I will insert any measurements that someone sends me (they have to be in the proper format, I'm not going to take arbitrary data), but for now its locked such that only I can use it and add data. I need to decide about the "public domain" thing, but clearly for the forseable future I am going to control the access.
Last edited:
The measurements are all in enclosures and there is compensation for the enclosure size to the extent that this is required (which isn't much). You have to understand that this data IS NOT what you are used to seeing because some very advanced techniques are used here that I am not willing to disclose completely. These techniques make "Compensation for baffle geometry" trivial to automatic.
Polar responses on totally nude drivers is rather pointless, but the program doesn't care if you have that data. I don't and never would take data that way.
Believe it or not if you wanted the dipole response, I could give that to you from its response in a box - as I said, its not your usual calculations.
This is exactly the issue. I do things one way, someone else another - no comparison is possible. I don't take THD so I don't show it, other don't takle polars so they don't show it. The measurements are a clear reflection of what we believe to be important - and very few in audio agree to any of that.
Cool, I know that X-over pro sold on PE has a fair sized database of FR plots. Its a decent little XO package. Of course when I used it, it didnt have my specific drivers so I had to important them.
I ended up just using the free speaker workshop XO package. Its not bad at all for being free (I found out it was easy for me to learn).
Well, it sounds like a neat idea.
The problem is that no one will have identical numbers because of a host of issues. Different equipment, techniques, environment, and knowledge make this an uphill example. Then there is always the issue of calibration of equipment.
I'm not sure what mcmahon48 is trying to do. Is he trying to test 2∏*half-space? If so, you need to bury the cabinet into the ground so that the baffle is flush with the ground.
Ignoring the problem that the ground is rough and a huge diffraction source, you can do that.
Otherwise, if the cabinet is simply lying face up you still have reflections from the ground from the waveforms that normally would be 4∏ full-space.
The only way to do it is to raise the cabinet up 30 or 40 feet into the air, probably with a crane. How many of us have that?
You can try a ground plane measurement, but the reflective surface probably is rough enough to cause diffraction artifacts itself.
Bottom line is that testing is a very, very hard thing to do and will exhibit great degrees of variability from person to person based on skill, environment, and equipment.
The problem is that no one will have identical numbers because of a host of issues. Different equipment, techniques, environment, and knowledge make this an uphill example. Then there is always the issue of calibration of equipment.
I'm not sure what mcmahon48 is trying to do. Is he trying to test 2∏*half-space? If so, you need to bury the cabinet into the ground so that the baffle is flush with the ground.
Ignoring the problem that the ground is rough and a huge diffraction source, you can do that.
Otherwise, if the cabinet is simply lying face up you still have reflections from the ground from the waveforms that normally would be 4∏ full-space.
The only way to do it is to raise the cabinet up 30 or 40 feet into the air, probably with a crane. How many of us have that?
You can try a ground plane measurement, but the reflective surface probably is rough enough to cause diffraction artifacts itself.
Bottom line is that testing is a very, very hard thing to do and will exhibit great degrees of variability from person to person based on skill, environment, and equipment.
Its not that hard to do, its just that everybody does do it differently. What I do is simple - a real no brainer - (the measurement side that is - analysis of the data is another ball game all together) but people still want to "do their own thing". So everybody does it differently and nothing is comparable. I suspect it will always be this way.
I will disagree with you in practice. Yes, there are plenty of documents that describe testing procedures and they may be good procedures, but what you have no control over is how well those procedures are executed, the equipment that is used, and the environment they are used in.
So, the theory is well founded enough, but you can't be sure that the execution is going to be statistically valid, it is just too subjective and with a high likelihood of variation.
It will always be that way because there are no public standards that testing be done to (at least for the hobbyist) nor any agency to certify that individuals are meeting a minimum set of criteria.
And... that is the very hard part of it.
The main thing is getting a testing standard procedure
I believe that there will be many variations, For even the manufactures FRD are based on the best box volume with a sized tuned port if need, Yet if a standard is set for doing the measurement will be valid if the software is as acurate as stated. For the program will paint the picture of what is going on. That is why I Stated that all the variables like box dimensions, vent length, addition box shape characteristics, bracing, padding, tube resonance frequency, tempature, baffle size, flush mounted, even though it may be a very large data base it would show the many characteristic for each speaker.
I wish that a Database would show many FRD for each speaker to paint a better picture of how the speaker performs. You also have to remember that the manufacturers are painting the best picture to sell their speakers. And the sound is going to be different do to the many variables outside of what they show.
Testing format should be like how I do it have the speaker outside away from and reflective walls and point up from the ground. I hate it when the manufacture says one thing but you dont know how they spec it out frequency +- 3db, 5db, 10dbthe only speakers that are a problem are the woofers due to the extending the bass range and being able to get very good readings below 300 hz .
I believe that there will be many variations, For even the manufactures FRD are based on the best box volume with a sized tuned port if need, Yet if a standard is set for doing the measurement will be valid if the software is as acurate as stated. For the program will paint the picture of what is going on. That is why I Stated that all the variables like box dimensions, vent length, addition box shape characteristics, bracing, padding, tube resonance frequency, tempature, baffle size, flush mounted, even though it may be a very large data base it would show the many characteristic for each speaker.
I wish that a Database would show many FRD for each speaker to paint a better picture of how the speaker performs. You also have to remember that the manufacturers are painting the best picture to sell their speakers. And the sound is going to be different do to the many variables outside of what they show.
Testing format should be like how I do it have the speaker outside away from and reflective walls and point up from the ground. I hate it when the manufacture says one thing but you dont know how they spec it out frequency +- 3db, 5db, 10db
the only speakers that are a problem are the woofers due to the extending the bass range and being able to get very good readings below 300 hz .Not everybody has the same space outside nor the same degree of environmental conditions.
I have all year round to work outside, but it is windy most every day. Some places have more ambient noises than others. Also, my temperature and humidity vary and they won't be the same set of conditions you test in.
Fails to install if using Google Chrome browser (says cannot find files needed for installation). Installs OK from IE8. Only get the Piston plot though, others say e.g. "Could not find the file \\www.gedlee.com\summa.txt" (though it is there on the website) followed by a few "Oops, should not be here". The contouring algorithm seems to struggle with the data somewhat.
You guys are making a mountain out of a mole hill. All that is required is a PC (its internal sound card is fine), a mic and pre amp (A Berhinger works, but a good mic like Earthworks is better, but not required, calibration is unnecessary, some sound cards have pre-amps), HolmImpulse, and a wooden test stand - the stand costs maybe $20 and takes an hour or two to make. I use my living room and almost anybody has enough space in some room if they clear it out. Measure the impulse response at some distance > 1 meter (distance isn't critical if its not too close), rotate the speaker on the home made test stand and measure again. Its hard to mess up Holm, a little practice and its a non-brainer. All of the details that people are worrying about are less important than the resolution of this system and this approach is as good as is required for "typical" drivers whose production variances are usually quite large.
Basically trying to get better than a dB or two is not required, but with a little practice its not hard. And thats just about as good as the drivers will be anyways.
Do all the measurements "relative" and calibration is unnecessary.
Basically trying to get better than a dB or two is not required, but with a little practice its not hard. And thats just about as good as the drivers will be anyways.
Do all the measurements "relative" and calibration is unnecessary.
If you cannot get to the data files then the program will certainly fail as you noted. Since, as you can see, the data files are there, if your browser or computer won't let you get to them then thats an issue with the browser and your connection, there isn't anything that I can do to correct it. But its worked for me everytime. Its hard to debug a problem that you cannot recreate.
many manufactures plots are measured on an infinite baffle with a non optimal box volume for the particular driver being measured. Final baffle geometry and box volume will have an effect on the ultimate FR of any particular driver. Not withstanding the variables as mentioned already, due to equipment and expertise, any measurements will give an idea of how that driver will perform in that particular configuration and not necessarily in another configuration. How is one to determine that the FR of a particular driver is due to inherent driver characteristics or due to interference effects from the baffle or possibly even the room and measurement setup. I recently took measurements and found that moving the speaker forward by around 25mm on my measurement platform made a huge difference to the smoothness of the measured tweeter response. I had the baffle flush with the edge of the 32mm thick platform (and the speaker has a chamfer and is on 30mm long legs) this caused a BIG diffraction problem. Moving the speaker forward so that the edge of the measurement platform was about 25mm behind the front of the baffle fixed the issue. But up until that point I thought that my baffle was causing the problem with the measured response!
Such a database would need to include the details of the box, both volume and baffle geometry, and details of how the measurements were done and whether the equipment was calibrated or not, in order to be useful in my opinion. I'd also say that a final measurement of the complete speaker system using the driver(s) with crossover would be invaluable as it will show what is actually possible with those drivers, as what may look bad to the untrained eye, once a proper crossover has been made may produce quite an acceptable result
Tony.
Such a database would need to include the details of the box, both volume and baffle geometry, and details of how the measurements were done and whether the equipment was calibrated or not, in order to be useful in my opinion. I'd also say that a final measurement of the complete speaker system using the driver(s) with crossover would be invaluable as it will show what is actually possible with those drivers, as what may look bad to the untrained eye, once a proper crossover has been made may produce quite an acceptable result
Tony.
I understand that, yet the goal is for The DIY Community
Then the idea of using Holmimpulse for doing DIY is useless is what you are saying, yet many say it is great, but the information is of no value. We need something to help us, since Many manufacturers are not willing and their specs are very vague. There is always going to be variables in the DIY and the speakers are going to perform differently so a testing standard for the DIY speaker testing procedure needs to be discussed that would be determined to give the most consistent results. Getting the most accurate with what we have available to many DIY speaker builders.
Like what would be most practical way and requirements for getting the desired measurements. A well developed guide for using Holmimpulse is also needed and how to interpret it, especially at the crossover points.
1 required distance to get the minimum reflections in the like in house/ large room ( the on the ground Idea) ( I have not heard of anyone having to hang their speakers 40 feet in the air in a room, That would be out side above the roof)
2 many times the speakers will have a wall behind them in real use
3 quick and simple set up for testing
4 what is the none reflective distance for 20 hz or a standard way of getting rid of the reflections to get a dependable response. That can be easily done by the DIY community, I don't know if the idea of using egg crates to cancel reflections actually works or not
5 manufacturers don't always take the time to make speaker box in unusual shapes to better shape the FR example Orb speakers did take the time to see the effect, this is why having many variations that are valid would be helpful.
6 to see the different capabilities of each speaker( the many FRD Files for each speaker, for the goal is to get the best sound we can get from the speaker)
7 what would be the best FRD octave smoothing as a standard I think 1/6
8 what specs should be also included ( other files- THD, Harmonics, polar) and how to gather the data properly, and what would be acceptable variations,
9 a group of people who do understand how to test would set up a program of different means of testing outdoor testing,non reflective chambered testing, the standard software for testing would be Holmimpulse, all microphones would be individually calibrated that would be able to be made or done by the DIY community
Then the idea of using Holmimpulse for doing DIY is useless is what you are saying, yet many say it is great, but the information is of no value. We need something to help us, since Many manufacturers are not willing and their specs are very vague. There is always going to be variables in the DIY and the speakers are going to perform differently so a testing standard for the DIY speaker testing procedure needs to be discussed that would be determined to give the most consistent results. Getting the most accurate with what we have available to many DIY speaker builders.
Like what would be most practical way and requirements for getting the desired measurements. A well developed guide for using Holmimpulse is also needed and how to interpret it, especially at the crossover points.
1 required distance to get the minimum reflections in the like in house/ large room ( the on the ground Idea) ( I have not heard of anyone having to hang their speakers 40 feet in the air in a room, That would be out side above the roof)
2 many times the speakers will have a wall behind them in real use
3 quick and simple set up for testing
4 what is the none reflective distance for 20 hz or a standard way of getting rid of the reflections to get a dependable response. That can be easily done by the DIY community, I don't know if the idea of using egg crates to cancel reflections actually works or not
5 manufacturers don't always take the time to make speaker box in unusual shapes to better shape the FR example Orb speakers did take the time to see the effect, this is why having many variations that are valid would be helpful.
6 to see the different capabilities of each speaker( the many FRD Files for each speaker, for the goal is to get the best sound we can get from the speaker)
7 what would be the best FRD octave smoothing as a standard I think 1/6
8 what specs should be also included ( other files- THD, Harmonics, polar) and how to gather the data properly, and what would be acceptable variations,
9 a group of people who do understand how to test would set up a program of different means of testing outdoor testing,non reflective chambered testing, the standard software for testing would be Holmimpulse, all microphones would be individually calibrated that would be able to be made or done by the DIY community
I'm not sure blaming the user for a failure of the software is constructive. It's not me, my browser or my computer that can't get to the files, it's your software. I and my computer have no difficulty browsing to the file and/or downloading it. The address reported in the error message "\\www.gedlee.com\summa.txt" is formed as a local network url but maybe that's just the way the code formats its error reports. If remote file access is problematic for your software why not allow the files to be downloaded and opened locally? They are already publicly accessible from your website.
1) a practical minimum is about 1 meter, but the optimal distance depends on the specifics of the room. The hardest to control is the floor - thats why you need a stand and if you are going to make a stand then why not let it rotate.
2) You want the speakers response NOT ANY room effects.
3) what I proposed is both
4) you simply cannot surpress reflections to the point where they are not a problem. You have to use gating so you need to have both the mic and the speaker as far away from anything reflective or diffractive as possible. But this isn't hard - just move the furniture out of the living room like I do.
5) not practical. A "typical" enclosure (closed) with large radiused edges is going to be fully representative.
6) I don't see the issue here
7) Toole says 1/20 octave, I use 1/12, sometimes 1/20, 1/3 is way to smooth, 1/6 is marginaly acceptable.
8) I only see a use for polar response and impedance and anything less is incomplete.
9) calibration is overblown - a knowledge of the FR of the mic is good to have because cheap ones can get rather bad, but absolute calibrated levels are unneceesary. Outdoors is a bad idea because its rare to be able to do that - I never have been able to, but I do dozens and dozens of measurements. Indoors, outdoors, it should all be the same if done right. I have already standardized on Holm - but I am concerned that Ask is no longer available. What if Holm all of a sudden went away? Ouch!
so 1/3 octave is too smooth, but 250Hz resolution is okay?
It isn't possible to not get room or baffle effects if your doing it indoors and want a meaningful resolution. I do my measurements outdoors on a regular basis, if I want to see anything outside the treble.
I can't get your software to work, but I have no problem generating polars in Soundeasy. HolmImpulse can't do overlays?
It isn't possible to not get room or baffle effects if your doing it indoors and want a meaningful resolution. I do my measurements outdoors on a regular basis, if I want to see anything outside the treble.
I can't get your software to work, but I have no problem generating polars in Soundeasy. HolmImpulse can't do overlays?
I don't follow the top comment - smoothing and rsolution are two completely different things.
What is the error running the software?
I can guarantee that none of those packages does polar maps and not like I do them. Looking at a single frequency polar is kind of useless isn't it? You need to be able to look at polar spectrums.
Having a measurement library is good to have as a reference to compare other drivers under the same conditions. That said the unavoidable fact is for you to do your crossover you are going to have to create your own measurement set having the drivers mounted in your box. You need to account for baffle width, placement, spacing, physical offsets and so on. Generic measurements won't really get the job done.
Rob
Rob
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.