PDA

View Full Version : DIY project: 2020H + 2447H living room speakers



catsnakebirdman
08-27-2018, 11:36 AM
Hello everybody!

This is my first post here, but I`ve been browsing the forum since around 2006. Back then I had discovered an auction for JBL 4311s on ebay and, in an attempt to find out more about them, stumbled upon the forum. Soon after that I got myself a pair of L150`s and a Yamaha PC2002 amp, both of which I still have and enjoy.

About a year ago, inspired by the DIY-section of this forum, I gathered the parts to build some speakers, too. I have always been impressed by the sound of PA speakers, i.e. the “punch/slam”, “effortless dynamics” and “impressive midrange” that some of these have and wanted to recreate this in a home system. The system is supposed to be used for music playback in a medium-sized (20 sqm) living room – mostly Jazz, Blues, Rock and Soul music. During the last year, the project has already progressed quite a bit.

The aim of this thread is to document the build as it progresses further, and to get some advice and assistance from the many experts and experienced builders on this forum. Currently, I`m stuck in a number of places and I am no longer sure whether the setup I have come up with can be made work satisfactorily.


It all started when one evening, after a few beers and while browsing ebay, I saw an ad for a pair of PA-speakers with components that seemed promising – 2447H large-format compression drivers and 2020H mid/midbass speakers, and, in a beer-fueled rush of euphoria :D, acquired them using “buy now”. In a similar way, a pair of miniDSP PWR-ICE 250 amps soon followed. These amps offer active crossover functionality with IIR and FIR (2300 taps for lf / 300 taps for hf). The 2447Hs got new 2452-SL diaphragms installed and were paired with STX825 waveguides.
82091
In order to save some space and have some more flexibility in placement, the system was supposed to consist of a pair of satellite speakers and a subwoofer. For the sub, an Audax PR380M2 15 inch woofer is available and might be used. Playing around with the TS-parameters of the 2020H in winISD revealed that the sub-sat concept (crossing over to the sub somewhere between 60 and 100 Hz) would not be viable without bass-reflex loading the 2020H, and that accordingly the satellite enclosures had to become a little bit bigger than estimated initially. I settled on an enclosure volume of 45 liters tuned to 65 Hz, and started planning and constructing the cabinets.
82092
82093
82094
Internal dimensions are WxDxH = 28.5cm x 24cm x 95cm, and all panels are 1.9cm MDF, except for the front plate, which is 3.6 cm MDF. To absorb the sound that is radiated into the enclosure, the internal surfaces are lined with felt of 1 cm thickness, and additionally a double layer of polyester damping sheets is used on the side and rear walls in the enclosure segment containing the 2020H. To fight standing waves, the bottom of the enclosure is filled with polyester damping sheets.

To get something up and running quickly, I decided to skip the bass reflex tuning and start with the crossover first, simply inserting the ports the way they were bought (10 cm diameter, 20 cm length). From conducting measurements of the 2020H and the 2447H at different angles, I concluded 1100 Hz to be a good crossover point. Indoor measurements with an IR windowing width of 3 ms served as a basis to EQ the HF and LF flat in the range between about 300 Hz and 20 kHz, and for crossover, 24 dB slopes were used. Delay between HF and LF was adjusted to get the deepest notch possible at the crossover point when reversing HF polarity.

Unfortunately, listening tests turned out to be not satisfactory. There seemed to be a multitude of resonances masking the direct sound. The midrange was not at all as clear as I had hoped for. This became particularly obvious in a direct comparison to the L150s, which sounded a lot cleaner and more detailed in the midrange.

As I found the side and back panels to vibrate considerably at certain frequencies, I decided to try to remedy this possible cause first. In a first attempt, the bracing in one speaker was extended as per the following drawing:
82095
A comparison to the unmodified speaker revealed that this had made things worse - panel vibration seemed to have become even stronger, now shifted to higher frequencies. Thus, in a second attempt, constrained layer damping was tried. Aluminum sheets (thickness 2 mm) were bonded to the sides and back of each speaker with a layer of Sikaflex 292i (thickness 1mm). Additionally, the woofer gasket that had allowed the frame to be in direct contact with the enclosure was replaced with a thick and rather rigid strip of foam to prevent such contact. After these measures, the enclosure walls now appear to be rather dead, with little tangible panel vibration left.

The sound, however, hasn`t improved much. That`s why I`ve decided to start all over, this time doing the internals of the enclosure and the port tuning right before moving on to the crossover. That`s where I`m stuck now and need your help.

The questions that I have relate to the enclosure size and tuning, to the construction of the enclosure and to the lack of midrange clarity that might be caused by resonances/standing waves and/or internal reflections.

catsnakebirdman
08-27-2018, 11:40 AM
1) ENCLOSURE TUNING:
Here are the winISD-simulations that the enclosure volume was based on:

RESPONSE
82096
The yellow curve shows the unassisted response in a 45 liter enclosure tuned to 65 Hz, and the blue one is the assisted response.

EQ
82097
In this diagram, the filter leading from yellow to blue is shown.

CONE EXCURSION
82098
With an input power of 50 watts, an output of about 116 dB is reached while cone excursion stays below 2.5 mm. At an input power of 160 watts, cone excursion hits 5 mm - the xmax of the driver, and the output level reaches about 121 dB. However, as the xmax given by JBL goes along with 10% THD, and the theoretical xmax (VC winding depth – top plate height)/2 calculates to 1.25 mm only, such levels might not be realistic. In any case, 116 dB should be loud enough for the living room.

GROUP DELAY
82099
With winISD giving the 2pi response only, the influence of nearby walls or room gain was not considered when making these simulations and deciding on enclosure volume and tuning. The parameters I finally decided on give a system that is a bit underdamped, so group delay might be too high and the whole thing might give a smeared sound / ring in the region of the tuning peak. Finally, the 2020H might just generally not sound good when used in a relatively large volume (45 liters) tuned relatively low (65 Hz), i.e. in that possibly much of the midrange clarity that this driver is capable of is lost.

What do you think? Can the 2020H be made to work satisfactorily in this enclosure? Is a tuning of 65 Hz too low? What would be a realistic tuning frequency for the satellite speakers when factoring in room/boundary effects (the enclosures will be most likely placed near a corner)? If necessary, the internal volume can be reduced to about 35 liters without much effort.


2) ENCLOSURE CONSTRUCTION / -DAMPING
As I had never built a speaker enclosure before, a lot might be wrong with the ones I have built now.
In particular, I`m unsure about the internal geometry (free space), bracing, component placement and damping materials – almost everything, basically. ;)

Issues that possibly need to be addressed might be the following:
- standing waves inside the enclosure
- reflected sound radiating through the woofer cone
- reflected sound radiating through the ports
- port resonances.

The enclosure was designed without taking width/depth/height-ratios into consideration much, other than avoiding equal lengths, so standing waves might be a problem. The port area appears to be a bit too large, possibly allowing for too much midrange to leak through.
The bracing might well be overkill – it leaves rather little free area connecting the segments between braces, which might weaken the low frequency output or even tune the segments themselves. Interestingly, when I tried different port lengths one time (between 0 cm and 20 cm), the effect of port length on the tuning frequency appeared to be rather small, with the dip in the woofer nearfield response only changing in the range between about 40 and 60 Hz.

Do you see any problems with the enclosure? What kind of damping material do you recommend? If necessary, the extra bracing in one of the speakers can be removed again in order to get more room for damping material.

Thanks a lot for reading this far! Any comments or suggestions are highly welcome!

Best regards, Eric :)

ivica
08-27-2018, 01:02 PM
Hi Eric,

I would first try to see how the speakers would soud as close box design, by closing vents. Ther would be no bass but midbass have to be present.
May be using such design, with subwoofers you can reach expectable sound....
I think you can reach better results using Jbl 2206H, or 2202H drivers, as You
Want to use up to about 1200Hz....

Regards
Ivica

speakerdave
08-27-2018, 01:24 PM
A few things that come to mind looking at your intrepid undertaking:

I'm not clear as to whether you have applied eq to the speaker. The 2020 has a 10 dB response rise over the range you are using it (in JBL's 10 cu ft baseline test setup). Don't know whether your setup corrects all of that.

I see a test mike in one of your photos, but you don't have an acoustic response graph?

Another unknown is the condition of the 2020 cones. Have they been used hard?

The interior box height and width are a close multiple.

You have mentioned resonances. Can you identify the pitch? What type of crossover are you using? Does your treble range encompass a resonance mode in the horn? I think JBL used that horn at 1.3kHz.

I'm not a big fan of particle board/MDF for speaker boxes; dense, but structurally it sucks. Maybe 1" thick or greater is ok. The Manley Labs 1.5 cu ft enclosure for the Tannoy 2558 DC uses 1.5 inch particle board. That really works well, but would be massively heavy for larger boxes.

Your previous speaker--does it not have a prominent bass? Maybe you are missing that. Can you set up your L150's as a bass module for your diy's to provide a full spectrum for another listening test?

I'd second the suggestion that the 2206H--new ones--would be a better choice for LF in a satellite, as in your application.

Ian Mackenzie
08-28-2018, 05:26 AM
You have put a lot of thought into the project and without seeing FR measurements your description of the crossover appears correct. I would confirm the lowest recommended crossover point for this particular waveguide. Do a search on the forums for this particular waveguide as it’s been discussed in detail.

There is possibly more than one reason for the resonance issues you describe.
It’s often difficult to predict the way an enclosure will react when loaded by a driver.

Looking at examples below of both 18 Sound and RCF (Bill Woods) 12 inch two ways systems your enclosure proportions may need to be modified. You may wish to try 18 mm birch plywood or 25 mm mdf and use 25-50mm fibre glass insulation on the sides, top & bottom and two layers on the rear as a starting point. Fibreglass has predictable absorption properties in loudspeaker and is highly effective.

Good luck with your project

http://www.rcfaudio.ro/suggested-designs_v2/Suggested-Designs-The-club-12.pdf

http://www.eighteensound.com/Portals/0/EnclosuresKits/18_Sound_12_2_Ways_v1.pdf

Tkroeker
03-21-2021, 10:06 PM
I know this is an old thread. First let me say great work on the cabinet. With that being said the first problem is the 2020 is a mid. I use them on top of 15s as woofers and 18s as subs. This is the reason you like the sound of a PA.

I cross them at around 300 and take them up to 800 and thats where I bring in the 2445's.

This is in no way the rule for them but is a great reference point for judging what changes will make what effect.

The previous comment about dropping them in a sealed box for testing is spot on. Try one that is about 30 liters or 1.5 ft3 and use birch Veneer. Try to give it 300 to 1k only and see if it takes your head off. Then after that excitement, work an eq untill the clarity is nice. If you want have a generic 15 in a different cabinet to warm it up.

Again very nice woodworking. I especially like how you braced the horn driver.