Take it from a mixing and mastering engineer……your favorite recordings were most likely mixed/mastered on speakers that DO NOT display constant directivity. This trend in design has risen from the home theatre segment and IMO is destroying the imaging capacity of folks who prioritize it and ignore the immediate need of some minor acoustic treatment of their space. These directive designs often disconnect the music from the acoustic space and the listening experience exposes it.
Diffraction…..another buzz word without meaning. You’re NEVER going to hear a measured 2db near field measured diffraction ripple in the listening window. Keep the drivers flush mounted and time aligned…..don’t worry about the rest. Narrow baffles in relation to the drive element do matter though…..my Decades of work have consistently revealed that a 1” dome tweeter on a 10” baffle is a great way to suck all the 3D spacial capability out of a recording. Look at what Vandersteen and Totem do…..or better yet, listen to them if possible. Your reality on diffraction will be forever changed.
Lobing….there‘s a term that IS NOT a buzzword and to be understood and respected in the design. Wanna know what lobes?……every two way speaker that employs a round high frequency element with a round midwoofer. They sound good……like French fries……cause everybody likes French fries. But step out of the comfort zone and use an AMT or ribbon with narrow vertical directivity and all is revealed in the crossover region instead of the mishmash of the forward lobe. Yes…….you have to take a seat to get the best performance.….and that’s a good problem to have if you ask me.
Enjoy your design exploration and the build……try and ignore the buzz…..it’s nothing more than a distraction. If your heart is set on a low Fs dome, take a look at the Scanspeak 3004 with an Fs of 420hz or so. I got a pair a few weeks ago and they can be crossed very low with minimal odd order components and the small face plate will allow you to get them close to the 8” where you need to be.
This is basically the only thing I really look at.
Just the point where the distortion shoots up.
Keep in mind that the actual crossover point is around -6dB less.
Which halves the distortion.
For home/studio hifi I don't care about much else.
Maybe I will have a brief look at sensitivity and power to see if it's not totally out of the ordinary.
Keep in mind that HF content in most music isn't that big to begin with.
So with a average 2nd or 3rd order filter we'll be fine.
It's a bit of a different story when 1st order filters are being used.
But I never use those.
It's a totally different story for PA/sound reinforcement stuff.
Weird statement without context.
Diffraction effects can be quite substantial, depending on the size of the baffle, drivers and rounding of the corners.
Sometimes it's fine, like you said, although 2dB to much energy around 1-4kHz CAN be audible.
Sometimes it's not so fine and creates all kinds of issues.
We most certainly can't just call it a buzzword, it's a physical and fundamental aspect of acoustics.
I do agree that in some cases it's not nearly as bad.
At the same time, I also don't see people using this a buzzword to begin with, so I am curious where you get that impression?
I'm late to this discussion, so this may have been covered above, but the basic tension IMHO is between dispersion and power handling.
In an ideal world, the direct answer to this questions is whichever is higher, the point at which the lower driver becomes directional or distortion begins to rise.
That is, if the 6.5" becomes directional ~ 2 kHz then I'd not try to go any further down with the tweeter. Then the question is, how low can the tweeter go with low distortion? If 1 kHz, then the answer is still 2 kHz because stretching the tweeter downward limits the output due to tweeter Xmax increasing as F drops. If the tweeter starts distorting at 3 kHz it's worth going upwards even if this means you have less than ideal off-axis output.
I should point out that many successful commercial designs push up the crossover point for the sake of power handling, leaving you with reduced horizontal output at the crossover point. Check Stereophile's 2-way reviews. Even some more expensive 3-way speakers like the B&W's show less than ideal off-axis performance.
Another tool worth using is WInISD or an online calculator for SPL vs. displacement. Enter the surface area and Xmax and frequency and you will get the output. It's useful to say, compare crossiing the tweeter at 2kHz vs. 3 kHz to examine maximum linear output.
In an ideal world, the direct answer to this questions is whichever is higher, the point at which the lower driver becomes directional or distortion begins to rise.
That is, if the 6.5" becomes directional ~ 2 kHz then I'd not try to go any further down with the tweeter. Then the question is, how low can the tweeter go with low distortion? If 1 kHz, then the answer is still 2 kHz because stretching the tweeter downward limits the output due to tweeter Xmax increasing as F drops. If the tweeter starts distorting at 3 kHz it's worth going upwards even if this means you have less than ideal off-axis output.
I should point out that many successful commercial designs push up the crossover point for the sake of power handling, leaving you with reduced horizontal output at the crossover point. Check Stereophile's 2-way reviews. Even some more expensive 3-way speakers like the B&W's show less than ideal off-axis performance.
Another tool worth using is WInISD or an online calculator for SPL vs. displacement. Enter the surface area and Xmax and frequency and you will get the output. It's useful to say, compare crossiing the tweeter at 2kHz vs. 3 kHz to examine maximum linear output.
Last edited:
If distortion is dependent on excursion, and excursion is dependent on frequency, then why look at a measurement taken at a single level to determine frequency?
I'm not sure I follow you, Allen. Please pardon my lack of experience with this.
Are you saying that the desired SPL level should be determined first... then work back from there?
It was more meant as a question for the others who are advocating the process.
Personally, I don’t normally use distortion measurements in speaker design.
Personally, I don’t normally use distortion measurements in speaker design.
You just hope that there are no issues going on or a driver's distortion doesn't shoots up?
I notice that theme. Of course if the issue is nothing but distortion then is it even necessarily an issue ?
Before that though.. if you measure at different levels, isn’t the frequency going to change?
Before that though.. if you measure at different levels, isn’t the frequency going to change?
I don't know, but I have seen even quality drivers from like Seas with severe issues and creating a huge distortion spike around 500+1500Hz (can't remember the exact frequency).
While the freq resp doesn't look tso much out of place.
Lately I have been working with a coax prototype that had serious issues at 200Hz. Something sounded already off, but only with measurements you can pinpoint the problem.
The only way of knowing is to do distortion measurements.
To determine where to cross a tweeter, we have to look where the distortion ramps up.
The rule of thumb of 2x fs sometimes works, but not in all cases.
Freq resp doesn't change when drivers do weird stuff.
Sometimes peaks or other problems are evident, but that's definitely no always the case or obvious.
Besides, why guessing if we can just run the measurements?
No, I mean doesn’t the frequency that tips the excursion past a certain point, depend on the level you measure at?
Well if anything, the impedance peak is a better mark here.
But both don't mean much with waveguided tweeters or even less for compression drivers in sense of distortion.
That ramping up obviously starts a little higher with higher levels, but not by that much.
It's a little hard to explain in words, so I guess I will find some examples.
Doesn't change the fact that you can guess this from just the freq resp and impedance measurements.
Even less so with other issues as I mentioned.
I mean I do get where you're coming from. Pixel peeping distortion plots isn't that useful. But that's definitely not what I am talking about.
Distortion measurement is very usefull for choosing your driver. If you go for a high output speaker system, even essential, but while you construct a speaker, it doesn't tell you anything more. So during development, if you stay in the reasonable range for your chassis, distortion measurement is not needed. Anyway, with a finished design, you should run it to be sure nothing went wrong. On the other hand, that is the time you start your listening test's. If you don't hear distortion as an unwanted component and you are pleased with the product, what does a distortion measurment tell you? Bragging right's?
And you just magically know?
I have tested and seen well over a hundred measurements (probably a lot more) at this point and there are still situations that throw you off guard.
And again, why taking the risk if the information can be gathered that easily?
That makes just absolutely no sense to me at all.
Let alone giving that as advice to other people.
Absolutely! Content such as classical/orchestral with high dynamic peaks will most often expose a driver or design not up to the task. Also consider the listening distance and desired levels at -6db per meter from typical 1w/m measurements.
This is why I highlighted directivity earlier……in an ideal world where one would choose a low crossover point for a tweeter to smoothly transition directivity, a compression driver would be used. That’s not to say a robust dome can’t do it…..but there’s other concerns now with SPL. Thermal and power compression also come in to play……..the dome may exhibit low distortion products but still compress from heat and motor stress. Domes are great…..but they have their limits. I’ve crossed 1” or so domes to 8” midwoofers but much prefer 8” midranges and cross them up around 3khz…..let the mid driver do it’s thing…..for me it easily trumps the perceived effects of the so called beaming.
Maybe we come from different worlds. When I started building speakers and did my own crossover's, there was no disttortion measurement system at any sane price. We talk about the price of a nice house in a suburb. So I had to use my ears, that I educated with live music, amplified and hand made.
Today, any school kid can measure anything, so educated ears may have become obsolete. That may explain the audio products people use today.
So my opinion "if it sounds right, it does what a good speaker should do" is outdated.
Maybe speaker that measure well, sound better than those that are a pleasure to listen to, but that would be paradox.
Today, any school kid can measure anything, so educated ears may have become obsolete. That may explain the audio products people use today.
So my opinion "if it sounds right, it does what a good speaker should do" is outdated.
Maybe speaker that measure well, sound better than those that are a pleasure to listen to, but that would be paradox.
No I know that world as well.
I just don't see why I would take a risk for 5 minutes more measuring time, or just browsing through some 3rd party measurements?
Even if it totally proves there are no problems.
If it all sounds well can only be determined after the fact.
Does it hurt or so to spend a tiny bit more time and effort?
Please do share the underlying point we seem to be missing... I've grown weary of the suspense 😛
Tools and learning evolve. When I first learned crossover design the idea of an affordable PC measurement system was laughable. Only manufacturers and university labs could afford even the most basic tools beyond a pink noise analyzer. Even commercial outfits like Infinity threw caution to the wind, tuned by ear and came up with some outrageously bad impedance curves as a result. They still made money.
Now we live in a time where there's at least 3 different free crossover simulations, many calibrated microphones and affordable and open source measurement tools available for the interested hobbyist.
The beginner only measures on-axis. She/He gets advice from others about drivers that may work well together and listens to some designs and decides to build a fresh design. Maybe they don't sweat distortion because a trusted ear or other designers used similar crossovers with their new tweeter. Later, seeking more independence and creativity you start worrying about off-axis response, and distortion, and at some point depending on your application power and maximum SPL become important as well. The needs of a PA or home theater speaker for a medium to large room are very different than a desktop. Buying $20 worth of drivers you may just whip it up fast, but if you spend $60 on the tweeter suddenly you start to invest more time and effort in getting the final design just right.
At the end of the day however this is a hobby site and you do what's enjoyable for yourself and if you learn about more ways to examine your designs and choices and you start to incorporate that into your design that's great. If not, great.
Lastly, a great place to find distortion measurements is in the no longer updated http://www.zaphaudio.com/ web site.
Last edited:
Maybe you got me wrong. I measure what ever I can, distortion as well.
It is only during the development process, where distortion measurements do not help me that much. I keep an eye on the tweeter I use and try to avoid the region the distortion starts. Maybe that is because I carefully plan my projects, before actual building takes place.
An example:
If I know that k3 at 1200 Hz and 95dB hits the ceiling, I simply don't use it in that region at that level. I can be pretty sure that anything will be OK, if I cut it at 2.500 Hz 12 dB/oct and the woofer will not be louder than 102 dB. So, the combination will fit. In the end, distortion measuremens make sense again, but only should show what I already know. If they are far off, I will ask myself why and what I have done wrong.
Anyway, at some point the actual sound of the combination comes into play, where the ears give the final judgement. Even if the result is distortion free, it may still not sound right to me. I build speaker for listening to music, not measuring samples.
Distortion measurements stress the component and the builder. Running the whole bunch of tests after any change in a capacitor, coil or resistor value is unrealistic for me.
I must admitt that there are chassis I don't like and use after buying and trying them. If things get too complicated, I quit and use an alternative. You will not find me building a passive 24dB crossover, decorated by three notch filters. It may measure fine in the end, but in most cases such extreme crossovers make the speaker sound dead. I'm not pointing fingers, but not all expensive chassis of Scandinavian origin need a bucket full of Jantzen parts. As profits on crossover parts are huge, compared to those on chassis, there sometimes may be a non technical interest in filtering the hell out of a speaker.
I'm not going to make a 25$ tweeter linear by using 125$ of crossover parts, I simply get a better tweeter that may cost 50$ and needs 20$ of parts in the end.
It is only during the development process, where distortion measurements do not help me that much. I keep an eye on the tweeter I use and try to avoid the region the distortion starts. Maybe that is because I carefully plan my projects, before actual building takes place.
An example:
If I know that k3 at 1200 Hz and 95dB hits the ceiling, I simply don't use it in that region at that level. I can be pretty sure that anything will be OK, if I cut it at 2.500 Hz 12 dB/oct and the woofer will not be louder than 102 dB. So, the combination will fit. In the end, distortion measuremens make sense again, but only should show what I already know. If they are far off, I will ask myself why and what I have done wrong.
Anyway, at some point the actual sound of the combination comes into play, where the ears give the final judgement. Even if the result is distortion free, it may still not sound right to me. I build speaker for listening to music, not measuring samples.
Distortion measurements stress the component and the builder. Running the whole bunch of tests after any change in a capacitor, coil or resistor value is unrealistic for me.
I must admitt that there are chassis I don't like and use after buying and trying them. If things get too complicated, I quit and use an alternative. You will not find me building a passive 24dB crossover, decorated by three notch filters. It may measure fine in the end, but in most cases such extreme crossovers make the speaker sound dead. I'm not pointing fingers, but not all expensive chassis of Scandinavian origin need a bucket full of Jantzen parts. As profits on crossover parts are huge, compared to those on chassis, there sometimes may be a non technical interest in filtering the hell out of a speaker.
I'm not going to make a 25$ tweeter linear by using 125$ of crossover parts, I simply get a better tweeter that may cost 50$ and needs 20$ of parts in the end.