One other note, is if these dont measure as good as expected, perhaps everyone has the placebo effect going on, or we aren't focusing on measuring the correct thing yet, even if we think we are because we can't explain why some drivers sound pleasing while others dont, and distortion measurements dont show that. In other words I dont think we can hear the difference of a few percentages of distortion difference, but thats one of the main measurements done now. I mean some drivers sound good and dont measure the best, while others measure great but dont sond that good. Hope we can find a way to measure things that correlate to what we here and have it be consistent.
I use the TD12M in two of my products.
My "target curve" is gently downward-sloping rather than perfectly flat, and I am able to keep the TD12M within 1.5 dB of my target curve out to 45 degrees off-axis. Of course it is falling off as we move off-axis, but it does so smoothly enough to blend well with my waveguide horizontally off-axis in the crossover region. I do not have to devote any special attention to "fixing" the response of the TD12M in the crossover network; it's "easy to work with".
I don't publish my measurements, as I've seen really good, honest measurements slammed because most people have highly unrealistic expectations of what good real-world measurements actually look like.
If I may wax subjective for a moment, the TD12M is very good at preserving dynamic contrast. While it's not a drop-in replacement for the TAD TL-1102 that I use in some of my products, that's due to differing T/S parameters; the considerably higher-efficiency TD12M doesn't go as low.
I would recommend the TD10X as a replacement for the TAD TL-1102, and I don't know of any other woofer that I could say that about. I have a pair, but haven't launched a product that uses them yet.
Duke
My "target curve" is gently downward-sloping rather than perfectly flat, and I am able to keep the TD12M within 1.5 dB of my target curve out to 45 degrees off-axis. Of course it is falling off as we move off-axis, but it does so smoothly enough to blend well with my waveguide horizontally off-axis in the crossover region. I do not have to devote any special attention to "fixing" the response of the TD12M in the crossover network; it's "easy to work with".
I don't publish my measurements, as I've seen really good, honest measurements slammed because most people have highly unrealistic expectations of what good real-world measurements actually look like.
If I may wax subjective for a moment, the TD12M is very good at preserving dynamic contrast. While it's not a drop-in replacement for the TAD TL-1102 that I use in some of my products, that's due to differing T/S parameters; the considerably higher-efficiency TD12M doesn't go as low.
I would recommend the TD10X as a replacement for the TAD TL-1102, and I don't know of any other woofer that I could say that about. I have a pair, but haven't launched a product that uses them yet.
Duke
I publish virtually all of my measurements and I take the slams of people who don't understand good from bad. You have to stick your neck out and do the right thing and the right thing is publishing the data. Not doing so can only be and IS suspect. Not doing it because others don't or there is no comparison base is a cop out. DO THE RIGHT THING and publish the data even if no one else does.
Agreed that most published measurements are useless and I completely agree with
Let me describe something that I witnessed. Knowing that the lower orders of nonlinearity aren't important audibly, but the higher orders are, I looked at some Klippel data on some of these higher orders. I was suspicious so I asked un-named manufacturer to do a "guage capability study" on the Klippel measurements. Long story made short, only the first two orders are capable, the others are meaningless, but the first two are inaudible, so basically the Klippel won't tell you a thing of any use as far as audibility is concerned. It can tell you if something is broke, but thats about all.
I'd love to see data on all the drivers out there, but someone has to start. I'd like to see the Thiel-Small values (but in MKS, as mass compliance, resistance, BL, etc. thanks - with error bars if you please) and then polar response. And Oh yeah, did I mention the need for polar response. (And let's not forget about polar response!) I'd like to see thermal data using a standard modulated noise signal (there is one on my website aimed at just this kind of thing).
You can keep the Klippel data and any THD or IM sweeps or any of that other "distortion" stuff. It doesn't mean a thing to me.
I'd be happy to post mine on the B&C drivers that I use.
Agreed that most published measurements are useless and I completely agree with
Let me describe something that I witnessed. Knowing that the lower orders of nonlinearity aren't important audibly, but the higher orders are, I looked at some Klippel data on some of these higher orders. I was suspicious so I asked un-named manufacturer to do a "guage capability study" on the Klippel measurements. Long story made short, only the first two orders are capable, the others are meaningless, but the first two are inaudible, so basically the Klippel won't tell you a thing of any use as far as audibility is concerned. It can tell you if something is broke, but thats about all.
I'd love to see data on all the drivers out there, but someone has to start. I'd like to see the Thiel-Small values (but in MKS, as mass compliance, resistance, BL, etc. thanks - with error bars if you please) and then polar response. And Oh yeah, did I mention the need for polar response. (And let's not forget about polar response!) I'd like to see thermal data using a standard modulated noise signal (there is one on my website aimed at just this kind of thing).
You can keep the Klippel data and any THD or IM sweeps or any of that other "distortion" stuff. It doesn't mean a thing to me.
I'd be happy to post mine on the B&C drivers that I use.
Hi Earl,
It is clear that you know exactly what data you are looking for when evaluating drivers. You are one of very few that knows what data they are looking for, why they are looking for it, and can understand the results. You clearly have a good reference in the B&C as well. Would you be interested in evaluating the TD15M in the same ways? This would provide a good comparison to a respected woofer and the data would be trusted coming from you. Did you also have similar data for the TAD's? I remember you had mentioned a long time ago that you had used them.
John
It is clear that you know exactly what data you are looking for when evaluating drivers. You are one of very few that knows what data they are looking for, why they are looking for it, and can understand the results. You clearly have a good reference in the B&C as well. Would you be interested in evaluating the TD15M in the same ways? This would provide a good comparison to a respected woofer and the data would be trusted coming from you. Did you also have similar data for the TAD's? I remember you had mentioned a long time ago that you had used them.
John
I'd be interested, but can't commit at this time, I'm too busy. But if you do send me a sample, I will measure it the next chance that I get.
I do have some old data from a TAD, but its not a complete set.
I generally don't do Thiele-Small values since they aren't important to me, but I certainly can.
Where I look most carefully in a woofer is its polar response at higher frequencies. This is where you will see a lot of differences. Almost all 15's are going to have excelent thermal characteristics, but thats one area where I don't have TAD data.
But sure, I'll step up to doing the measurements in a standard way.
There won't be any THD or IMD or any other such unimportant measurements. Although it is important to find X-max, but this is easily done without a Klippel or anything else.
John, I may ask you to make me a standard enclosure if thats OK. You could send the driver in the enclosure. Rounded edges of course, actually a cube would be good. I can draw it for you if you like.
I do have some old data from a TAD, but its not a complete set.
I generally don't do Thiele-Small values since they aren't important to me, but I certainly can.
Where I look most carefully in a woofer is its polar response at higher frequencies. This is where you will see a lot of differences. Almost all 15's are going to have excelent thermal characteristics, but thats one area where I don't have TAD data.
But sure, I'll step up to doing the measurements in a standard way.
There won't be any THD or IMD or any other such unimportant measurements. Although it is important to find X-max, but this is easily done without a Klippel or anything else.
John, I may ask you to make me a standard enclosure if thats OK. You could send the driver in the enclosure. Rounded edges of course, actually a cube would be good. I can draw it for you if you like.
gedlee said:
We measure T/S on every driver, so that shouldn't be an issue. I went through a LOT to be able to do repeatable and accurate T/S parameters.
Just want to make sure we clarify what you will be doing. You mention polar plots. You mention no THD or IMD, but I assume you will be looking more at the higher order distortions individually is this correct? What all does your standard data set consist of?
As for the box, just let me know the dimensions and I can whip it up. I would also like to assume that the B&C and TD15M will be measured in the same enclosure to eliminate that variable. Is this correct?
John
John
I would propose that all drivers be measured in the same box. That takes the enclosure out of the picture.
I do not do any nonlinear distortion measurements at all. I know of none that have been shown to correlate to perception.
I do a near field response for data below 200 Hz. and a far field response for data above 200 Hz which is done at every 7.5 degress away from the axis up to 90 degrees. This is done at about 2 meters. My system is not calibrated to the point where I can state a dB sensitivity with a very high accuracy, but the numbers would all be comparitively accurate.
I do an impedance curve and can do this with the driver out of the box so that the TS data can be calculated if desired.
I would measure the voltage and current spectrum for a power run of the modulated white noise signal (from my web site), power level TBD, but I don't have a huge power amp, maybe 200 watts. From this data I would plot the Re values as a function of time from this signal (from which the VC temp can be calculated). The goal is to cross correlate the VC temp curve with the input power curve to plot a signal which would be the thermal time constants "impulse response". This later aspect is not fully developed at this time, but is based on several studies that I have done in thermal effects and should be quite illuminating as a single plot which has all the thermal aspects imbedded in it. It may take some time to learn how to "read" this data.
I might do a time trace of the nearfield at the driver resonance at high levels since this can be used to calculate the Xmax.
I would propose that all drivers be measured in the same box. That takes the enclosure out of the picture.
I do not do any nonlinear distortion measurements at all. I know of none that have been shown to correlate to perception.
I do a near field response for data below 200 Hz. and a far field response for data above 200 Hz which is done at every 7.5 degress away from the axis up to 90 degrees. This is done at about 2 meters. My system is not calibrated to the point where I can state a dB sensitivity with a very high accuracy, but the numbers would all be comparitively accurate.
I do an impedance curve and can do this with the driver out of the box so that the TS data can be calculated if desired.
I would measure the voltage and current spectrum for a power run of the modulated white noise signal (from my web site), power level TBD, but I don't have a huge power amp, maybe 200 watts. From this data I would plot the Re values as a function of time from this signal (from which the VC temp can be calculated). The goal is to cross correlate the VC temp curve with the input power curve to plot a signal which would be the thermal time constants "impulse response". This later aspect is not fully developed at this time, but is based on several studies that I have done in thermal effects and should be quite illuminating as a single plot which has all the thermal aspects imbedded in it. It may take some time to learn how to "read" this data.
I might do a time trace of the nearfield at the driver resonance at high levels since this can be used to calculate the Xmax.
Brandon
Don't be too quick to do that as nothing is resolved yet. I'm not going to do any measurements if people are going to write them off. I want the parties involved, mostly John, to buy into the measurement plan and not say later on that this was not what he thought was going to be done.
Then there is my willingness to just measure drivers for people - I'm not willing to get into the habit of doing this for free as that would be rather absurd. I'll do a few to get the ball roling but then either someone else will have to take up the load or I'll have to charge.
Don't be too quick to do that as nothing is resolved yet. I'm not going to do any measurements if people are going to write them off. I want the parties involved, mostly John, to buy into the measurement plan and not say later on that this was not what he thought was going to be done.
Then there is my willingness to just measure drivers for people - I'm not willing to get into the habit of doing this for free as that would be rather absurd. I'll do a few to get the ball roling but then either someone else will have to take up the load or I'll have to charge.
If they get tested great, if they don't, that's fine too. I just figured you may be interested since you are looking at the thermal performance and both the TD15M and 15NW76 tackle this issue, but from different angles. One using well vented VC/gap, the other using a large heatsink (the magnet).
soongsc said:
Did not think reaching test condition agreements would be that difficult.
Well Zaph testing is suited to drivers from 7" inches and under. His infinite baffle is a 4' x 8' sheet with a 11 inches cutout and his VAS testing box is a 11 inches cube, so maybe he could stretch that to test some 10 inches drivers.
To correctly test 15 inches drivers, he would need to redo many things, check the testing protocol and was unwilling to do it, for obvious reasons. So he was suggesting to John to do a smaller testing protocol but the full capability of the driver wouldn't have been shown. John suggested to send a bigger baffle and such with many other drivers to compare to. John also suggested a new testing methodology for that comparison. Zaph said he doesn't have time and the interest to test other drivers since he doesn't have much time, and that John needed to agree to his methodology or go elsewhere.
Then both couldn't reach an agreement.
It would be hard to image large signal analysis would require a large baffle. Normally these are done near field. VAS could also be done using added mass method couldn't it? The real catch is the baffle should be faily stiff to avoid it's vibration from poluting the test data with drivers this size I would think.simon5 said:
soongsc said:
The added mass method for Vas measurement is very poor. It introduces all kinds of other non-linearities into the equation so results tend to be inaccurate. Here are some things I found when measuring T/S parameters:
https://www.aespeakers.com/phpbb2/viewtopic.php?f=4&t=927
Doing distortion and or response measurements nearfield has limitations too. You end up with diffraction issues off the surround that change with excursion as well as issues off the edge of the frame, etc. Also with a phase plug driver like the TD's, things look a little different with the mic right up close to the plug than they do at large distances.
EV measures everything at 2m or greater distance. From what I understand, their test baffle is something like 10ft x10ft with a very large sealed enclosure behind it.
The key is to take out all variables other than what the driver is doing. You want to eliminate the enclosure as a factor, diffraction from edges, reflections from the room, etc. Ideally for myself I plan to make a baffle that can go in the overhead garage door opening here. Drivers will be mounted in that baffle right at the ground and mic will be 2m away laid on the ground. The mic on ground eliminates reflection issues up until you get to frequencies that have a small wavelength with respect to the mic diameter. You are at a distance where diffraction off the surround doesn't have a variance with excursion, etc
John
John
Ground plane measurements have their problems too. The surface has to be completely rigid and nonporous. Concrete is porous unless you apply something like epoxy on it. You have to remember that any imperefections in the ground plane, like roughness, etc. will continuously affect the sound wave and that this will be worst case when the microphone in also placed on that boundary. This measurement ASSUMES a perfect ground plane and those are hard to come by.
Ground plane measurements have their problems too. The surface has to be completely rigid and nonporous. Concrete is porous unless you apply something like epoxy on it. You have to remember that any imperefections in the ground plane, like roughness, etc. will continuously affect the sound wave and that this will be worst case when the microphone in also placed on that boundary. This measurement ASSUMES a perfect ground plane and those are hard to come by.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.