DIY Lightweight 5-meter tall Delay Tower PA System

The pole deflects under load but shows no signs of instability, cracking, or uncontrolled oscillation.

Carbon fiber is fine, until it isn't. It doesn't undergo plastic deformation under load before failure like metals experience. CF might be OK for a horn.

Using only half the original 24′ carbon pole, the system benefits from a significantly higher load margin — likely 3–5× the nominal safety factor, since we're working within the thickest and strongest sections of the pole. The moment arm is shorter, and the flexural stiffness higher, reducing the risk of dynamic failure.

Likely 3-5 times isn't a statement you are capable of defending if you ever end up in court.
 
maybe a round and angled piezo array down to 1.8khz, like with this one would be lightweight enough and help the dispersion of highs and mids.
 

Attachments

  • images-3.jpeg
    images-3.jpeg
    3.5 KB · Views: 14
I’m hoping to generate a lighter-weight horn design with comparable or slightly better frequency performance than the Goldwood GM-450PB, ...
My takeaways are to pay more for a better driver, one designed for midrange with a bigger coil, such as the B&C DCM420. Also to consider a folded horn, or perhaps a longer one
I think you should accept advice from weltersys and buy TOA TU651/TH660 combos - they are better performer than yours JBL D250-X/Goldwood GM-450PB, with much more low frequency extension.
I am not aware of any commercial folded horns for B&C DCM 420, BMS 4592, BMS 4594, or available data for folded horn DIY printing.
 
Last edited:
  • Like
Reactions: EvenLouder
My takeaways are to pay more for a better driver, one designed for midrange with a bigger coil, such as the B&C DCM420.
Wrong takeaway.
A larger diameter coil does not mean more displacement, smaller diameter "PA" drivers have much more excursion potential which can offset the larger coil size.
A larger coil can handle more power, but at the low end, the driver's displacement limit, rather than power will limit output.
Each doubling of displacement allows +6dB more low end output.
Also to consider a folded horn, or perhaps a longer one.
Optimally, the horn length would be one wavelength long, 4.5 feet for 250 Hz.
Folding reduces the volume the horn occupies considerably, and if the corner reflectors are correctly done, can allow good reproduction up to the acoustical bandpass limitations imposed on a driver capable of adequate excursion for loud lower speech range out put.
I also had the impression that the folding was a compromise-laden way to simulate a longer horn with out taking up a longer space. Is that right, or does the folding have its own benefits?
The potential to array with adjacent mouths and the acoustical LP reducing upper harmonic distortion are other benefits, but in general, folding high frequency horns is an acoustic compromise for volume and material conservation.
I think you should accept advice from weltersys and buy TOA TU651/TH660 combos - they are better performer than yours JBL D250-X/Goldwood GM-450PB, with much more low frequency extension.
I'm not saying they are the ultimate or best solution, but certainly the extra octave of extension makes them closer to covering the speech range with an available commodity.

I'd use horn loaded full range drivers rather than standard compression drivers for the low mid range, far more displacement for a given cost, but more design work.

EvenLouder should consider horn design from the bottom up, possibly using Ath.

Ath is a piece of software for designing waveguides and horns. At the same time it makes it possible to easily simulate their acoustic behaviour by means of FEA (BEM) via ABEC/AKABAK tool.

Art
 
Thanks for the corrections!

Is The TOA TU-651a good choice, even on the Goldwood GM-450PB smaller horns? the driver is affordable, and goes down to 150hz, and I wonder if they even bothered to test the Goldwood GM-450PB that low.

Regarding ATH it is a very exciting option, and I would love to explore it. I will admit the learning curve looks steep, and I have some fear.

If anyone is interested in working together on this, I have the CAD skills to thicken geometry and do the mechanical parts, but I think I am years from knowing enough about acoustics to operate that software. My DMs are open!
 
Is The TOA TU-651a good choice, even on the Goldwood GM-450PB smaller horns? the driver is affordable, and goes down to 150hz, and I wonder if they even bothered to test the Goldwood GM-450PB that low.
Yes, TU-651 is a good choice, but it can't go down to 150 Hz - absolute minimum is 250 Hz with a steep high-pass filter (3rd order) on a TOA TH660 horn. On a GM-450PB horn it will not go down that low, because of much smaller horn mouth and length.
 
Yes, TU-651 is a good choice, but it can't go down to 150 Hz - absolute minimum is 250 Hz with a steep high-pass filter (3rd order) on a TOA TH660 horn.
Interesting! Can I ask how that was determined?

The data sheet for the driver says
“Frequency Response: 150 Hz – 6 kHz”
And the frequency response graph seems to be 90dB at 200hz with the toa th-650 horn

Graph source:
 
The data sheet for the driver says
“Frequency Response: 150 Hz – 6 kHz”
And the frequency response graph seems to be 90dB at 200hz with the toa th-650 horn

graph.JPG


At 250 hz, the horn and driver combo will do about 102 dB, which is already down more than 10dB from it's peak output level. This will sound about half as loud, more or less. Around 175 hz, it's going to be about half as loud as it would be at 250 hz, about 92dB or so according to the chart.

Excursion goes up 4x every time the frequency is halved, assuming all else is held equal. At 1000hz, if you needed 0.08 mm excursion for a given sound pressure level, at 500 hz, it would be 0.32mm. At 250 hz, this would become 1.28mm. At some point the diaphragm is going to whack the phase plug or break.
 
  • Like
Reactions: Sonce
I'd personally look for the best deal you can find on some neo magnet higher sensitivity fullrange drivers around 4". Buy as many as you can to array them vertically and hammer the FR into shape with dsp. I'd build the boxes out of 3/8" (front baffle) and 1/4" (back and sides).

The lower mids are critical to get right, as most of your speech intelligibility is based from 300 hz - 5k. Power handling will be great, so will overall sensitivity with a decent number of drivers.
 
  • Like
Reactions: Horneydude
@Horneydude If the coverage needs to be vertically expanded, a curved array could be appropriate. The critical array height would need to be fairly long for a sensitivity boost all the way down to 300 hz, but it would alleviate the need for a separate LF driver, given the collective cone surface area is sufficient. A line with 16 to 20 drivers would be very capable of doing the job with decent low end cutoff. Thats about the surface area of a single 15" cone.
 
  • Like
Reactions: Horneydude
Thanks for the corrections!

Is The TOA TU-651a good choice, even on the Goldwood GM-450PB smaller horns? the driver is affordable, and goes down to 150hz, and I wonder if they even bothered to test the Goldwood GM-450PB that low.
You need to look at the cut off frequency of the horn (Fc) as well as the driver response.
Below Fc, the horn response is useless for high power PA use, the driver's excursion goes up as output goes down, basically no more output than the driver with no horn.
The Goldwood GM-450PB (knock off of the JBL 2380 horn) says "cutoff (Fc) of 500Hz.
Fc Cutoff frequency.png

The 2380A horn, not the 4" dome diaphragm 2445 driver sets the low frequency limit, but with only .5mm 1 way excursion, a 2445 can't do much output down low, regardless of the horn's Fc.
Regarding ATH it is a very exciting option, and I would love to explore it. I will admit the learning curve looks steep, and I have some fear.
Hornresp is probably better to start with, though the learning curve is still tough.
The easiest, but most limited in terms of driver selection is Celestion's Horn wizard:
https://celestion-horn.netlify.app/
In a matter of minutes you will be able to see how different horn expansions, profiles, and sizes result in different Fc, beamwidth, Di (directivity index) frequency response, and impedance curves.
Screen Shot 2025-05-20 at 12.37.50 PM.png

That said, as I previously mentioned, horn loaded full range drivers rather than standard compression drivers for the low mid range are far more cost effective.
A $25 dollar "full range driver" has far more displacement capability than the $1000 dollar Celestion Axi2050, which only has a bit more displacement than the 4" diaphragm JBL 2445, 2450, etc.

Art
 
The GW horn mentioned doesn't do that well below 800 hz. I've used it with a few other large drivers ie. B&C DCX50, PRV D3220PH. It does OK with xover above 700 hz but doesn't load the driver lower than 500 hz, as Art already mentioned.