Hello,
I’ll soon be taking my first ever FR measurements (using HolmImpulse). I’ll be measuring the individual drivers in a 3-way system that is crossed at 500hz & 5khz, and I’m wondering about a few things:
1) Protecting the midrange and tweeter during a log sweep---
I naively thought that I could simply use the software to filter the log sweep during the individual measurements of the mid & tweeter. But it appears that that software filter is intended just for post-measurement display adjustment. So then, would a 12db/octave filter be the way to go? …for the mid, perhaps a 28uf cap in series, followed by a 4.7mh inductor in parallel….and for the tweeter, a 6uf cap in series, followed by a .40mh inductor in parallel. I pulled these numbers out of some 2-way crossover calculator, but don’t know whether they would apply to this situation.
2) Polar measurements with Holm---
It’s really nice that you can take some measurements with Holm and then, with a single click, save the graph as a GIF. However, it appears that Holm will only display three measurements at a time, no? Did I miss something? It looks like a lot of polar measurements are taken at 7.5 increments. So, if you were limited to only three polar measurements, what would they be? 0, 30 & 60 degrees? Some other combo? Or is three simply not enough to be useful for anything?
Thanks
I’ll soon be taking my first ever FR measurements (using HolmImpulse). I’ll be measuring the individual drivers in a 3-way system that is crossed at 500hz & 5khz, and I’m wondering about a few things:
1) Protecting the midrange and tweeter during a log sweep---
I naively thought that I could simply use the software to filter the log sweep during the individual measurements of the mid & tweeter. But it appears that that software filter is intended just for post-measurement display adjustment. So then, would a 12db/octave filter be the way to go? …for the mid, perhaps a 28uf cap in series, followed by a 4.7mh inductor in parallel….and for the tweeter, a 6uf cap in series, followed by a .40mh inductor in parallel. I pulled these numbers out of some 2-way crossover calculator, but don’t know whether they would apply to this situation.
2) Polar measurements with Holm---
It’s really nice that you can take some measurements with Holm and then, with a single click, save the graph as a GIF. However, it appears that Holm will only display three measurements at a time, no? Did I miss something? It looks like a lot of polar measurements are taken at 7.5 increments. So, if you were limited to only three polar measurements, what would they be? 0, 30 & 60 degrees? Some other combo? Or is three simply not enough to be useful for anything?
Thanks
SBA, download REW at the HTShack. The new Beta version can do like 18 overlays or something. I'm not really familiar with Holm. Maybe someone else can chime in on that.
REW is really easy to use.
Dan
REW is really easy to use.
Dan
One othe thing SBA, I go in 11.25 degrees as do some others. It will show you what you need. 7.5 is better, but more work and shows little more IMO. It just depends on how OC you are.
Dan
Dan
I do 7.5 only because my "turntable" for turning the speaker has lines 0, 15, 30, 45, 60, 75, 90 and its easy to find 7.5 incriments.
Holm is very clean and easy. Geddes has posted many times that its the first measurement package that he has switched too in 20 years. Its nice that its free and very, very easy to use. Sadly, the developer isnt around any more at all.
Im not sure REW is good at anything else then bass measurements. Maybe Im just not comfortable with it outside of subwoofer measurements. I do have the latest version too, it definitely has some cool overlay features including some nice CSD stuff. HOLM lacks overlays and CSD.
ARTA is another choice that I think is very powerful in the end thinking back though after buying WT3 and some other products I should have just purchased SoundEasy, its takes measurements much further then these free choices.
So the "compact impulse response" or lack of "tail" or other names for the same things should be called signal to noise ration at a particular frequency and SPL. Speakers with no drive make little noise (noise being anything including distortion not in the original signal) however, the noise increases with increasing drive. This idea names a way to actually specify the results with numbers rather than "see it is better!"
Have long measured signal to noise ratio and dynamic range at different SPL and frequencies to quantify driver behavior. This idea moves these concepts from pseudo-science into quantified data. Now please, I am not saying any of the non-quantified method is bad but again, it produces no numeric values so different transducers cannot be compared.
Just to help a little have attached a simple explanation of dynamic range WHICH IS NOT signal range. This is written for RF amplifiers but the concepts are valid for transducers as well.
Have long measured signal to noise ratio and dynamic range at different SPL and frequencies to quantify driver behavior. This idea moves these concepts from pseudo-science into quantified data. Now please, I am not saying any of the non-quantified method is bad but again, it produces no numeric values so different transducers cannot be compared.
Just to help a little have attached a simple explanation of dynamic range WHICH IS NOT signal range. This is written for RF amplifiers but the concepts are valid for transducers as well.
Attachments
Thanks SUM. IOW, that Behringer looks good at the volume it was measured at(which is actually quite loud by my normal listening standard, but I made no actual SPL measurements), but will look worse at higher volume and at some point will crash and burn. I'm sure we all understand that, but quantification would be nice so that we all know what we're getting. I'd bet money that it would hit a brick wall of self noise before most PA speakers even start to show signs of distress.
The thing that strikes me about the measurements I just took is how much they relate to what you said earlier. This:
Subjectively sounds so much louder than this:
The average of those graphs respectively look like this:
So is it the couple decibels in the 2-5k region, where the ear is sensitive, or all the additional self noise. In anycase, with that little difference in output, my ear tells me, or perhaps it's just my brains interpretation, that the DIY 2 way was considerably louder on the measurement stand. Enough so that I wanted to turn it down. The drivers also ended up being much higher off the floor d/t the size difference of the enclosures. I wasn't sure how to adjust the output so that total average was actually equal. Should I do that with an SPL meter and pink noise? "A" weighted, or "C"?
Thanks,
Dan
The thing that strikes me about the measurements I just took is how much they relate to what you said earlier. This:
Subjectively sounds so much louder than this:
The average of those graphs respectively look like this:
So is it the couple decibels in the 2-5k region, where the ear is sensitive, or all the additional self noise. In anycase, with that little difference in output, my ear tells me, or perhaps it's just my brains interpretation, that the DIY 2 way was considerably louder on the measurement stand. Enough so that I wanted to turn it down. The drivers also ended up being much higher off the floor d/t the size difference of the enclosures. I wasn't sure how to adjust the output so that total average was actually equal. Should I do that with an SPL meter and pink noise? "A" weighted, or "C"?
Thanks,
Dan
Last edited:
Thanks dantheman-
SPL measurements... very tricky question for broad band. First thought is the bandwidth would need to be the same so... a bandwidth limited measurement or (maybe better) drive signal. Then the audibility curves like "A" or "C" weighted might be used like you suggest or just bandwidth limited pink noise.
I recall calibration of frequency on a particularly (1.5B$) expensive install and it came down to me and pink noise. Everyone agreed after I did the calibration it worked. No one could come up with a better way so that is what was done. All the science between all of us failed us and it ended up being my calibration everyone accepted as the most accurate. Strange days for sure 9-11.
That aside I would pink noise, A weight noise, and C weight noise two speakers next to each other and flip between them quickly and decide when they seemed to be of equal overall loudness by ear and by the meter and see what comes out for data. Maybe someone will quickly come to a good conclusion.
At least there is a chance and hope the "equal loudness" could be determined.
=SUM
SPL measurements... very tricky question for broad band. First thought is the bandwidth would need to be the same so... a bandwidth limited measurement or (maybe better) drive signal. Then the audibility curves like "A" or "C" weighted might be used like you suggest or just bandwidth limited pink noise.
I recall calibration of frequency on a particularly (1.5B$) expensive install and it came down to me and pink noise. Everyone agreed after I did the calibration it worked. No one could come up with a better way so that is what was done. All the science between all of us failed us and it ended up being my calibration everyone accepted as the most accurate. Strange days for sure 9-11.
That aside I would pink noise, A weight noise, and C weight noise two speakers next to each other and flip between them quickly and decide when they seemed to be of equal overall loudness by ear and by the meter and see what comes out for data. Maybe someone will quickly come to a good conclusion.
At least there is a chance and hope the "equal loudness" could be determined.
=SUM
Dan & Doug, thanks for your replies.
Geez...there’s a lot of S#%& involved in measuring speakers.
I downloaded REW and found it to be a very intuitive, well documented, and cool looking program. However, I couldn’t even get past the first step, which is calibrating the soundcard (Echo Indigo I/O with an ASIO driver). The manual said to disable “playback monitoring“ to avoid getting a feedback loop; but the program wouldn’t recognize the ASIO driver, so I couldn’t get access to the soundcard to make input adjustments.
Going back to HolmImpulse I got the same feedback loop, but now knew, and was also able, to turn off the playback monitoring. [ Device & Control > Payback/ASIO control > uncheck monitoring, which will then allow a proper soundcard calibration.] Had I not read the REW manual, I would have never found such information at Holm. So thanks again.
I also saw in Holm’s Device & Signal section that there is indeed a log sweep filter. Previously, I had been looking at the bottom of the “measurement options” section and wondering why I couldn’t get those filters to work on the log sweep.
Geez...there’s a lot of S#%& involved in measuring speakers.
I downloaded REW and found it to be a very intuitive, well documented, and cool looking program. However, I couldn’t even get past the first step, which is calibrating the soundcard (Echo Indigo I/O with an ASIO driver). The manual said to disable “playback monitoring“ to avoid getting a feedback loop; but the program wouldn’t recognize the ASIO driver, so I couldn’t get access to the soundcard to make input adjustments.
Going back to HolmImpulse I got the same feedback loop, but now knew, and was also able, to turn off the playback monitoring. [ Device & Control > Payback/ASIO control > uncheck monitoring, which will then allow a proper soundcard calibration.] Had I not read the REW manual, I would have never found such information at Holm. So thanks again.
I also saw in Holm’s Device & Signal section that there is indeed a log sweep filter. Previously, I had been looking at the bottom of the “measurement options” section and wondering why I couldn’t get those filters to work on the log sweep.
High sba,
I had problems after calibrating with holm, make sure you do some measurements uncalibrated and also calibrated, and use whatever works better 🙂 I got very strange phase shifts and frequency spikes at low freq's after calibrating.
Also when measuring multiple drivers with holm, after measuring the first driver make sure you use the zero lock feature (can't remember off the top of my head). I think you have It detects time zero for the first measurement, and then you lock it at that, so that measurements of subsequent drivers which may not be time aligned are accurate with respect to the first one, as if it automatically adjusts for each driver, unless they are perfectly time aligned you will get phase shifts I think.
Tony.
I had problems after calibrating with holm, make sure you do some measurements uncalibrated and also calibrated, and use whatever works better 🙂 I got very strange phase shifts and frequency spikes at low freq's after calibrating.
Also when measuring multiple drivers with holm, after measuring the first driver make sure you use the zero lock feature (can't remember off the top of my head). I think you have It detects time zero for the first measurement, and then you lock it at that, so that measurements of subsequent drivers which may not be time aligned are accurate with respect to the first one, as if it automatically adjusts for each driver, unless they are perfectly time aligned you will get phase shifts I think.
Tony.
It's true, it is complicated. I think so far that reading this thread won't help get you started. You're welcome for the help and if you are interested in setting up a rig for polar graphs just ask. Wait till you get thing calibrated first. Sorry REW didn't work for you. I'm still learning how to use all the features of the old version. I for one love looking at how particular speakers/types of speakers measure. Learning what I've learned during the course of this thread, I know I need to go forth and do more measuring on my own design. I've just been lazy d/t their size/weight and my time.
Good luck and keep us posted,
Dan
Good luck and keep us posted,
Dan
I had a chance to briefly listen to the Behringer 2031A, just a quick listen with one channel out of the box. My initial impressions are:
1. First sound coming out, I could hear image depth, so that was a good sign. However, the image depth seemed to shift with the same instrument. I'm guessing mostly due to the room.
2. Cymbal attacks were very nice, but the trailing timbre seems a bit strong and irritating. Since the frequency plots only went up to 20KHz (or 15KHz?), such that I could not see the high frequency slope characteristics, so I could not figure out whether this was due to the class D amp or the metal dome tweeter.
3. The sound had just a very slight boxiness to it. It seemed to occur in a small low frequency range, probably close to 50Hz.
4. My guess at the CSD is that the initial 0.1ms range is not more than 6db lower than time 0. The rest probably decays pretty good, except for the tuned port.
If I can get a look at the frequency between 10KHz and 30KHz, I may decide to get the P version if it's well behaved in that region.
Looking at the data of 2031, there seems to be a mild breakup mode around 18KHz. This is probably what's causing cymbals to sound funny, but looking at CSD would probably be better. Looking at the impulse dan has posted. I'm going to guess the 0.3ms of the CSD is probably going to look like the Summa. I'm also wondering how close the sound of the Summa and 2031 are going to be.
Siegfried Linkwitz did a nice listening test comparison between his Orions and the Behringer's. See here:Spatial distortion
Looks like when they are properly mounted in the listening space they may be the speaker bargain of the year.
It seems like they have potential. But one would have to study it closer to see what modes need to be made. Cannot complain at that price, but I think some slight changes might improve things significantly with probably a 5%~10% change in cost (not price).
Browsing through what Siegfried Linkwitz posted, I get a feeling we are hearing the same things, just different way of expressing it. He attributes the cause to things different than I would.
Last edited:
I don't know whether they will allow me to test them or not. I called for a listening test first, and they never got back with me. So I just showed up at the store when I was in the neighborhood. They did not have the P version, and they did not know whether there was any left in Taiwan or not.
- Status
- Not open for further replies.