Speaker build measuring questions

Hello all.
I've built from plans before and some enclosures, but this is my first scratch build, and I have some questions.
The build is with a pair of Radian 5210 coaxials, in a 3ft^3 box, and a 4" vent. I have factory SPL graphs and the factory T/S numbers, but I decided to go all the way (within limits) and bought a Dayton DATS3 to get more precise measurements to work with. I also set up my UMIK-1 and ran some sweeps today, super close, right flush with the baffle to try to eliminate as much room interaction as possible lol. I tested in an unlined box, and next I'm going to test again with some decent degree of lining to tame things down a bit.
Here's my questions
1. I took T/S measurements both free air and mounted in my test box. Which set should I use for the crossover inputs?
2. The SPL graph is somewhat different from the factory test (these were matched and tested at Radian), which is understandable since my conditions are so different and they may have either tested free air of infinite baffle, I don't know. But does it make sense to keep playing with lining the box and adding fill until the reading looks as good as possible, and then use those plots in Vituix? It makes sense to me, but I don't know. I'll post some bad photos I took as I was struggling along today, FWIW. The first photo shows the factory results compared to mine, apologies for the amateur graphic arts display. The second is the woofer obviously, the third is woofer/tweeter, and the last includes the port. The reason you see two lines for each in the later photos is because I measured with and without a bunch of panels surrounding the box/mic to see if there was any improvement in taking some reflections out, at least in the higher regions. The most dramatic difference was in the 12-20k region, where the panels definitely smoothed things out a bit.
Thanks in advance.
Bryan
71304467044__90CE9E9C-D8C8-4589-ADEB-0C49A4049E62.jpeg
71304068296__89DFCC59-81B6-4DDF-BDC4-0246BCCC662E.jpeg
71304314181__D86298B9-F67A-4404-BD4F-7DF9EB06D95A.jpeg
71304122984__AAC82565-E321-40D2-BA3B-04F7F6F05252.jpeg
 
For crossover design, I like to measure with the microphone about 18 inches from they baffle next to each driver. I actually have a stick I use to set the distance each time. This gets the distance and thus the timing / phase the same for each driver. I really stuff a box so it is very dead. I hate resonances and reflections.
 
  • Like
Reactions: 1 users
Nearfied measurements are only part of the solution.

You actually want the baffle response in your measurements, otherwise you won't be able to compensate for diffraction effects from the baffle and arguably more important - baffle step rise / losses.

You need to measure at distance with an impulse gate. This removes the room reflections but allows you to see the baffle step losses and diffraction caused by the baffle.
 
  • Like
Reactions: 3 users
For crossover design, I like to measure with the microphone about 18 inches from they baffle next to each driver. I actually have a stick I use to set the distance each time. This gets the distance and thus the timing / phase the same for each driver. I really stuff a box so it is very dead. I hate resonances and reflections.
Thanks I'll remeasure out as you suggest, and as Dave is saying I'll want the baffle step in there. I just have to figure out how to do a gated impulse in REW, shouldn't be too hard (I hope). This is a huge world to dive into even though I've been studying for a couple of years. Even just diving into PCs again after 15 years of Macs is tough, but part of the reason for this build is to learn the process from one end to the other. I can always refine things as I learn more. I appreciate your responses guys.
 
......so after reading many pages of Scott Hinson's work, and searching and reading on the topic of both ground plane and nearfield measurements, I've realized how difficult it would be for me to perform the GP with anything close to accuracy, since I don't have a mobile system I can take into a big lot or gymnasium etc. And doing this in the house in a 16'x18' room seems like a real compromise. Nearfield/farfield also seems fraught with compromise, "don't trust these numbers below/above etc".
I've written to Radian and asked them to provide me with the exact conditions under which these drivers were measured, hoping I might be able to make some basic assumptions about baffle step etc if they tell me they're measured in a 1cu ft AR box, or an infinite or IEC baffle. Since the drivers were individually tested and I have the results, I could trace them into VCad and maybe adjust for some assumed baffle step and go from there?
It doesn't look like it's ever going to be easy to subtract the room for now, until I put together an outdoor rig, which I just can't do this summer.
Any suggestions for a "second best" approach like what I'm describing?

By the way, kudos to all you guys who've really learned this stuff and know what you're doing. I'm working my way in that direction, but it seems like there might not be enough years left in my calendar to become proficient! The more I learn, the less I know kinda thing.
 
Don't discount your ears. You can do quite accurate assessment of FR if you know what your hearing. Differences of a half decibel in the mids are not hard to discern. I've designed a few successful builds using only my ears. Its not easy, but it can be done with patience and taking breaks in between listening. The key is using a simple SPL meter to get the levels equal. The rest is all in your ears. Not optimal, but possible.

That being said, nowadays I couldn't fathom working without some form of measuring setup. Its so much more efficient anf repeatable, considering you have all sorts of free software available, needing only a simple audio interface and a basic 50 dollar mic with cal file.
 
  • Like
Reactions: 1 user
......so after reading many pages of Scott Hinson's work, and searching and reading on the topic of both ground plane and nearfield measurements, I've realized how difficult it would be for me to perform the GP with anything close to accuracy, since I don't have a mobile system I can take into a big lot or gymnasium etc. And doing this in the house in a 16'x18' room seems like a real compromise. Nearfield/farfield also seems fraught with compromise, "don't trust these numbers below/above etc".
I've written to Radian and asked them to provide me with the exact conditions under which these drivers were measured, hoping I might be able to make some basic assumptions about baffle step etc if they tell me they're measured in a 1cu ft AR box, or an infinite or IEC baffle. Since the drivers were individually tested and I have the results, I could trace them into VCad and maybe adjust for some assumed baffle step and go from there?
It doesn't look like it's ever going to be easy to subtract the room for now, until I put together an outdoor rig, which I just can't do this summer.
Any suggestions for a "second best" approach like what I'm describing?

By the way, kudos to all you guys who've really learned this stuff and know what you're doing. I'm working my way in that direction, but it seems like there might not be enough years left in my calendar to become proficient! The more I learn, the less I know kinda thing.

I've had trouble getting a reflection free groundplane.

I usually rely on the nearfield + baffle step applied plus splice to gated farfield measurement approach. I'm happy my speakers sound "uniform" and certainly measure that way when a groundplane is done or I can get the speaker high enough to get down to say 100Hz showing there isn't a rise in the response (and bafflestep is appropriate).

1. Take a nearfield of the woofer
2. Simulate the bafflestep from the bafle (use something like Jeff BAgby's tools or The Edge)
3. Subtract 2 from 1. Jeff's tool actually does this for you if you load 1 and 2 into his spreadsheet
4. Take a farfield gated on the woofer (no reflections)
5. Splice 3. with 4. using Scott's guide about where best to do it

Voila - you have your "quasi anechoic" woofer response, complete with baffle step, realistic low end response and ripple with baffle diffraction (what you want / need).
 
Last edited:
  • Like
Reactions: 1 user
"Don't discount your ears." Malarkey, human hearing perception is fraught with poor performance...The same shortcomings with human hearing also extends to human vision, human taste, and unfortunately, reasoning. This is why we all should rely on the discipline of science...science doesn't care if we are afraid of spiders, high places, the water, dark nights, doesn't care if it's hot outside, rainy, cold...doesn't get depressed, tired , sad, frustrated, bored...all the foibles of the human being.
No, no one is going to be able to notice a one-half decibel change in sound pressure level. We have machines and science to hear better, see better....we never would have made it to the moon if engineers said, "sounds about right to me"


-------------------------------------------------------------------------------------------------------------------------------------------------------------------------Rick...
 
I've had trouble getting a reflection free groundplane.

I usually rely on the nearfield + baffle step applied plus splice to gated farfield measurement approach. I'm happy my speakers sound "uniform" and certainly measure that way when a groundplane is done or I can get the speaker high enough to get down to say 100Hz showing there isn't a rise in the response (and bafflestep is appropriate).

1. Take a nearfield of the woofer
2. Simulate the bafflestep from the bafle (use something like Jeff BAgby's tools or The Edge)
3. Subtract 2 from 1. Jeff's tool actually does this for you if you load 1 and 2 into his spreadsheet
4. Take a farfield gated on the woofer (no reflections)
5. Splice 3. with 4. using Scott's guide about where best to do it

Voila - you have your "quasi anechoic" woofer response, complete with baffle step, realistic low end response and ripple with baffle diffraction (what you want / need).
Thanks a lot Dave, this sounds like the way to go. I have a couple of questions:
1. What kind of distance would you suggest for farfield? 3m? 10m? 1m? And would .4ms be a good place to gate the measurement?
2. I also have a port, which obvioously I have to measure nearfield, but the question is about how to sum it with the woofer. I've read a couple of places where the author adjusts the level, I think Scott Hinson does as well. But if the mic level and distance is the same for the nearfield woofer measurement and the port, shouldn't the relative SP be already correct?
3. Where do you measure the tweeter? Can't really gate this measurement, is that right? I guess I'm not too concerned with things being a bit out of wack above 10k because I'm probably using 2nd order that will have more of an effect from around 10k to the xover point (1100-1200). Does that make sense?
 
Hi Bryan,

Re 1. Most advice says at least 3 times the largest driver diameter to be measured to get reasonable baffle behaviour and summation characteristics when doing multi-driver measurements. In practice many use 1m. If you are doing groundplane then you usually double to 2m

Re 2. you measure the port with the mic centreed on the port just on the outside of the box (i.e. don't breach the port with the mic). The levels are likely different and you need to bring down the port so the low frequency tails align between nearfield and port measurements. I think REW can do the merge for you. I can't recall specifics but I'm sure one of Scott's papers will be far more useful here.

Re 3. Tweeter measurements are always farfield gated (1m as per point 1). you want to see what is happening up top. Nearfield makes no sense for tweeters as they are playing in a range where you want to see what the baffle is doing (and below this tweeters have no output if the crossover is implemented correctly). I'm talking conventional 1.5Khz and up crossover points. (dome / ribbon tweeters).

for your farfield measurements - you want to keep the voltage (amplifier output) and mic in the same place. that is - mic on tweeter axis - measuring the tweeter, then switch without changing any amplifier setting or mic position to measure the midrange then midwoofer etc...

are you using a USB mic or XLR mic with loopback? As this affects whether you use relative or measured phase for crossover design

Edit: be careful with tweeters without any low frequency protection. You can use a higher voltage (e.g. 1v+) if you begin your sweep at a safe frequency (e.g. 400Hz). I personally do a full frequency sweep but I keep amp voltage to 0.5v which all dome tweeters I've tested so far can handle. The only implication here is the noise floor goes up, so maybe not useful for distortion measurements, but perfectly fine for frequency response measurements (in a quiet room)
 
Thanks Dave. I'll remeasure today hopefully according to your info.
In answer to the mic question, I'm using a UMIK-1 with USB, and the correction file installed in REW.
I'm trying to recall the suggested cutoff for gated tweeter measurements, would .4ms be OK, or should it be a bit longer, .5? I seem to recall .4-.6 being the suggested range.
I've also added some internal damping to the enclosure, pretty modest but the least I would expect to use. I think I see some noticeable cab resonances in the first set of measurements I took, so hopefully this will tame things down a bit and make the crossover design less difficult.
Last question: I took impedance and T/S measurements with the Dayton DATS3, both on the bench and in the box. The bench numbers are closer to the Radian spec sheet, but wouldn't I want to use the measurements taken with the box?
 
Hi, few quick tips: be sure to know if the USB mic ruins timing between measurements or not, and how reliable / useful the resulting data is then for crossover work. VituixCAD measurement manual and user manual are great resource for successful measurement procedure at home condition, doesn't mean couldn't be done otherwise as well, you just need to know caveats of what you are doing and you'll be fine. VituixCAD can do any processing of the data, including merging nearfield responses with far field responses, baffle simulation and so on. Manual(s) and software available here https://kimmosaunisto.net/

Quick observation from data so far, your port has very loud output below 200Hz, as loud as the helmholz resonance itself. It also shows up in the other measurements and will certainly affect sound quality, I'd say it ruins it. You'd want that peak at least 10db below the main helmholz resonance, preferably lot more down but might be hard without making new box, perhaps system plan altogether.

Don't consider these let downs, important stuff! Your system will sound better the more you try and fail and try again!:) been there done that, slowly getting better sound. Remember to have lots of fun with the hobby!:)
 
Last edited:
  • Thank You
Reactions: 1 user
I'll dig into the Vituix manual today before proceeding. The port dimension was determined with Eminence sim software by the supplier of the drivers, so I considered it a starting point and was hopeful it might be OK! Right now it's 100mm dia with no pipe, just the 18mm ply thickness (baffle is double thickness above the port). I can cut a smaller port 'donut' and apply it over the existing port and see what happens for comparison, maybe 75mm, this should tell me something? I could also just drill a new port hole just below the existing one. But first I'll see how it measures now with some damping, hoping that will at least smooth the peaks, but it probably won't reduce the amplitude much I'm guessing unless I pack a lot more material into the base of the cabinet, does this all sound correct?
REW and VCad are really amazing tools, but not easy for relative beginners like me.
Note: When the guy at USSpeaker ran the sim for box size and port size, he also said this would only work if I 'stuffed the box half full'. I was trying to see if I could get an 85 liter enclosure to work and extend the bass, but I know this driver really is optimized for a smaller enclosure, something like 30l.
 
Hi, if the port is so short then the peak is resonance within the enclosure. I suspect the enclosure is ~180Hz wavelenght / 2 tall, ~90cm or so, and port is close to bottom or top? You might be able to reduce the peak leaking out by moving the port further from ends of long dimension, also if driver was middle of the long dimension. Perhaps some heavy damping other end of the box would work. Hope it helps! have fun!:)
 
Close! The enclosure is 98cm, but I did place the port closer to the center of the baffle to get it out of the higher pressure near the base, and also to allow for filling if needed. I may have to add more damping to the base, I'll see how my measurements today compare against the naked enclosure.
It looks to me like the three distinct port peaks at 46, 180, and 550 hz are also present in the woofer levels, and assumed these were all box resonances? That's why I figured I'd keep adding damping until they smooth out enough. If I have to put a couch cushion in there I know it's time for a different enclosure!
 
"Don't discount your ears." Malarkey, human hearing perception is fraught with poor performance...The same shortcomings with human hearing also extends to human vision, human taste, and unfortunately, reasoning. This is why we all should rely on the discipline of science...science doesn't care if we are afraid of spiders, high places, the water, dark nights, doesn't care if it's hot outside, rainy, cold...doesn't get depressed, tired , sad, frustrated, bored...all the foibles of the human being.
No, no one is going to be able to notice a one-half decibel change in sound pressure level. We have machines and science to hear better, see better....we never would have made it to the moon if engineers said, "sounds about right to me"


-------------------------------------------------------------------------------------------------------------------------------------------------------------------------Rick...
I suspected getting a reply like this. This isn't a black and white type of argument. I'm not advocating using just your ears to design a speaker or throwing away your test equipment. I'm just saying that its possible to do more with your ears than you think.

You can absolutely hear half a decibel in the midrange if you tried. If you can't, that's too bad and I'm sorry you don't have the finesse to do so. Just don't say others can't just because you can't. I rely on my ears for my livelihood and wouldn't have a roof over my head if I didn't hear as good as I did.

All I'm saying is, give your ears more credit than they get, as most people are capable of hearing more details than they think. Just don't use absolute statements based on your perception alone. This isn't a discussion about rocket science or quantum physics. Its about speaker building, which is a hobby for most of us.

Your ears are the reason you enjoy audio. Everyone's ears are different, but they decide for you, not just specs and measurements. Just looking at graphs and specs alone doesn't tell you the absolute performance of a speaker. It will tell you alot, but not the entire picture. Driver specs or even common measurements won't tell you what a finished design will sound like in the final product. Only relying on measurements to tune your crossover won't guarantee it will sound good to YOUR ears.

For example, just because people tell you the Harman FR curve is the gold standard doesn't mean it sounds good to everyone's ears. I like a little bit of a BBC dip in my designs, so those mid-forward, in your face 80s recordings won't cut off your head at higher volume levels.

Of course you won't likely be successful in designing steep slope, complex passive filters without the help of measurements and simulation tools. Your just making hearing comparisons and correlating those with your measurements. Thats really the main thing I'm referring to here.
 
  • Like
Reactions: 1 users