So far the conversation has revolved around TS parameters. I'd like to ask about some other microphone related measurements.
I tend to measure distortion in order to set some limits on my intended crossover points. I am almost always using drivers that other people have measured and used in a design, so I'm not doing critical tests on whether to actually select the driver or not.
But beyond harmonic distortion plots, do people rely much on other measurements such as impulse response and waterfall plots? Is this for crossover design, or just determining if the driver is adequate to use or not?
I tend to measure distortion in order to set some limits on my intended crossover points. I am almost always using drivers that other people have measured and used in a design, so I'm not doing critical tests on whether to actually select the driver or not.
But beyond harmonic distortion plots, do people rely much on other measurements such as impulse response and waterfall plots? Is this for crossover design, or just determining if the driver is adequate to use or not?
Let's review the goal of this thread
Ok, regarding Bl(x) and Kms(x): The vast majority of online driver reviews and driver tests do not give us any information about Bl(x) or Kms(x). So this parameter is not particularly helpful in selecting a driver. Only in the case where we might have several drivers under consideration, and all of them have published Klippel data, would this parameter be useful in selecting the driver.
So let's assume we have a driver, and we can measure Bl(x) and Kms(x). How do we use that information to design a better speaker? Does it change the crossover simulation? I don't think so, but i am open to the possibility. Does it change the way we model low frequency behavior? Change the vent size? Perhaps. I would like to hear the thoughts from @b_force and others on how you use Bl(x) in your speaker designs, and what would be the performance impact if you had to design a speaker without that information?
I also think it would be helpful to show us some test procedures or diagrams on how this test is performed. My understanding is that it is rather difficult, but I could be misinformed.
In that case;
- nearfield measurements with the woofer
-> those include distortion measurements (to determine any break up issues or other issues if you don't have those yet)
-> Burst decay measurements (= waterfall in periods). Useful to find out where the internal cabinet resonances will be.
This will give you all information you need to determine the crossover.
The original posts was quite open about just "selecting the best drivers" for a specific project.
The way I was reading it, wasn't in particular for just ONLY the crossover filter.
See point (2) of the goal of this thread.
The more tight BL(x) and Kms(x) are, aka less shift in either Fs and (Zmax-Re), the less shift there is in the low-end system response as well as distortion (in general, not specific).
The unfortunate side of the story, is that we don't have all information available.
So in that case we have to investigate driver performance ourselves.
Which I also consider part of finding and selecting the right speaker for the task.
Which unfortunately will cost us money
But hopefully we can all help each other with this as well!!!
To answer your questions;
- No it doesn't crossover simulation
- It can changes the way you want to design a box, I can show this best with some simulation graphs, but will do this later (don't have time now )
- The performance impact is rather asking the question the other way around.
Again, with linear I don't mean just distortion, I mean actual change in system response!
Frequency Response:
When selecting drivers, we examine the manufacturer's frequency response curves. These are typically made at 0, 30, 60 degrees off axis, and sometimes 45 degrees is included. If we are fortunate, there is a third party test result available online... @HiFiCompass sets the gold standard here, in my opinion.
These curves are sometimes measured in 2pi (infinite baffle conditions), but more often they use a standard IEC size baffle. Sometimes the test reviewer has a non-standard size baffle which is used for all the testing by that reviewer.
Of course the best option would be to purchase every single driver we might be considering, and test them all ourselves. Then discard all the ones we don't want to use. For most projects, this is quite unrealistic.
So I would like to discuss what we should look for, what is our criteria, when looking at frequency response curves... either from OEM or from third party test reviews.
j.
When selecting drivers, we examine the manufacturer's frequency response curves. These are typically made at 0, 30, 60 degrees off axis, and sometimes 45 degrees is included. If we are fortunate, there is a third party test result available online... @HiFiCompass sets the gold standard here, in my opinion.
These curves are sometimes measured in 2pi (infinite baffle conditions), but more often they use a standard IEC size baffle. Sometimes the test reviewer has a non-standard size baffle which is used for all the testing by that reviewer.
Of course the best option would be to purchase every single driver we might be considering, and test them all ourselves. Then discard all the ones we don't want to use. For most projects, this is quite unrealistic.
So I would like to discuss what we should look for, what is our criteria, when looking at frequency response curves... either from OEM or from third party test reviews.
j.
Testing frequency response with USB microphones
Some of us use a USB microphone to conduct frequency response measurements of drivers installed in cabinets. I did it this way at one time.
I don't think it is too controversial to say that a 2-channel setup with an XLR mic provides more information and enables high resolution polar response measurements. But some people use the single-channel USB mic, and they have their reasons for doing so.
So I would be very interested if those folks would discuss their methods, procedures, and equipment when using a USB mic. I would like to hear how they implement their data into simulations, and any other considerations or observations they have...
Thanks!
Some of us use a USB microphone to conduct frequency response measurements of drivers installed in cabinets. I did it this way at one time.
I don't think it is too controversial to say that a 2-channel setup with an XLR mic provides more information and enables high resolution polar response measurements. But some people use the single-channel USB mic, and they have their reasons for doing so.
So I would be very interested if those folks would discuss their methods, procedures, and equipment when using a USB mic. I would like to hear how they implement their data into simulations, and any other considerations or observations they have...
Thanks!
Voice Coil Magazine (Vance Dickason) and some others have measured drivers's Bl(x) symmetry etc. parameters with Klippel system for many years. Some of those are public domain, some only for registered (for free)
https://audioxpress.com/page/Voice-Coil-Magazine
https://audioxpress.com/page/Voice-Coil-Magazine
I very rarely look at freq resps these days, but go straight into distortion graphs.
As long as these look fine and smooth, the rest if often also pretty good.
The fact that someone uses a different size baffle is totally not important for the performance of the woofer.
We also don't have to buy each driver to figure this out.
Therefor IEC size baffle measurements (or equal) or perfectly fine to use and gives us a fair comparison with measurements from other people.
What you should look for in general;
- A nice even response that doesn't look like an alpen landscape
- Distortion graph is totally depended on the goal of the project, but in general one that doesn't rise up all of a sudden, and ideally with no weird peak/spike around de 800-3000Hz area.
Also, I rarely look at absolute distortion numbers, but just look for a nice smooth distortion graph without to much weirdness in it.
Most important of ALL of this combined: Check multiple sources for data!!!!!
Not all manufactures (read: most) have fair frequency graphs and 3rd party sources sometimes have a specific baffle or have little mistakes.
I personally always look at Hobby Hifi first, followed by Voice Coil Magazine, https://hifi-selbstbau.de/ and HifiCompass.
Or otherwise sometimes http://www.dibirama.altervista.org/ or Klang und Ton (also German magazine)
Especially Hobby Hifi, Hifi-selbsbau and Hificompass seem to be very close and consistent.
With Voice Coil Magazine you have to be VERY careful with odd, non-standard graphs and sometimes bad mistakes.
If none of this is available, I sometimes have a look at forums, although the reliability of measurements can often be questionable.
There are a few exceptions from users that I can trust and have the experience/knowledge.
Erin has a couple on his website as well btw!
I used just single channel measurements for the longest time back in the day with WinXP.
There were some small minor artifacts, but nothing so crazy that it would give different results.
With the latest Windows 10/11 even with ASIO drivers delay/latency seems to be all over the place.
Making single channel measurements so much a hit-or-miss that I just find it flat-out unreliable anymore.
A dual channel audio interface can be bought for about 40-55 bucks these days, so I don't really see good reasons why someone wants to take the risks?
So I will try to stay on topic ...
Impedance measurement - are there any resonances in the targeted frequency range?
My most important measurement - THD over frequency ofer level. Simply a few frequency responses at different level. THD tells us what the limits of the river are! How deep can a tweeter play. How high the midrange. Which resonances have effects for lower frequencies. Does it scale properly to low levels. Does there occur power compression?
It's not about the sound of the driver but about the usability. And when a driver excells here - it at least doesn't sound bad.
The hard thing is to keep the distance perfectly the same over wide angles, otherwise phase gives errors ...
Make out of phase measurements between drivers to check perfect phase addition - easy and precise.
I use an Audio Precision system for most measurements but export to VituixCad can be a pain, so for some measurements I also use RAW.
And break in IS an issue with SOME drivers. So just do it, it doesn't hurt.
TSP of course to estimate volumes and Q. As I mainly use closed cabinets variation over power and movement is not as tragic as with portet/other higher order enclosures. Always make a prototype with a portet enclosure and drive it to the level you need!
Impedance measurement - are there any resonances in the targeted frequency range?
My most important measurement - THD over frequency ofer level. Simply a few frequency responses at different level. THD tells us what the limits of the river are! How deep can a tweeter play. How high the midrange. Which resonances have effects for lower frequencies. Does it scale properly to low levels. Does there occur power compression?
It's not about the sound of the driver but about the usability. And when a driver excells here - it at least doesn't sound bad.
Impedance and frequency response in the ACTUAL CABINET at different degrees.
The hard thing is to keep the distance perfectly the same over wide angles, otherwise phase gives errors ...
Make out of phase measurements between drivers to check perfect phase addition - easy and precise.
All of them are needed for a good design. I aim for no resonances in the audible bandwidth, low THD cuase it for sure is a good thing to have and even off axis behaviour cuase these speakers are less sensitive to different rooms. I prefer close distances between drivers (coax fan).
I use an Audio Precision system for most measurements but export to VituixCad can be a pain, so for some measurements I also use RAW.
I always do impedance measurements at different levels. Does it stay stable, how big is the change?
And break in IS an issue with SOME drivers. So just do it, it doesn't hurt.
Well I think first, you’d have to set up some accepted standards around transducer type…..compression drivers and horns aren’t going to pass the impedance sweep/resonance tests that we’d normally apply to cone drivers.
I’d agree with the consensus here so far where close mic frequency response, distortion and impedance sweep along with T/S parameters get you 90% of the way when selecting drivers for a design and purpose.
Next step for me is to use the best measurement system I can……my ears. Using my DAW and a multi pEQ plug in, I set the intended passband of the driver on a test baffle and just listen…..first with a tone generator to look for the obvious offensive stuff and then complex music……..95 db dynamic peaks with an average around 85-88db….pretty typical of todays studio created content.
My pretext here is I’ve already got the bass covered from 20hz up to 250hz with 4 subs and DSP where I can stack everything on top of this flat foundation. That and the room decay is minimal already so I know I’m hearing the driver and not the room.…..I’m a voicing guy
My above process has revealed some startling surprises over the years and I consider myself fortunate to have a background in audio engineering to have experienced the offensive stuff that is embedded in the recordings themselves…….you can’t erase those with playback……just mitigate them. A great example is todays preferred tilted response curve……it avoids most of the baked in crap that exists on an average……and adjustable HF trim is really nice to build in at home for those recordings that a really well done.
I’d agree with the consensus here so far where close mic frequency response, distortion and impedance sweep along with T/S parameters get you 90% of the way when selecting drivers for a design and purpose.
Next step for me is to use the best measurement system I can……my ears. Using my DAW and a multi pEQ plug in, I set the intended passband of the driver on a test baffle and just listen…..first with a tone generator to look for the obvious offensive stuff and then complex music……..95 db dynamic peaks with an average around 85-88db….pretty typical of todays studio created content.
My pretext here is I’ve already got the bass covered from 20hz up to 250hz with 4 subs and DSP where I can stack everything on top of this flat foundation. That and the room decay is minimal already so I know I’m hearing the driver and not the room.…..I’m a voicing guy
My above process has revealed some startling surprises over the years and I consider myself fortunate to have a background in audio engineering to have experienced the offensive stuff that is embedded in the recordings themselves…….you can’t erase those with playback……just mitigate them. A great example is todays preferred tilted response curve……it avoids most of the baked in crap that exists on an average……and adjustable HF trim is really nice to build in at home for those recordings that a really well done.
Last edited:
The other one requires a scope and a frequency generator?Solutions which require (1) a computer and (2) and audio interface for it
I don't know but I think basically any person these days had a computer/laptop anyway.
Probably can even be done on an smartphone these days actually.
Unless you live in some kind of cave.
But than I wonder how one can communicate on forums like this?
There's a really simple quick and dirty way to measure impedance peaks and Fs points (not the amplitude itself). Simply connect a frequency generator with a 50 ohm resistor in series, listening for max volume at specific frequencies. It works very reliably and quickly. You can measure LF resonances this way as well. The peaks are located where the amplitude sounds the loudest. This method also allows you to hear any distortion from non linearities in the motor and other parts of suspension, surround and cone. You can even pinpoint the exact location of any buzzing and noise using a stethoscope with its main diaphragm removed (using only tube opening). I've found some very difficult to locate noises with this method.
Hi Hörnli,
So you say. I'm just being honest. I did warranty for speaker manufacturers, and you can melt/burn the voice coil adhesive (or worse) without knowing it. Or you can warp the voice coil former. So the driver can be damaged a fair amount before it fails completely. For driver testing, you need it to be non-destructive for hobbyists, and you do not want to change any parameters.
I can also test driver linearity and a number of other things that matter. We use an oscilloscope for that too. I had a workstation set up to test drivers and complete systems. There was a lot of experience gained in doing this.
So no. I am not gaslighting anyone here.
Hi b_force,
Most DVMs do not respond very high up in frequency. The better handhelds have a -3 dB around 100 KHz, bench up to 300 KHz for most. I would pay attention to your equipment in making measurements. You may know this, but many people not experienced with calibration haven't even looked at the performance data for the meter (or whatever) they are using. People new to this probably aren't even aware of this issue.
Anyway, if making measurements the method I use is extremely accurate, and easy to perform. I did it so often I made up a jig for the purpose.
So you say. I'm just being honest. I did warranty for speaker manufacturers, and you can melt/burn the voice coil adhesive (or worse) without knowing it. Or you can warp the voice coil former. So the driver can be damaged a fair amount before it fails completely. For driver testing, you need it to be non-destructive for hobbyists, and you do not want to change any parameters.
I can also test driver linearity and a number of other things that matter. We use an oscilloscope for that too. I had a workstation set up to test drivers and complete systems. There was a lot of experience gained in doing this.
So no. I am not gaslighting anyone here.
Hi b_force,
Most DVMs do not respond very high up in frequency. The better handhelds have a -3 dB around 100 KHz, bench up to 300 KHz for most. I would pay attention to your equipment in making measurements. You may know this, but many people not experienced with calibration haven't even looked at the performance data for the meter (or whatever) they are using. People new to this probably aren't even aware of this issue.
Anyway, if making measurements the method I use is extremely accurate, and easy to perform. I did it so often I made up a jig for the purpose.
Erin has Klippel tests and acoustic measurements https://www.erinsaudiocorner.com/driveunits/
HifiChoice has many acoustic measurements incl. CSD that are useful for evaluation and xo design https://hificompass.com/en/speakers/measurements
Good ole Jim Krutke measured hundreds of drivers http://www.zaphaudio.com/
https://loudspeakerdatabase.com/
https://data-bass.com/#/drivers?_k=ukga89
I have designed only dsp/multiway-multiamp active speakers, sealed or dipole. Driver selections are based mainly on distortion profile, directivity and volume displacement estimated to be needed. I have chosen "good" drivers from major brands that are on sale or second hand, and I'm happy with results. Search for the ultimate best is mission impossible...