy is collected to form a number of the energy level for that frequency. Longer gate, more energy.. it's like the shutter for a camera - long time, more light (but more shake ;-) - I'm not quite sure how that analogy came out - have never heard or tight about it))
Butt please correct me if I'm wrong..
//
https://blog.teledynelecroy.com/2013/09/back-to-basics-what-is-fft.html
"The capture time, T, determines the frequency resolution of the FFT (Δf = 1/T)."
"The capture time, T, determines the frequency resolution of the FFT (Δf = 1/T)."