I have a DIY build that is effectively a Cura CA10 - the cabinets were remaindered when the company went out of business, and it seems they had PL18s and Seas H519 25TFFC tweeters. I managed to get new old stock of both, and I have them set up active with minidsp crossovers. I never found a crossover schematic for them. They've been 'on the bench' for a while but are now back in-system.
I've had a fairly simplistic crossover running on them with a very steep filter at 2.5k but I'm wondering if I could do better.
The tweeter nominally resonates at 1200, has stored energy between 1200 and about twice that, and it rolls off second order with -3db at about 1800 and flat at about 3500, it looks like Q is about 0.5.
I'm wondering if I could:
The PL18 cone geometry means that the 60degree plot diverges from about 1800 and the 30degree plot from about 2500, so I would like to keep the crossover as low as I can given the tweeter's limitations.
Does that sound sensible? I don't think the woofer is up to higher than 2.5k without directivity issues
The 'ears' on the tweeter are next to each other and the cabinet is cut for that rather than the more spaced out terminals on newer tweeters, otherwise I'd probably replace with an SBAcoustics or similar. I may yet take a saw to the cabinet, but I thought I'd try first.
Not sure how I'd approach a passive crossover for these without using the tweeter's rolloff and accepting all the stored energy and raised distortion, but that's not my immediate problem.
I've had a fairly simplistic crossover running on them with a very steep filter at 2.5k but I'm wondering if I could do better.
The tweeter nominally resonates at 1200, has stored energy between 1200 and about twice that, and it rolls off second order with -3db at about 1800 and flat at about 3500, it looks like Q is about 0.5.
I'm wondering if I could:
- use an LT to flatten the tweeter's response and push the natural rolloff down so its flat above about 1000
- then add a 4th order (might try 2nd) LR at about 2300, maybe a bit less.
The PL18 cone geometry means that the 60degree plot diverges from about 1800 and the 30degree plot from about 2500, so I would like to keep the crossover as low as I can given the tweeter's limitations.
Does that sound sensible? I don't think the woofer is up to higher than 2.5k without directivity issues
The 'ears' on the tweeter are next to each other and the cabinet is cut for that rather than the more spaced out terminals on newer tweeters, otherwise I'd probably replace with an SBAcoustics or similar. I may yet take a saw to the cabinet, but I thought I'd try first.
Not sure how I'd approach a passive crossover for these without using the tweeter's rolloff and accepting all the stored energy and raised distortion, but that's not my immediate problem.
The stored energy depends on the result you achieve, nothing else.
Your method seems a reasonable way to achieve it, perhaps not the simplest.
Your method seems a reasonable way to achieve it, perhaps not the simplest.
Whichever way I did it, the response would be the same. That is the goal and you shouldn't allow it to be anything other than what you decide it has to be.
I won't say you can't do it through a LT, I guess I prefer to avoid anything that runs through a stage of greatly increasing the low end on a tweeter.. plus the LT is unlikely to bear a relationship to the wanted target since many other things are taken into account when arriving at the filtering needs.
If you're asking how I would do it, I'd create a target curve and work the response to achieve it with an active or passive filter set.
I won't say you can't do it through a LT, I guess I prefer to avoid anything that runs through a stage of greatly increasing the low end on a tweeter.. plus the LT is unlikely to bear a relationship to the wanted target since many other things are taken into account when arriving at the filtering needs.
Last edited:
It won't, though - by the time the filters are applied, there will be nothing raising it at any frequency, because the impact of the crossover will more than cancel that of the transform, especially 4th order. Some care would be needed to ensure there is no artificial clipping but once you get into the floating point domain, there should not be a problem so long as the final gain is always well below unity, as I understand it anyway.
Well, huh, who would have thought of that? Getting an on-axis target is not really a problem with the PL18 if the eventual breakup is notched out. Its the off-axis continuity that's tricky, but that seems to be an issue in any 2-way with a 6.5" and an older/smaller tweeter.
Absolutely. SL himself promoted this use of the "Linkwitz-Transform". The LT is just a filter that cancels a second order HP response and replaces it with a different second order HP response.
What SL did was to change the tweeter's second order response into one-half of the LR4 filter (the LR4 is two identical second order Butterworth filters in series, e.g. each has Q=0.707) and then add the "other" BUT2 filter. The combination will have an LR4 acoustic response.
You can also do what you mentioned, but you would need to extend the tweeter's passband much lower than 1k, for example to 500Hz when you want a 2.3k crossover point. Don't even actually operate the tweeter without a HP filter and with this sort of LT applied because the gain around 500Hz will be pretty high. When you add the 2.3kHz LR4 there will overall be attenuation below 2.3kHz but it will have a shelved region. The reason this works is that when the passband corner frequency is pushed far below the intended crossover point Fc by the LT, the phase response is flat around Fc instead of changing with frequency. In any case, SL's approach of using an LT+BUT2 is more direct so I would recommend doing that instead.
Thank you Charlie - I did think of that but maybe more complex to set up, but it does have merit as well. I guess its not often one sees an LT used to raise the rolloff, but I suspect that's also of value in some sealed 2-ways if there's a subwoofer available, or for a midrange in a 3-way. I wish it was easier to estimate the Fs and Q from a graph for a tweeter - do you know of a model?
And yes - flat-to 500Hz would be about 16dB of gain. Getting 1 octave above the natural rolloff wfor a passive crossover is challenging, let alone 2. How important do you think it is to have so much leeway?
And yes - flat-to 500Hz would be about 16dB of gain. Getting 1 octave above the natural rolloff wfor a passive crossover is challenging, let alone 2. How important do you think it is to have so much leeway?
FWIW it seems that the MiniDSP LT calculator seems to match the curve on the Seas plot if I use Fs=1200 (the tweeter resonance) and Q=0.5, matching at 2k, 1k, 500. I'll go with that as 'good enough'.
You bought the tweeter, so you know the model. I would use the MFG datasheet's value for Fs and Q:
https://www.seas.no/images/stories/vintage/pdfdataheet/h0519_25tffc.pdf
They list Fs = 1200 Hz but unfortunately no Q value. You could try to compare the response of a second order highpass filter with Fc=1200Hz to the driver's response while adjusting the filter Q until you get a decent fit. It won't be perfect but you just need to get in the ballpark.