Tweeter crossover point

Hi,

I'm new to the forum.

I have a pair of passive ATC SCM20 Pro PSL Mk2 studio monitors I wish to make active, but only if I can cross the tweeter at 1.2khz, 1.4 maximum. Reason being, the provided crossover is a well designed 3rd order butterworth at 2.1khz, and in my opinion, a properly implemented 3rd order butterworth is the best passive crossover for loudspeakers. I don't think I'd be getting enough of an improvement/one of the improvements I want if I just switched from passive to active. Besides a slightly quicker upper midrange, I'd like greater dispersion at the crossover point. Currently it's not bad, but if I could turn ~+-50 degrees horizontally to +-65-70, I'd be very happy.

The tweeter is a ferrofluid free dual suspension design, I think it's the only tweeter of its type. It's an underhung design with a 2mm coil in a 3mm gap - the gap is saturated (even magnetic field throughout). Theoretically it is capable of 1mm travel completely distortion free. I put 3khz for 1 second at 100w through the thing, which should have moved it 0.75mm, and it was distortion free. I assume due to the tweeter's design, 1.2khz with .75mm excursion would have a similar level of distortion as 3khz with the same excursion. 1.2khz moving 3/4mm makes 100db with a 1 inch driver, which is good enough for me most of the time. I'd move the crossover point up or switch to the passive if I needed more volume.

The only problem I think I might run into with the lower crossover point is power compression. I want to keep the greater performance with transients of active crossovers. The tweeter is 91.5db/w, quite efficient, but not exceptionally. Edit: to clarify, I don't want thermal compression to offset the improved transient response gained from using an active crossover, nullifying its benefit. As mentioned below, the improvement is small in the first place and I'd like to keep all of it (this isn't a cost-free endeavour).

Does anyone know how to figure how much more power I'd be sending to the tweeter with the crossover approximately 1 octave lower? I want to keep its average power to under 5 watts.

In case anyone is wondering, I think the resonant frequency of this tweeter is between 700 and 800hz. I will be testing it soon to confirm. Interestingly, there is no rear air chamber lowering the fs. This should be a good thing, like a sealed box woofer. I believe the low resonant frequency of this tweeter is due to the weight of the dual suspension. The magnet is very, very, very powerful - high quality neodymium weighing many ounces. At least 6.
 
Last edited:
It's a thing. It's a small thing, and an even smaller thing with a 3rd order butterworth. But it's still a thing.

Anyway, as outlined in my post, what I'd like to know is: roughly how much more power would be going to the tweeter with the crossover point lowered by an octave? I want to minimize the potential for thermal compression - details above
 
I don't quite get why you worry about power compression. My initial idea would be that not the average level would be the problem here, but the distortion-free headroom.

I also think bringing the X-over down could lead to side effects regarding the acoustic behavior of the baffle. Interference patterns will change and probably not for the better. Did you measure the polar plot of the speaker at 1 to 2 kHz?
 
As I described in my original post, at a crossover point of 1.2khz I can get 100db with 3/4 xmax. If I need higher levels, I'd switch to a higher crossover point, either electronic or passive.

I'm not sure if you've looked at a picture of the speaker, so I'll describe it: the baffle is flat for most of its width and almost 10 inches wide, with the tweeter centered near the top/the dome recessed a bit (as is done to boost the top octave a bit/keep things more directional up there, and reduce interaction with the baffle at lower frequencies). I haven't measured the polar response of the speaker, or the woofer/tweeter separately between 1 and 2khz to see which performs better, but I have looked at the audiosciencereview review of the Mk1 version (same woofer different tweeter crossed at the same frequency) which includes the polar response from bass to 20khz. It shows narrowing of the woofer starting a bit above 1khz. I also have done listening tests which corroborate my speaker narrowing like the one measured in the review

From what I've observed, a ~10 inch baffle is good for half space radiation down to at least 700hz - half wavelength. Even if I doubled the frequency to a full wavelength to be extremely conservative, 1.4khz is still within my target range.

Knowing what I stated above (tweeter's design, review, my observations), do you still think that it's likely the tweeter would be worse than the woofer for dispersion between 1.2 and 2.1khz?
 
Last edited:
Thank you Pano! Those graphs show for my usual genre the tweeter will be getting about double its usual power, and 4x its excursion if played at the same maximum level and the crossover reduced to 1.2khz. So headroom is reduced 6db, average power 3db

Edit: oops, 6db divisions. Make that ~7.5 and ~4
 
Last edited:
Knowing what I stated above (tweeter's design, review, my observations), do you still think that it's likely the tweeter would be worse than the woofer for dispersion between 1.2 and 2.1khz?
Well, a wider dispersion invariably leads to more diffraction at the baffle edge. And since the cabinet hardly is optimized for that, I can't rule out you end up with worse directional behavior than you have now. Trouble is, no crossover or DSP can fix poor directional behavior. So my suggestion merely boils down to: check it, and check it again.

[Edit]Some quick and dirty calculating. Baffle 10" wide would lead to on-axis interference dip around 1,4k (I noticed the tweeter is nearer to the top of the baffle, so that doesn't help either). I think The Edge could show you that. Off axis of course is another story.
 
Last edited:
Ah, "The Edge" is a tool for baffles. I'll look into that.

If there's a dip at 1.4khz, this active design I'm pondering might not be viable. If I could only lower the crossover point by 300hz to 1.8khz there wouldn't be much of an improvement - definitely not enough of one to justify all the trouble and expense of switching to active. Plus the potential for unintended consequences..

Though I could make a new box with a better baffle. Possibly even just a new baffle -I think it's removable.

It's a sealed 20 litre cabinet. Easy enough to design around, but not something I planned on doing. Mostly because I didn't think I'd need to

Anyway, this is a really, really nice tweeter which I'm sure would sound great crossed lower: I think it was a consideration during design, but ultimately they settled on 2.1khz.

In an interview, ATC said that the reason they don't use a lower crossover point for the tweeters in the hifi versions of their speakers is just consistency across product lines. Pro audio equipment (the good brands anyway) are more designed for long term reliability and indestructibility than hifi usually is - reason being, things can be in use for 12+ hours a day for years on end, with abuse sprinkled in here and there from (most often) accidents, and sometimes idiots. Using a higher crossover point reduces a tweeter's exposure to heat and movement which lengthens its useful life. Hifi equipment is used less and is less likely to suffer neglectful use or abuse, so the average heat and movement of a tweeter with a lower crossover point in a hifi system is still less than what a tweeter in the pro world is exposed to.

I have the pro version of this speaker. There is a similarly sized and more nicely finished SCM19 with a slightly warmer voicing, which is hifi. Both versions have similar tweeters. ATC makes two tweeters in house, the SH25-76 and SH25-76S. The model with S in its name, termed S-Spec, has a 33% stronger magnet and (I think) a narrower magnetic gap. It has lower distortion and is just over 3db more efficient, but otherwise isn't notably different. It's used in the SCM20/45/50/100/110/150 speakers. Basically everything from $6,000-$50,000. The less efficient version is used in their smaller speakers, and hifi versions up to the SCM40 (a sealed 3 way tower with an ~8.5" woofer).

I'm not sure of most of the specifications of the SH25-76S because unfortunately they're unavailable. But I do know that it has a very strong motor, is very, very well damped due to its dual suspension/associated mass, and that the ratio of its motor strength to moving mass is high compared to the woofer. Its BI must be at least 2.0, and moving mass is very likely under 1g.

So, the BI of the woofer is 8. Not extremely high, but its a huge 20mm gap for an underhung 8mm coil. To pull this off the magnet is somewhere around 300oz - its diameter matches the woofer itself. The woofer has a moving mass of either 47 or 57g (can't remember which exactly, saw it in a datasheet with many missing fields once). Anyway, its moving mass is twice a typical woofer's of the same diameter. For this purpose we'll go with 50g and BI of 8.0

Woofer's magnetic strength to mass ratio is 0.16BI per gram.

Tweeter is 2.0BI and 1g (or less) which is 2.0BI/g

2.0/0.16 is ~12

So the tweeter is about twelve times stronger than the woofer, which is the other reason I'd like to lower the crossover point (first is dispersion). I should say that I don't feel there is anything wrong with the quality of the 1-2khz range of the woofer, it's great. Almost perfect really, except for the aforementioned directionality. I just think it could be a bit more detailed coming out of the tweeter.

I'm going to do some polar measurements from 700-2800hz as soon as I can
 
Which will be some time this week, probably on the weekend. I have no software and only a SM58 right now, so the results will only be good for relative amplitude at different angles.

I'm thinking 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800, 2000, 2200, 2400, 2600, 2800hz

0, 15, 30, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90 degrees.

Woofer, then tweeter, then both through the included passive crossover.

I haven't done this before... Basically I plan to mount the microphone on the axis of the driver being measured, and at the center of the woofer when doing the speaker as a unit (as is recommended for listening to this speaker), and turn the speaker in the above increments, going through the range of frequencies listed above each time. I'll do everything at the same voltage, and with the speaker outside and at the edge of a table for minimal reflected sound. I'll be typing results into excel. I think it'll take a couple hours, maybe 3 or 4 with setup/takedown.

I can upload the results here, or a screenshot of everything if xls files can't be uploaded.

Any pointers?
Should I do anything else?
 
Last edited:
Do you have (access to) an USB audio interface? Then use an app like REW or ARTA (I prefer the latter). I should mention HolmImpulse too. This kind of software is suited for taking such measurements and can produce the desired signals in one go (sweep or periodic noise). Saves a lot of time. And all are usable as freeware, ARTA only needs a license for saving and combining results.
 
Yes, I do - I have an RME Babyface Pro. It's a pretty good interface, 117db SNR ADC, 114db SNR DAC. 4 channels in, 4 out.

I was going to use an online tone generator and use my scope to verify the voltages in case the amplitude changed with frequency (because of bad design lol)

Those programs seem interesting. Why do you recommend ARTA over REW for this?

Also, it occurred to me that it might be beneficial to measure the harmonic distortion of both drivers if I could, especially 2nd and 3rd in the 1-2khz range. Do you know if that's doable (for free)? I can borrow a ~$6k Manley microphone if a really clean mic is needed.
 
Last edited:
Yes, I do - I have an RME Babyface Pro. It's a pretty good interface, 117db SNR ADC, 114db SNR DAC. 4 channels in, 4 out.
That would do quite well. Even cheap interfaces from Behringer actually are good enough for speaker measurements.
Those programs seem interesting. Why do you recommend ARTA over REW for this?
I'm used to it. Also, ARTA has more options for measuring gear. But they are kind of alike.
Also, it occurred to me that it might be beneficial to measure the harmonic distortion of both drivers if I could, especially 2nd and 3rd in the 1-2khz range. Do you know if that's doable (for free)? I can borrow a ~$6k Manley microphone if a really clean mic is needed.
ARTA can measure harmonic distortion (HD). Modern insight is that higher odd-order distortions (5th and 7th) could be more problematic than 2nd or even 3rd. Microphones (when used within their level range) often don't distort that much. Linearity (or applied correction for linearity) is quite important when assessing HD. You could consider buying a not too expensive Dayton or Behringer measuring mike.
 
That would do quite well. Even cheap interfaces from Behringer actually are good enough for speaker measurements.

I'm used to it. Also, ARTA has more options for measuring gear. But they are kind of alike.

ARTA can measure harmonic distortion (HD). Modern insight is that higher odd-order distortions (5th and 7th) could be more problematic than 2nd or even 3rd. Microphones (when used within their level range) often don't distort that much. Linearity (or applied correction for linearity) is quite important when assessing HD. You could consider buying a not too expensive Dayton or Behringer measuring mike.

Looks like ARTA it is. I won't need to save and combine for my purposes will I?

I'm going to do the polar measurements before buying a mic, because if the baffle causes too much of a problem for the tweeter at around 1.4khz, I'm not entirely sure where I'm going with this project.

If the tweeter does well though, I'm going to spend the $80 to measure its THD and frequency response. For both my project and to put online, because from what I can tell, a lot of people are curious about this tweeter's measurements, but they aren't available.

For the tweeter, if the level of odd harmonic distortion remains low to the crossover point and up to the maximum level, the crossover point is good.
You say 3rd harmonic isn't as detrimental to sound quality as 5th and 7th? Do you know if there's a weighting that's used, or should I just give them a bit more consideration when looking at their level? They wouldn't be very important above 3.5khz, right?

To put online I was thinking of putting the tweeter in the middle of a 4x4ft sheet of plywood and measuring it on axis from 2m, moving 10, 20, 30, 40, 50, 60, 75 degrees horizontally. I'd give it 2 watts, start at 500hz and go up to 22khz (published upper -3db). I'd be outside on grass, the 4x4 foot board lifted 2ft/tweeter up 4ft from the ground. Minimal reflections otherwise.

For frequency response at different system powers, I'm thinking of doing descending sweeps at 5, 10, 20, 30, 40, and 50 watts. The frequency at which THD reaches 3% is the bottom of the range for that power rating. Does the program allow something like this (and quitting at 3%THD to avoid driver damage)?

And then there's the general THD measurements. Basically, if I'm measuring this tweeter, what is it I would need to do to be thorough? Do you see any issues with what I plan to do above? Would you add anything?