In another thread I posted my theory of using Fs as a guide for the lower cut off of a driver such as a tweeter or midrange. I posited that it's not so much the Fs we care about but distortion. Unfortunatley most manufacturers don't really post distortion specs.
The esteemed @AllenB posed the question as to whether or not this was really true and honestly I'm not really sure! 🙂 so I thought I'd ask. In a couple of drivers I've looked at this seemed to be a good rule of thumb. Take the plots from the Scanspeak 4" Illuminator 12mu/8731T00:
I happen to also have the distortion graphs from Zaph:
If this pattern holds then using a general guide of 2x FS is not a perfect substitute for distortion but it's not a terrible method either. As mentioned, I've not written an academic paper or thesis on the topic but to me it's fair to say this is a useful guide in the absence of all other data. The impedance can always be flattened, but not the distortion.
What do the more experienced members think of the statement?
Best,
Erik
The esteemed @AllenB posed the question as to whether or not this was really true and honestly I'm not really sure! 🙂 so I thought I'd ask. In a couple of drivers I've looked at this seemed to be a good rule of thumb. Take the plots from the Scanspeak 4" Illuminator 12mu/8731T00:
I happen to also have the distortion graphs from Zaph:
If this pattern holds then using a general guide of 2x FS is not a perfect substitute for distortion but it's not a terrible method either. As mentioned, I've not written an academic paper or thesis on the topic but to me it's fair to say this is a useful guide in the absence of all other data. The impedance can always be flattened, but not the distortion.
What do the more experienced members think of the statement?
Best,
Erik
Great question Erik!
As always, a difficult answer.
I investigated this myself some time ago. Here is the post
If it's TLDR, then I would summarize that yes, a quick and dirty way to aim to cross over at 2 x Fs. If you have no measurement equipment, than I would take this as a good-enough rule of thumb.
To empirically test this, one can do very low voltage testing at Fs, and IF you are able to, you can also determine maximum SPL.
But be careful! If you accidentally test at a high voltage- you can damage your driver:
The cursor position shows the maximum SPL of 77dB dB for 1% (-40dB) THD at 600Hz. The maximum SPL for 3% THD(-30dB) at 600Hz is 87dB (no cursor shown)
Recall that by first principles, a high pass filter with an acoustic LR4 response will reduce the SPL by 24dB at 1 octave below the crossover point.
ie.. acoustic LR4 1200Hz, the -6dB point is 1200Hz, and the -24dB point is 600Hz.
So in reverse, the maximum SPL is 77 + 24 = 101dB (with THD <1%), OR 87+24 = 111dB (with THD <3%)
Advantages: Avoid iterative testing at various crossover points and slopes. Can specify end points eg. 10%, 3% or 1% THD, thus testing is short in duration.
Disadvantages: Relies on some assumptions that may not be correct in practice. e.g. driver's behavior above or below Fs.
As always, a difficult answer.
I investigated this myself some time ago. Here is the post
If it's TLDR, then I would summarize that yes, a quick and dirty way to aim to cross over at 2 x Fs. If you have no measurement equipment, than I would take this as a good-enough rule of thumb.
To empirically test this, one can do very low voltage testing at Fs, and IF you are able to, you can also determine maximum SPL.
But be careful! If you accidentally test at a high voltage- you can damage your driver:
NOT RECOMMENDED
Do not try this at home
Do not try this at home
The cursor position shows the maximum SPL of 77dB dB for 1% (-40dB) THD at 600Hz. The maximum SPL for 3% THD(-30dB) at 600Hz is 87dB (no cursor shown)
Recall that by first principles, a high pass filter with an acoustic LR4 response will reduce the SPL by 24dB at 1 octave below the crossover point.
ie.. acoustic LR4 1200Hz, the -6dB point is 1200Hz, and the -24dB point is 600Hz.
So in reverse, the maximum SPL is 77 + 24 = 101dB (with THD <1%), OR 87+24 = 111dB (with THD <3%)
Advantages: Avoid iterative testing at various crossover points and slopes. Can specify end points eg. 10%, 3% or 1% THD, thus testing is short in duration.
Disadvantages: Relies on some assumptions that may not be correct in practice. e.g. driver's behavior above or below Fs.
Easy way to destroy your driver if not careful e.g. accidentally sending your drivers a few volts whilst testing at Fs can destroy it.
NOT RECOMMENDED
NOT RECOMMENDED
Might be worth pointing out that we're looking only at "small signal" distortion here. We assume distortion only gets worse as the signal amplitude increases up until the limits of the linear Xmax. That limit is another issue and true at any F.
D
Deleted member 375592
It is a bit more complicated. Let's leave distortions curves alone, they are of no help. Here we measure Focal ps130F, on music or music-spectrum-shaped noise excitation noise excitation applied, with varying HPF filters: 50:50:300, 80dBSPL RMS at 1m
and measuring residual. The results speak for themselves: too low freqs modulate the mids and highs.
BTW, John REW implemented FSAF measurement technique in his latest beta, so now you can check everything yourself.
and measuring residual. The results speak for themselves: too low freqs modulate the mids and highs.
BTW, John REW implemented FSAF measurement technique in his latest beta, so now you can check everything yourself.
Not really, but the old ROT of not crossing < 2Fs is a useful guideline for those who can't / won't measure & have no other data to fall back on. Not so much about distortion in those cases though as safety, especially for tweeters, as it reduces the risk of burning out the VC. Which I suppose is a form of distortion. 😉
Beyond that -all other things being equal, the lower the flat BW extends, so HD will be higher for obvious reasons. Which is why you sometimes have to be a bit careful when comparing HD plots, as it may appear some units are at a disadvantage, but in practice that can (can) vanish as soon as they're filtered properly.
I looked up Ps130f- is this a small 5” midwoofer with flax cone designed for car use?
This is a somewhat busy graph with what appears like 21 lines
Can you please expand on what I am looking at?
What is foAX80?
What are the three groups of curves?
Is there a one line conclusion which is “best” and why?
Member
Joined 2003
It is only coincidental that Fs and increasing distortion both occur at the lower end of a driver's frequency spectrum. They are not otherwise related, one is not a result of the other.
Right!! But that coincidence happens enough to be useful. Like the warning that "behind a running ball comes a running child." The ball is not the child, and yet...
Member
Joined 2003
At the risk of spoiling the party, I have to present some fundamental issues regarding the use of smaller HF domes at lower crossover points.
With almost all average 25 - 28mm dome tweeters the main deciding factor dictating max SPL is the surface area vs. frequency. The suspension on most tweeters isn't linear enough to allow for excursions greater than about 1mm pk-pk and even so, the VC lead in wires won't be reliable with that amount of repeated flexing unless they're braided and of sufficient length. The adhesives are also being stressed more and can fail sooner, which can further affect the VC connection integrity. Adhesives also dry out and deteriorate by themselves, becoming less flexible. The process is accelerated with harder driver use.
Dome rocking and VC shifting the gap can be another problem at lower frequncies around Fs during higher excursion. This can provoke VC windings to contact the upper pole plate, especially with tighter VC gaps and slightly offset VC alignment caused by production tolerances. I've seen this with various HF soft domes as the suspension ages and sags.
With only a few rare exceptions, the crossover point should be above 2000 hz with a 3rd order slope to play the speaker at any decent level of volume. With a 2nd order slope, that point needs to be raised 300 hz and even a little higher with a softer kneed LR2 slope.
If you want more volume, the most reliable solutions are to either build a 3 way system, use a larger diameter tweeter or raise the crossover point with a high order filter.
IMO, needing to push the tweeter harder and damaging it can translate into a greater monetary loss with higher end HF drivers, especially if no replacement VC is available. At a minimum you'll wear the tweeter out faster. There may be a few budget tweeters available which can play a bit louder at lower HP points and in those cases the risk may be worth the reward. Tweeters have come a long way over the years, so this would offset the argument a bit.
The big exception to all this would be when using a WG, which helps to increase sensitivity closer to Fs on the HF driver. The abrupt change in directivity on most 2 way designs with a larger cone midbass and dome tweeter does warrant a lower filter point.
Many times, better performance can be had for similar or less money with a 3 way system using lower cost drivers. It depends.on how elaborate the crossover needs to be, but the increased performance maybe worth it.
With almost all average 25 - 28mm dome tweeters the main deciding factor dictating max SPL is the surface area vs. frequency. The suspension on most tweeters isn't linear enough to allow for excursions greater than about 1mm pk-pk and even so, the VC lead in wires won't be reliable with that amount of repeated flexing unless they're braided and of sufficient length. The adhesives are also being stressed more and can fail sooner, which can further affect the VC connection integrity. Adhesives also dry out and deteriorate by themselves, becoming less flexible. The process is accelerated with harder driver use.
Dome rocking and VC shifting the gap can be another problem at lower frequncies around Fs during higher excursion. This can provoke VC windings to contact the upper pole plate, especially with tighter VC gaps and slightly offset VC alignment caused by production tolerances. I've seen this with various HF soft domes as the suspension ages and sags.
With only a few rare exceptions, the crossover point should be above 2000 hz with a 3rd order slope to play the speaker at any decent level of volume. With a 2nd order slope, that point needs to be raised 300 hz and even a little higher with a softer kneed LR2 slope.
If you want more volume, the most reliable solutions are to either build a 3 way system, use a larger diameter tweeter or raise the crossover point with a high order filter.
IMO, needing to push the tweeter harder and damaging it can translate into a greater monetary loss with higher end HF drivers, especially if no replacement VC is available. At a minimum you'll wear the tweeter out faster. There may be a few budget tweeters available which can play a bit louder at lower HP points and in those cases the risk may be worth the reward. Tweeters have come a long way over the years, so this would offset the argument a bit.
The big exception to all this would be when using a WG, which helps to increase sensitivity closer to Fs on the HF driver. The abrupt change in directivity on most 2 way designs with a larger cone midbass and dome tweeter does warrant a lower filter point.
Many times, better performance can be had for similar or less money with a 3 way system using lower cost drivers. It depends.on how elaborate the crossover needs to be, but the increased performance maybe worth it.
Well this may be all very true but also very specific. My hypothesis was much broader than that. 😉
Yeah, understood. I'm just pointing out some things which often don't get considered or are forgotten. The modern trend to push little domes far down in xover sort of puts less emphasis on better midbass. Thats where most efforts IMO should be placed when shopping for drivers, as it does most of the heavy lifting.
As an addendum / extention to post 5 above:
That bit is wrong. The LF distortion reduces when the driver is HP filtered, details / extent depending on the frequency & order. For a given power level maximum excursion occurs above the crossover frequency for HP filters > 2nd order, with LR4 providing the highest power-handling in excursion terms of any conventional flat-summing slope -higher orders, with their narrower transition band require the driver to stay flatter to a lower frequency, although they may (may -not necessarily invariable) claw some advantage back in terms of steeper electrical transfer functions, reducing how much power is getting to the VC. But it's generally good design practice to ensure that it's the LF units that run out of steam first, with the HF last.
@eriksquires Just think of all the sweating those Dynaudio D21s and D28s had to do with those 1st order filters. Then they put in all those complicated delay circuits which ruined the phase purity. I admire their crazy thinking but think of.how much better sounding of a speaker they could have put together using proper LR2 and elliptical filters along with some BSC and the all to necessary low Q notch on the midbass.
As expected from you, Scott, everything you wrote is correct, but as you start your back of the envelope calculations for a new project, do you not use Fs to understand the range of HP filters, with the expectation that you'd be avoiding excess distortion as a result?
Not meaning to tease you, just kind of wondering how builders like you go about making initial guesses/assumptions before the part even arrives. 🙂
That's exactly my point. Proper distribution of bandwidth with the LF dictating the xover is of higher priority when designing a higher class 2 way system.
Many times people focus on the tweeter first which almost guarantees other compromises to be made downstream. Something often overlooked is the interaction of the HF padding network with the tweeter impedance curve, messing with the FR. Even small tweaks to the pad can affect things enough to require changes to the HP filter.
Having steeper filter slopes in the area of tweeter Fs, the peak has to be addressed if keeping distortion under control is important.
Member
Joined 2003
I would agree for the most part. Perhaps a better ROT for 25mm dome tweeters would simply be to keep crossover 2kHz or higher, regardless of Fs. Tweeter can have Fs of 1kHz or 500Hz and be equally as bad <2kHz. I personally don't think anyone should be relating Fs to a ROT for crossover design, drivers need to be evaluated individually for best implementation.
Member
Joined 2003
Those tweeters are not particularly stellar performers either (by today's standards at least), they suffer the same fate at low frequency as most others.
Not really. Although that's just me -I don't speak for anybody else!
I agree, something like 2Fs is a useful ROT for those who don't / can't / won't measure, mainly for safety, but it's the FR, HD (IMD if possible, although at a pinch you can get a reasonable idea how that's going from the HD plots) & nominal Xmax that I tend to pay most attention to, along with the intended listening distance, target average SPL and desired headroom. Caveat to that is that I'm not a big fan of low order filters with relatively delicate drivers. For e.g., If I'm doing a trad. 2-way with a low-crossed mid-tweet, I usually default to 2nd order, partly because it fits with tradition (& I usually have a trad. XO frequency for such system targeted). For a more contemporary 2-way or > with dome tweeters etc., then I'll rarely use anything less than LR4; sometimes steeper if it brings some advantages on the polars / stopband suppression & the tweeter's power-handling & distortion performance can handle the requirements. Have to watch the GD of course, but that's another issue.
John (Zaph) actually mentions a good example of this in his tweeter test mash. The now-deleted H1211 (27TDC) can actually be crossed as low as its H1189 (27TDFC -same tweeter, but with a rear chamber) stablemate, even though Fs is about 80% higher. Once you flatten the impedance & filter it, distortion on the low end you're using is much of a muchness with arguably a small advantage to the 1211 as frequency rises, although that's largely off the radar. YMMV as always of course -we all do what works for us.