Low THD Tang Band RBM drivers?

Status
Not open for further replies.
I've been testing my speaker with a first order crossover about 360Hz. The treble disperses really well around the room and frequency sweeps show a pretty smooth crossover. However, bass dispersion suffers from the expected nodes and antinodes so there are bass sweet spots where it sounds great but substantial areas where the bass is unacceptably weak. This seems worse than a regular stereo pair. The 4R mid-tweeter is roughly 3dB more efficient so at the moment its padded with a 3.9R 5W resistor. It has 33uF as a high pass with a -6dB point at 350Hz. The 4R woofer has a 3mH inductor which is -6dB at 370Hz. The other issue is a noticeable peak in output in upper vocal notes. So, a work in progress!
An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
Well, look what turned up two months after ordering it. Spot the vent holes in the rear cone. I'm breaking it in now.
An externally hosted image should be here but it was not working when we last tested it.

In the meantime, I made a toy from a spare Bluetooth. After the last BT didn't work, I bought three different ones to test, and one of them had an amplifier on the board. Plus I had some leftover bits so I glued the mess together and ... it won't win any prizes but it works and sounds pretty okay.
An externally hosted image should be here but it was not working when we last tested it.

It's a 3" woofer with a 1" tweeter and a first order crossover, mounted into a upvc pipe bend with a 3" passive radiator at the other end. The remote control bluetooth is this
An externally hosted image should be here but it was not working when we last tested it.

I wanted mono and this is stereo, so I removed the 20K resistors for the right channel amp (R7 and R9? near the headphone connector) and the 8002 soic amp chip. I shorted C18 and C19 (they are 10uF each) on the input side to the left channel 8002 to create a mono signal. I increased C17 to 5uF to reduce the turn on pop. Finally I bumped the gain to over 12 from just 2 by reducing Ri to 10K and increasing Rf to 67K instead of the remaning pair of 20K resistors (R13 and ?). It's powered by a 3,7V lithium battery that has an input charger and an output boost to 5V. Finally a switch can turn off the amp/bluetooth but still leave the charger connected. I haven't tried the USB audio but the TF card reader works with 32Gb and the Bluettoth works with no fuss. The lady speakers English with a Putonghua accent.

The battery charger is from Sanwu . I tried 3 different voltage boosters but two of them created some kind of motorboating-type noises and only this one worked without problems.

Anyway, it's a "JBL" type of thing to compare to the JBL beater I'm still working on. I don't mind admitting I'm having some difficulty getting balanced well dispersed sound from it. I resolved the upper mid frequency response issue but getting decent smooth bass to balance and disperse well simply isn't easy.
 
Last edited:
Very nice nanoloop. I really enjoy reading your updates. My project is still progressing as well... just have been having some difficulty finalizing the cabinet design.
 

Attachments

  • FinalCrossSection.jpg
    FinalCrossSection.jpg
    52.7 KB · Views: 180
  • FinalSurface.png
    FinalSurface.png
    836.5 KB · Views: 94
Last edited:
That cabinet looks ace. One thing that might throw a curve is weight - the TB driver is several kgs - combined with a hollow but thankfully ribbed plastic mount. Plus of course it's a sub so it creates some serious pressure in the cabinet. Anyway, I saw in the left pic that the wall is rather thinner on the top left side of the woofer enclosure so I thought it worth mentioning that weight and pressure might create issues?
 
No doubt that the driver will be very secure where I mount it. Theres alot of material there and its extremely resilient / stiff. That thin wall in the top left is a result of a cutout I made in the wall to place absorber. I figured it would make sense to create a cutout for 0.5in absorber / damping to help with any possible cabinet resonances.

I'm 3D printing the parts and thus... I needed to slice up the cabinet into 5 different pieces (And then glue together for assembly) I'm securing the parts on the X and Y plane with wooden dowels. The Z plane will be secured with 2-part mix epoxy. But now I'm thinking maybe I should remove the cutouts so that the glue has more area to bond to.

It really just depends on how well that epoxy holds. I'll have to test the bonding strength.
 

Attachments

  • 20180407_113552.jpg
    20180407_113552.jpg
    1 MB · Views: 77
Last edited:
Looks great tbh. If you're using a suitable epoxy then bond strength can exceed the strength of the material but the wrong kind won't penetrate sufficiently, can crack, break, etc. Its been a long time since I studied this so I can't remember the different formulations but I guess Google knows.

Reflex speakers benefit from minimal damping material and stiff panels. Rod Elliot did some testing of this. Volume filling a reflex box
 
I forgot to say - I guess the longest dimension in your woofer enclosure is only slightly greater than the woofer width itself so the lowest resonant frequency should be above 500Hz? Which is double the -6dB point of the crossover? So resonance shouldn't be an issue hopefully.

My JBL killer is nearing the end of its setup/testing and I'm amazed at how good it sounds now. The bass has really good extension even on bass heavy stuff like Massive Attack - a very satisfying sound and way better than expected. I'm seriously surprised - defies my expectations for what a 1.2l/3" driver can do. I'm still working on the notch filter but the pic shows the current components.

I avoided the need for any attenuation resistors by using the TDA8932 full bridge to drive the woofer (through a 3mH low pass inductor), and only half the bridge to drive the mid-tweeter through a 100uF DC blocking high pass. This means the woofer gets 6dB more than the mid tweeter. As the woofer is 3dB less efficient, it means it's actually getting a 3dB boost and this has compensated nicely for the poor dispersion. Although this makes for an unconventional crossover, it works well with these components and sounds great. The two drivers share one side of the bridge as their positive but the ground return for the half bridge goes to the power ground. It also means I don 't need to bi-amp. The battery+amp combo provide enough power to push the woofer beyond its comfort zone.

I also noticed that if the speaker is on a coffee table and I walk up to it, it sounds fine until I'm about 30 degrees from leaning over the very top. And above the top cone, the sound changes noticeably - the sound immediately above the speaker top cone is quite different from the dispersed sound - very shrill. I experimented with ways to control this and found rings did the job - so then I tried various rings in various places and found two places where they make a noticeable improvement. So I have used fish tank air pipe as rings, painted black as usual ;-) The problem is much less and the dispersed sound is more balanced too.

So, just need to perfect the notch and I can finish the build.

p1000470.jpg

jb-killer.jpg
 
Last edited:
Glad to hear the JBL Killer is doing well. Has a really interesting look to it. What frequency can it get down to?

Do you know any good resources regarding cabinet resonances? I'm trying to get a better understanding of resonances so I can better predict if they will occur. I can't seem to find a good article or book on the subject. Thanks.
 
Glad to hear the JBL Killer is doing well. Has a really interesting look to it. What frequency can it get down to?

Do you know any good resources regarding cabinet resonances? I'm trying to get a better understanding of resonances so I can better predict if they will occur. I can't seem to find a good article or book on the subject. Thanks.

At full volume, listening at a room node, using the most dynamic EQ of the 7 settings on the Bluetooth, it produces audible notes down to 24Hz. However, this is v quiet so would not be audible in music, and in fact, the poor little thing is having a hard time not simply bouncing about. For music, the bass response is reasonably flat to 91Hz and after that there is a noticeable roll off in response before a tail end boost between 77Hz and 62Hz. So in realistic listening terms, it sounds like it goes down to 60Hz as a -6dB point, but it's very uneven below 90Hz. It seems to match an extended bass shelf design where the enclosure is about 20% larger than ideal damping and the reflex is tuned very low to give a tail end hump in the response. I can't change this cos it would mean taking weight off the passive radiator, which is likely to damage it in this case. But that's fine - it's not hifi by any means but it's a pleasant signature. And that design has come about just as a result of some some lucky guesses and purchases. :)

And it keeps playing for hours - it was in use most of yesterday and the battery voltage has dropped from 12.5V to 11.1V. The datasheet says the amp shuts down at 10V, which is above the deep discharge voltage of each of the 3 batteries (2.7 to 3V). The amp is super efficient at over 90% - it seems ideal for this application.

The full-bridge + half bridge works really well - it should work for your design too so it would mean you don't need to attenuate the full range units with resistors as much. The 80dB woofer gets a 6dB boost and that just about matches the full range drivers efficiency right? If you try it, just remember the half-bridge must have a capacitor in the path to the full range speaker or DC will flow and bye bye voice coil. But the capacitor will be in the high pass crossover anyway so it doesn't need an extra one.

I'm afraid I don't know of any books on resonance - I have learned about it from multiple scraps over the years. But if you Google quarter wave resonance, helmholtz resonance bass traps, enclosure bracing, etc and go down those rabbit holes that should broadly cover it? Pro design can use finite element analysis but of course this is beyond most diyers so just general mech eng principles apply of avoiding stress concentrations with rolled off not sharp edges, extra thickness in high stress areas, etc. If you need extra stiffness, you can probably epoxy on steel strips and other such measures.

One last note to record for my project - the parallel notch filter is done at 3.9R/24uF/0.15mH. So I can get to work on the W5 2053 tube sub :)
 
Last edited:
And here is the rough build of the W5 2053 ready to be tuned. I keep thinking it needs breaking in, but of course, no spider, so perhaps not. But this is as far as I'm going to get for now. This afternoon is the start of a blitz at work so no hobbies for a month or so. :-( I plan to fill and smooth the roughness, then cover with titanium coloured vinyl and finally a stainless steel outer mesh to give it an industrial look. But wait and see... work calls.
An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.
 
I currently have three Peerless Tymphany 3.5inch passive radiators operating in a 2.2L enclosure with the W4-2089 (Same as W5-2053). After doing some last minute research I'm having second thoughts....

I'm worried that 3x 3.5inch Passive Radiators will not do the job based on what others are saying online. In my simulations it appears that 25 grams added to each Peerless 3.5in Passive Radiator will do the trick (They already have 38 grams attached). Yet some people online are suggesting that the W4-2089 would be better suited with 3x or 4x 5.25inch Tymphany Peerless passive radiators.

So I'm considering adding 3x 2inch Dayton Audio passive radiators in addition to the 3 I already have. So 6 passive radiators total... 3x 3.5inch and 3x 2inch. Thoughts? I don't want to starve the W4-2089
 
Last edited:
Have you measured the actual surface areas? How about the x max of the passives? If they add up to/equal at least the same as the main driver in their ability to shift a volume or air, then it should be okay, even when the driver is at x max. However, up to 50% more for the passives is the usual advice because you don't want them introducing non linear issues near x max, esp given this driver is designed to be low distortion at such limits.
 
So here are my calculations. Statistics first...

Passive Radiator (3x of These):
Surface Area: 50.2cm^2
Max Excursion: 6mm

W4-2089 (1x of These):
Surface Area: 75cm^2
Max Excursion: 8mm

Calculations:
Total surface area of the 3x passive radiators is 150.6cm^2 which is double the W4-2089 which is 75cm^2... so that's pretty much perfect since I've read your passive radiators' combined surface area should be anywhere from equal to or double the woofer's surface area. Regarding volume, the passive radiators can displace 90.36cm^3 of air. The woofer displaces 60cm^3... so that should also be good.

Going to add around 25grams to each passive radiator (Not including the 38grams that are already installed). This is a simulation estimate... to tune the passive radiators, I will basically just add weight until at 40Hz the driver isn't moving but the passive radiators are. I believe this is how you tune passive radiators.
 
Volume and size look fine. in my experience, the more weight I add, the more the output from the PR reduces with increasing frequency, leading to a PR that only adds the lowest notes. It's probably affecting the transient response which might muddy the bass, which is why I prefer a larger cone and less added weight. So I start with too little weight and do frequency sweeps to see what's happening with the output as weight is added. I've also tried tuning it by measuring impedance before and I recommend this as the best way. I don't think your proposed way will produce very accurate results but no harm in trying and it may work just fine in the end. If it sounds right.... Anyway, here's an easy way to test impedance: Loudspeaker impedance measurement using a multimeter and 2 resistors
 
Last edited:
When tuning the passive radiator, how does the impedance plot factor into the tuning? So I have the impedance plot from the manufacturer... or should I re-measure once it's installed with the passive radiators?


The article was useful for showing me how to measure impedance, Just trying to understand how to then use that plot to tune the PR's. Thanks!
 
the TB plot shows one impedance peak at 40hz iirc because it's in free air - when you have the driver reflex loaded "correctly", the single peak will split into two equal and smaller peaks. The valley between them is the new resonant frequency, and shows the driver is being damped at its resonant frequency for the enclosure, so the driver excursion is reduced and power handling increased for that enclosure. So yes, measuring impedance when the driver is in the enclosure is usually necessary because it will be different from the TB plot.

In generalised terms, if you put the driver in an undersized enclosure, the resonant frequency will be higher and the mid bass output will be stronger. In an oversized enclosure, the mid bass will roll off gently instead of rising and the resonant frequency will be lower, allowing more deep bass but at reduced output (an extended bass shelf design). With an ideally sized enclosure, the driver will be correctly damped and produce a theoretically flat response in the mid bass until it gets near the resonant frequency roll off. The enclosure size + driver sets this up and it can be tweaked by adjusting the reflex loading, but is largely determined by the driver + enclosure. This is why I got so lucky with my 3" woofer. I guessed around 1.2 litres but it seems to be around 1L is ideal so I got an extended bass shelf, and then the PR is tuned "too low" producing a tail end boost around 65Hz, giving the impression of a deepish bass capability, albeit very uneven and a tendency towards having one deep bass note.
 
Last edited:
It's been awhile since my last post. Still working on this project. Things have been a bit finicky with the 3D printing so I haven't posted much. Producing the cabinet is proving to be the hardest part. I did put together a prototype with 3D pieces I decided I didn't want to use for the final build. The Full Range crossover sounded pretty fantastic. I might do some frequency response testing with the prototype. Biggest concern is the baffle for each driver will be much shorter given that there is only half a cabinet.
 
Last edited:
Status
Not open for further replies.