So I'm having a bit of trouble with getting any sort of useful output from a Fountek NeoCD3.5H ribbon tweeter (transformer-coupled) in a 3-way design. Just free-air testing the crossover design (shown below) revealed that the midrange is not attenuating at all (passing even 12-16kHz tones without attenuation), and the tweeter output is far less than expected.
The crossover design is a Linkwitz-Riley 2nd-order 3-way with a 6-7 ohm woofer impedance, 8 ohm midrange impedance, and a 7.2 ohm (flat in the usable range) impedance for the tweeter. The drivers are:
Woofer - Dayton Audio PA380-8 15"
Midrange - Fostex FE166En 6.5"
Tweeter - Fountek NeoCD3.5H Horn-Loaded Ribbon Tweeter
I've tried everything I can think of, hence why I'm here. I know it's probably some big, easy-to-spot type of thing, but I'm the type of person that "loses" his keys, only to have been holding onto them the whole time.
Cheers,
-Kev
The crossover design is a Linkwitz-Riley 2nd-order 3-way with a 6-7 ohm woofer impedance, 8 ohm midrange impedance, and a 7.2 ohm (flat in the usable range) impedance for the tweeter. The drivers are:
Woofer - Dayton Audio PA380-8 15"
Midrange - Fostex FE166En 6.5"
Tweeter - Fountek NeoCD3.5H Horn-Loaded Ribbon Tweeter
I've tried everything I can think of, hence why I'm here. I know it's probably some big, easy-to-spot type of thing, but I'm the type of person that "loses" his keys, only to have been holding onto them the whole time.
Cheers,
-Kev
I have looked at a few, but none of them have used this specific midrange and tweeter combination. I selected the drivers based on their high sensitivity and good impulse response, all right around the 95dB ballpark range that I'm attempting to achieve. I might try a different midrange (Dayton PA-165) that is a drop-in fit, but would require a notch filter around the crossover point, and different zobel values, as it has a much higher Le. It's almost as if the actual board layout is wrong (something connected where it shouldn't be), or the zobel isn't doing its job of presenting a matched impedance to the tweeter. The low-mid crossover portion works very well.
What tool did you use to design your LR2 cross-over?
Also I am wondering at the lack of provision for adjusting levels to the mids and tweeters, unless they're absolutely perfectly matched I would expect there to be some deviation in sensitivity driver to driver.
I'm using an LR2 3 way xo I designed 12 years ago in my system, because of disparate sensitivities I designed the mid and tweeter xo to run into LPADs. (This helped me to ignore the driver impedance variations over frequency for the most part)
Also I am wondering at the lack of provision for adjusting levels to the mids and tweeters, unless they're absolutely perfectly matched I would expect there to be some deviation in sensitivity driver to driver.
I'm using an LR2 3 way xo I designed 12 years ago in my system, because of disparate sensitivities I designed the mid and tweeter xo to run into LPADs. (This helped me to ignore the driver impedance variations over frequency for the most part)
I used the ERSE crossover calculator, along with a couple Zobel network calculators from a DIY site to flatten the impedance of the LF and midrange and allow for a fairly smooth transition. This is my first time ever using a tweeter with a matching transformer (the original design used an Audax tweeter that apparently doesn't exist anymore), so I'm fairly clueless with respect to that. They're both around the same sensitivity at the crossover region, but the actual DC resistance of the matching transformer is extremely low (as it should be).
The driver sensitivities are as follows:
Woofer - 96-ish dB
Midrange -94dB
Tweeter - 95dB
What sort of test equipment do you have on hand? It seems like you might actually want to measure the drivers to determine their TS parameters and perhaps a simple mic and sound card set up so you can do measurements as you refine the design.
I am wondering about the zobel across the mids, where does it become effective and is that frequency too low?
My first XO design was absolutely dreadful, I did not realize just how bad it was until I designed the second one which with minor refinements is what I use to this day.
What I am saying is you've got some work ahead of you and success may not be instant.
My concern is that the midrange driver is a couple of dB less efficient than the woofer - this may or may not in fact be true in the real world. The fact that the tweeter is probably slightly more efficient than mids means you should be able to get them to match well acoustically.
My approach was to use mids and tweets that were significantly more efficient than my woofers (horn system with diy Onken bass bins) - I then padded down the mids and tweeters to match using crappy LPADs. Going to fully active (tube based) very soon.
I would look at some alternative cross-over simulators online and would recommend modeling the current one in LTSpice if you are familiar with it or are willing to take on a bit of a challenge.
Worst case if you can consider active you could use one amp to drive the woofers and another with passive XO to run the mids and high, and use an active XO to divide between lows and mids-highs. (Hopefully not needed)
I am wondering about the zobel across the mids, where does it become effective and is that frequency too low?
My first XO design was absolutely dreadful, I did not realize just how bad it was until I designed the second one which with minor refinements is what I use to this day.
What I am saying is you've got some work ahead of you and success may not be instant.
My concern is that the midrange driver is a couple of dB less efficient than the woofer - this may or may not in fact be true in the real world. The fact that the tweeter is probably slightly more efficient than mids means you should be able to get them to match well acoustically.
My approach was to use mids and tweets that were significantly more efficient than my woofers (horn system with diy Onken bass bins) - I then padded down the mids and tweeters to match using crappy LPADs. Going to fully active (tube based) very soon.
I would look at some alternative cross-over simulators online and would recommend modeling the current one in LTSpice if you are familiar with it or are willing to take on a bit of a challenge.
Worst case if you can consider active you could use one amp to drive the woofers and another with passive XO to run the mids and high, and use an active XO to divide between lows and mids-highs. (Hopefully not needed)
Recalculated Le is ~.5mH, not the .053mH that the datasheet specified. I thought .053 was super low for a driver that hits 32 ohms at 10k. I'll see if a properly sized cap fixes the zobel, and thus, the crossover (otherwise, the crossover "sees" the midrange driver as a 16-18 ohm driver and throws everything off balance. Oops!)
Last edited:
Cut up some Rockboard 80 and got them lined. Cornwalls seem small in comparison.
So I spent six hours in XSim hell, got some pretty graphs and modified the schematic. Lower 2800Hz third-order crossover point to the ribbon, which reduced the amount of midrange cone breakup present in the response.
The following graph shows the response of the original crossover:
Aaaaaaaand this one is the much-improved not-lumpy-bumpy version with a lower crossover point, using a third-order high pass and a second-order low pass.
I decided to spring for a different midrange driver in the end, as the Fostex was just too unpredictable and difficult to work with electrically. Went with the Dayton PA-165. (Dirt cheap, good construction, why not?)
The following graph shows the response of the original crossover:
Aaaaaaaand this one is the much-improved not-lumpy-bumpy version with a lower crossover point, using a third-order high pass and a second-order low pass.
I decided to spring for a different midrange driver in the end, as the Fostex was just too unpredictable and difficult to work with electrically. Went with the Dayton PA-165. (Dirt cheap, good construction, why not?)
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.