Cool project!
I was wondering if you have considered the driver setup seen here: https://www.diyaudio.com/community/threads/suitable-wg-xover-for-b-c-dcx50.387590/post-7100030
I was wondering if you have considered the driver setup seen here: https://www.diyaudio.com/community/threads/suitable-wg-xover-for-b-c-dcx50.387590/post-7100030
Great question!
This is one of those decisions during design and engineering where reasoning is the only tool and not enough facts are avaiable to be 100% certain (except if you have unlimited resources).
I think for the DCX50 design to work as expected (if im wrong please let me know), the crossoverpoint between the MF and HF driver needs to be high enough where the HF is already directional and barely gets directed by the horn. The reason for this constraint is probably caused by the fact that there is no smooth transition from the HF phase plug to the surface of the horn. You could say the HF phase plug is a separate horn.
With the high crossoverpoint and needed capabilities from 400hz and up, compromises are needed in the MF driver sellection and total performance. Especially when there is little posibility, in terms of resources, of having custom drivers.
Let me know if you have any suggestions or a different vision on this subject!
This is one of those decisions during design and engineering where reasoning is the only tool and not enough facts are avaiable to be 100% certain (except if you have unlimited resources).
I think for the DCX50 design to work as expected (if im wrong please let me know), the crossoverpoint between the MF and HF driver needs to be high enough where the HF is already directional and barely gets directed by the horn. The reason for this constraint is probably caused by the fact that there is no smooth transition from the HF phase plug to the surface of the horn. You could say the HF phase plug is a separate horn.
With the high crossoverpoint and needed capabilities from 400hz and up, compromises are needed in the MF driver sellection and total performance. Especially when there is little posibility, in terms of resources, of having custom drivers.
Let me know if you have any suggestions or a different vision on this subject!
Last edited:
I wish you the best of luck and determination! Very interested to see how this goes.
The DCX50 recommended HF crossover point is 10kHz, that wavelength is slightly longer than the HF exit width of ~14mm, so should diffract to "fill" a 2" (50mm) horn throat. I'd expect the recessed HF phase plug would also diffract HF up to it's 16kHz range at a roughly 90 degree angle, something like the red arrow lines depict here:
That said, I've seen no HF polar response of this driver on any horns to confirm that.
I'd think the multiple channel paths available for your mid and high frequency drivers would cause destructive interference and resonant peaks in the response:
I'd really like to see the sensitivity, raw on axis and polar response of your prototype on the SB H280 horn, and the SPL limits before diaphragm to phase plug contact.
Print looks nice!
Art
Thanks for the insights! What would then be the reason the crossover would be 10khz instead of for example about 4khz in the DCX464? Would it be power handeling of the HF driver?
What would your choice be when developing a diy coax CD? DCX464 or DCX50 design concept?
Thanks for pointing this out. Would this be caused by MF sound reflecting back from the start of the HF path at the HF driver? For example: the length of the HF path to the MF port is 15 mm. This means that there would be a reflection of MF sound 125 deg out of phase at 4 khz. Have not thought of this potential problem during design.I'd think the multiple channel paths available for your mid and high frequency drivers would cause destructive interference and resonant peaks in the response:
Yes me too! I only printed the HF section without MF ports to first evaluate HF performance and set a baseline to get good info on the effect of the MF ports on the HF. I have planned to make the first measurements this weekend (work, parter and life don't leave much time for audio stuff), so stay tuned!
On such junction there is impedance change, part of sound reflects back where it came from and part transmits through. So, some of sound from HF driver goes to the mid driver, some reflects back to HF driver and only some get out toward device throat. Sound that went back towards drivers reflects there and goes back toward the junction and same thing happens again.
Same thing happens within regular compression driver, only part of sound enters a slot and rest exists later from another slot. Inside a compression driver next slot is typically closer than 1cm, so most of the sound exists relatively fast, mqybe within 2cm or so. On the combiner though, reflections back inside a device is roundtrip so very long extra path before sound fully exit.
When wavelength is long enough that most of the sound has exited within say 45deg phase there is about no effect, but at higher frequencies there is some. Fundamental problem is to cover wide bandwidth of sound (very long and very short wavelenghts) with one static size physical object. This same problem limits performance of any type of loudspeaker, and you'd just work around it somehow so that it isn't that audible.
You likely need to experiment with the channels. Just imagining how it would work out gives you few variations to get started and build more intuition. Copying existing designs would be good, combiner very close to transducers with particular sized openings. Another could be to combine further down the device, like in a MEH. I haven't thought or simulated this any further, but hopefully there is some food for thought. You might be able to play games with impedances at the junction using absorption like melamine foam to "block highs" some. Have fun!🙂
Same thing happens within regular compression driver, only part of sound enters a slot and rest exists later from another slot. Inside a compression driver next slot is typically closer than 1cm, so most of the sound exists relatively fast, mqybe within 2cm or so. On the combiner though, reflections back inside a device is roundtrip so very long extra path before sound fully exit.
When wavelength is long enough that most of the sound has exited within say 45deg phase there is about no effect, but at higher frequencies there is some. Fundamental problem is to cover wide bandwidth of sound (very long and very short wavelenghts) with one static size physical object. This same problem limits performance of any type of loudspeaker, and you'd just work around it somehow so that it isn't that audible.
You likely need to experiment with the channels. Just imagining how it would work out gives you few variations to get started and build more intuition. Copying existing designs would be good, combiner very close to transducers with particular sized openings. Another could be to combine further down the device, like in a MEH. I haven't thought or simulated this any further, but hopefully there is some food for thought. You might be able to play games with impedances at the junction using absorption like melamine foam to "block highs" some. Have fun!🙂
Last edited:
The DCX50 tweeter power handling is low due to the 32mm voice coil, and lower frequency output limited by the SD of the small annular diaphragm.
It's frequency response below 10kHz is poor, likely because of the HF entering the mid channels and resonating.
I'd prefer the DCX464 over the DCX50 design concept.
The DCX 354 and 464 use what Bennett Prescott calls an "impedance mis-match device", an important part of the design to allow much more overlap region between the mid/high elements.
That presents a high impedance path to the HF, restricting it's entering the MF channel.
I overlooked you (sort of) also planned to use the "impedance mis-match device" concept in your design:
That said, your HF side openings of the MF ports look too large (over 50% of the area is open to the HF), the entire cross section area of all the slots could be reduced to a fraction of the size, which would also raise the MF's acoustic bandpass frequency.
Ideally, the MF and HF would each cover only ~half the bandwidth.
Good idea, I also did that with the SynTripP in the design phase.
Unfortunately, forgot to evaluate the effect of the oversized bass reflex ports on the upper MF response until years later 😢
The XT25SC40-04 spec sheet lists it's Xmax at only 0.1mm, only about 20% of most HF compression drivers, it will be interesting to see how it does.
Art
First results
HF Plane Wave tube
19.4 mm inside diameter tube
1 meter long with 30 cm dampening cloth
imm-6 mic as close to the exit as possible in the tube section
I'm lost on how to evaluate the results and make improvements on the design. from 10k and up is effected by the PWT inner diameter. there are some dips and peaks between 2 and 6k. then a dip at 8k. Could the length of the channels compared to the "horn" area be the cause like below or am i completely talking BS?
HF Plane Wave tube
19.4 mm inside diameter tube
1 meter long with 30 cm dampening cloth
imm-6 mic as close to the exit as possible in the tube section
I'm lost on how to evaluate the results and make improvements on the design. from 10k and up is effected by the PWT inner diameter. there are some dips and peaks between 2 and 6k. then a dip at 8k. Could the length of the channels compared to the "horn" area be the cause like below or am i completely talking BS?
The compression ratio of the MF with lavoce 3 inch is already 6.64:1. i would need to go higher in the compression ratio to make the ports smaller. Ill try to tinker the design some what in cad to make them smaller while keeping compression ratio in check.
Thats one of the drawbacks of using off the shelf drivers. they need to be small for this design to work. Also, the current goal of the CCDIY is to be used in home audio so huge powerhandling and Xmax is not that critical.
A properly terminated PWT should not be responsible for dips and peaks between 2 and 8k.
For comparison, an EV DH3 1.25" dome diaphragm compression driver on a 1" PWT:
Some diaphragm breakup above 9kHz, and second harmonic distortion is around -25dB below the 150dB SPL level (~ 5.6%) with 2 watt input.
Yours has around half the distortion at ~45dB less output with only a tiny fraction of a watt input.
The iMM-6 test mic has a Max. SPL for 1% THD @ 1000Hz of only 127dB, you would need a mic capable of lower distortion at higher levels for PWT testing of drivers that have much output.
Hard to tell the cause of the peaked response from only one test with no comparison to the raw driver.
Displacement (Xmax times Sd) directly affects the clean output level achievable from any diaphragm, a critical design factor.
Doubling displacement is equal to +6dB more SPL.
The XT25SC40-04 has an Sd of 8 cm^2, times .1mm Xmax, = .8 cubic cm displacement.
As an "off the shelf" comparison, the SB26ADC-C000-4 dome tweeter has an SD of 6 cm^2, times .6mm Xmax =3.6 cubic cm., 4.5 times the displacement, over 12 dB more clean output potential. More than an order of magnitude difference between the two domes.
The SB26ADC-C000-4 has been used directly attached to a waveguide and measured with low distortion at "home audio" levels, achieving perhaps around an order of magnitude (-10dB) below "pro" SPL levels :
You may want to define the engineering benchmarks you are trying to achieve with your coaxial compression driver in terms of SPL, bandwidth and directivity.
Art
The ports will be very long which cause for a big change in expansion rate at the horn entry, with such a "big" driver compared to the XT25. would this be a problem because of diffraction at the horn entry by impedance change?
thanks for giving direction.
SPL: 100 dB should be fine for home audio
Bandwidth: 400 Hz to 18 kHz, just like DCX464
Directivity: no clue, consistent at least.
Next measurements will be with the H280 horn for directivity evaluation
I wasn't really suggesting that particular "big" driver as a co-axial choice, simply pointing out how limited the output of the XT25 is compared to others of similar diaphragm size.
Those goals should be easy to reach, though a co-ax driver is not required to produce only 100dB/1m from 400 to 18kHz.
The purpose of my comment was to provide the limmiting factor which resulted in sellecting the XT25. For now my main priority is to get the concept up and running. Will note it for later improvements.