Thanks for this. Looks like your build a tuned a little lower than the sim (note where the lowest dip occurs. The shape of the peaks is also very different, and the impedance curve deviates significantly above about 200 Hz (alignment would be better if you use the semi-inductance parameters for the driver). The measured impedance peak at 60 Hz is actually HIGHER than the sim suggests it should be, which is very unusual. There do appear to be some noticeable box losses as well, maybe a leak around the driver? Small leaks could cause the lowest peak to be shifted upwards in frequency (happened with a TH that a cousin of mine built - there was a leak around the driver and a corresponding drop in output around the resonance frequency).
So, things I'd suggest to do:
1. Remeasure the driver's t/s parameters. DATS measurement while hold the driver in free air, away from any boundaries. Determine Vas by the delta compliance method.
2. Export the measured impedance curve for the driver as a zma file and use the spreadsheet from my site to calculate the semi-inductance parameters, and use those in the Hornresp sim.
3. Use the driver's published parameters in the sim to see how the results compare to what you measured.
4. Confirm that the driver is properly sealed to the baffle when mounted in the box. Typical error in this situation is a mounting hole that's cut a bit too large for the driver because it's spec'd for the driver being mounted through it rather than on top of it, resulting in leaks around the screws that are holding the driver to the box.
I did not calibrate the level between the old measurement and the simulation. It is just a simple comparison of the shape of the response curves. I have no idea what drive voltage I used when I measured the ROAR12.
That actually doesn't matter that much. The Hornresp sim is a lossless one, so the measured response should always be less than the sim'd response, unless what you've built has little or no relation to what you sim'd.
Going by the graph, it looks like the measured response needs to be adjusted at least 2~3dB downwards for a more realistic comparison against the sim.
Yes, suspended in free air. I am confident that the current TS specs are correct
Thanks for the detailed reply, Brian. I will have to find another computer to play with your spreadsheet. My work laptop has blocked all the spreadsheets on your site (something I didn't know was possible), due to them containing macros. Damn IT departments!
I will put some rubber grommets under the srcrews and see if it makes any difference, however the fact that I have gotten a very similar frequency response with two different drivers, that were each mounted differently (see below) makes me almost certain I've stuffed something up in the box construction. This is likely a leaking inner panel - when we put the top on the box, the cheap ply used for the rest of construction had warped a little and required a clamp to pull it back square. I am thinking that when I did that, it might have smeared enough glue away from the join that I have a leak.
My problem is going to be that cutting the box open will require removal of all the screws, which have been filled and sanded. I don't think it will be an easy task! I'll probably ruin at least one saw blade finding a screw that I didn't know was there, which will cost as much as the ply that went into building the box...
So this indoors measurement shows 50hz 8db down on 100hz. Taken outdoors, would you expect the difference to be even greater?
I'm wondering whether 100db 1w/1m at 50hz really is achievable out of this design.
My plan was to build two using better quality ply, with this being a proof of concept, but I am a little hesitant to do so at present. Drivers I have on hand are the aforementioned Beymas and JBL 2217
While you are not able to use the boxplan, you could try an alternative design tool using FreeCAD as you can see in the link below:
https://freeloudspeakerplan.rf.gd/
Yes it is. But it is not 100dB/2,83Volt/m... The driver is very firmly held between two series coupled resonators which does increase efficiency greatly.
The simulated efficiency peak at 56 Hz seems to coincide with the dip in your measured response curve. This does add some credence to my box-leak-theory.
Here is the simulated Diaphragm pressure at 28,3 Volt drive voltage. Black curve is the tapped throat (start of the tapped pipe section) and grey curve is the tapped mouth where it exits into the large front resonator. The series coupled QW design works with very large internal pressures and any leak will hurt the performance greatly.
If you want to check for leaks, get a 30$ 2 MP inspection camera from Amazon or eBay. One of these "Endoscope Inspection Snake Camera" things.
Cheaper than cutting it open and once you have such a tool, you may use it for other tasks around the house and in the car as well. If not, return after a test 😎
Cheaper than cutting it open and once you have such a tool, you may use it for other tasks around the house and in the car as well. If not, return after a test 😎
No. The focus of my camera is far too close (it's designed for rifle barrels), so while I could see inside the box, it was with such fine detail that I couldn't make out where I was.
I will probably cut the box open today and look for a badly sealed internal panel.
Another way to check for leaks is just run pure sine waves through the box at low frequencies. You may hear the characteristic puffing noises caused by tiny leaks.
Thanks Brian, I've spent hours (literally) doing that, and can't seem to find any. I've really tested my wife's patience in the process, too!
I've tried 4 different drivers of two different makes. I've stood on panels while measuring to ensure bracing isn't the problem, every join I can access has had a liberal dose of silicon sealant, and speaker screws have rubber grommets in them. The only thing left I can think of is a leaking internal panel that I can't access.
One last run around with silicon today before I commit to major surgery.
I used an expanding polyurethane glue, which has previously hidden a lot of sins with less than perfect cabinetry skills, but I can't get the PL premium stuff that comes in a caulking gun tube where I live - the stuff I use has more of a honey consistency. Perhaps it's the source of my woes.
When I say low, I mean really low, lower than the box's lowest resonance frequency. Like around 10 Hz, if not lower. Don't drive the box at a level that overloads the driver. You won't hear the pure tone, but you should hear if there's any distortion caused by leaks and the like within the box.
That's right where I was running it.
After using nearly a whole tube of silicone making certain that nothing was leaking around the speaker or external jokns, I did some major surgery on the box, cutting some holes to examine the glue joins of internal panels.
The internal joins all looked perfect.
I'm going to cut my losses here. I really wanted this box to work and selected it from the hundreds of designs out there for the benefits of the design.
I won't rule out that it was entirely a construction error that stopped this box from performing as it should, but the only roar it's going to make now is on the bonfire.
Thanks for this conclusion and your efforts, it will be of great value for me!
Intersting. I imediately found it so superior to ‘paraflex’ on the bottom end and it may likely be design details that ultimately affect roll off and thus room gain. Timing and overlap of both outputs really rears its ugly head or shines in many designs it seems. we ask for so much and then the devil wants Those Details In order to get it?