Site Links
Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
Categories
An experimental transmission line . . .
. . . with dual line chambers and ports. The woofer, a TB W6-1139SIF, will be buried inside the line. Gross internal volume, including ports, will be about 1.4ft3. This will leave about 0.1ft3 for a sealed midrange chamber, if needed. (Or I may go with an open back midrange instead).
I modelled the line in the Leonard Audio TL program. For validation, I'm building an initial test line with a removable top panel and slot ports, held together temporaily with bar clamps and weather stripping. The slot ports will start out with oversized dimensions of 1.25x7x14." Then I'll slowly trim them down to size based on measurements. At the same time, line stuffing will be adjusted until mid-bass peaking matches the model.
I just found another cheap stack of particle board shelves at Mendards, so, it's time to make sawdust. I'll report back periodically with measurements as I try to pound the response curve into shape. I'll put a side view drawing and a few model pics in the next post.
Comments
Here is my drawing of the TL:
Here is the modelled SPL at 2.83V:
Here is the modelled SPL with 60W input. It will make about 98dB before reaching xmax:
Here is the modelled front and rear port velocity at 60 watts:
Here is the cone displacement at 60 watts:
Here is what the basic LA-TL model looks like. I don't know what kind of line this is. It is a mass-loaded TL on the front and rear, maybe call it a band pass mass-loaded TL (BP-ML-TL).
Looks like a 6th order bandpass with a large and very long port.
Here is how this alignment looks in WinISD, modelled as a dual chamber 6th order bandpass, using the same front and rear chamber volumes and port tuning frequencies. Much sharper roll offs, but WinISD does not account for line length or port resonances:
Here is what it looks like when you check WinISD's "Use transmission line model for port simulation" check box. This accounts for the port resonances, but not chamber length:
Here is how this alignment looks in Hornresp, modelled as a parallel 6th order band pass. Lots of dips and peaks above 100Hz, but I could not find a place to enter damping material for a parallel 6th order band pass:
Here is the Hornresp input parameter screen that I used to generate the above graph:
Clicking around in LA-TL, I was able to come up with a better model by moving the woofer 4" closer to the front of the line and then re-arranging the stuffing toward the end of the line. My first LA-TL model (above) shows the first peak @ 380Hz, only 14dB down; the 2nd peak at 550Hz, only 17dB down; and the 3rd peak at 800Hz, only 11db down. The revised model (below) has a flatter pass band with first peak @ 500Hz, 20dB down; and 2nd peak @ 1.7kHz, 25dB down. Much improved. I've seen this type of thing mentioned in several TL threads; the importance of proper woofer placement along the TL for the best results.
re: offset driver
Yeah, for the basic TLs I model in Hornresp, the exact offset of the driver from the closed end is critical to eliminating much of the 2nd and 3rd harmonics dips and peaks.
I agree. I remember you pointing this out in the other thread, which prompted me to try this. So, because this is so critical, I've decided to glue 4 small mounting tab blocks to the four corners of the baffle board. This way, I can adjust the woofer position in the line during the testing process. Thanks for pointing this out.
EDIT: Here is a link to the other thread going over this topic: https://diy.midwestaudio.club/discussion/2236/ported-vs-mltl-box-for-ciare-hw251n-woofer
I have now mostly completed my move in to a new home and will resume a TL experiment with stuffing - both materials and density. I have a TL that I can remove a side for the experiments.
Cool. Keep us posted.
Ready for glue up.
Does mama know that you have "borrowed" the kitchen counter top for this project?
Menard's "value shelving." $1.33/shelf after 11% rebate. This is some really good stuff, very dense particle board with very durable formica type lamination on one side and glossy epoxy type surface on the other side. Much better than the flaky particle board "handy panel" crap that they sell.
Looks like the stuff I've used off and on for years for baffles. At least the pattern isn't gross.
InDIYana Event Website
Gluing it up: I put temporary spacer boards between the transmission line panels, to hold all panels in the correct position as the glue sets. I suppose I could have cut dado type channels for alignment, but that would have been too much work. Simple butt joints are so much faster. I just have to remember to tap the temporary spacers with a hammer every 10 minutes or so, to keep them from sticking to the glue.
Clamps & spacer boards removed:
The internal baffle board is now complete.
Front view of baffle board. I added four 90 degree mounting brackets so that I can adjust the woofer's position along the transmission line during the testing process.
Rear view of baffle board:
Here is a pic of the baffle & driver, positioned in the 8 inch offset location in the line. I haven't drilled the 4 mounting holes yet, so the driver is held in place by a piece of particle board for the photo. The plan is to drill sets of 4 mounting holes every 1/2" from 4 inches up to 9 inches. I will then be able to re-position and measure the woofer in 1/2" increments, in an effort to find the best spot.
Here is what the woofer looks like in the 4.5" offset position:
very interesting project !
Excellent thought process and execution Bill!
Thanks, John & tajanes. The plan is starting to come together!
Adjustable-removable slot port glue-up:
Note that the long board clamped along the center of the port is there only for the purpose of flattening the long port board underneath it. All of the "value shelves" that I picked up were slightly curved or warped along their length.
Both slot ports are now done:
Closeup detail of slot ports (front & rear views):
Here is how I plan to clamp the main cabinet, top panel, and slot ports together to make NF port measurements. This is just a dry test fit to make sure all the panel edges line up properly . I still need to pick up a couple rolls of window gasket material to seal all the edges. I also need to pick up a few bags of poly fill.
Gasket material is in place and I'm getting ready to make some initial test measurements. All of my initial measurements will be made without stuffing, changing the the woofer offset position and noting the change in bandpass tuning frequencies and the change in line harmonics, especially the 2nd and 3rd. But before starting, I decided to make it easier to adjust the slot port tuning. My slot ports measure, internally, 1.25"x7"x17." My initial plan was to trim the 1.25" dimension down incrementally on my table saw between measurements. Instead, I decided to create several 1/8" and 1/4" thick inserts that will be held in place with carpet tape. I will then be able to adjust the 1.25" dimension in 1/8" increments by simply inserting and removing inserts. Kind of like the way you change the thickness of a dado saw blade by inserting & removing the number of chipper inserts.
I mounted the woofer in the 7.5" offset position and then sealed it into this position with screws, gasket, and mortite caulking. Went through about 18 feet of window gasket material elsewhere on the TL and port edges. Clamped it all together and ran up a set of front and rear NF port measurements with the line completely unstuffed and both slot ports at their initial 1.25x7x17" dimensions (no 1/8" or 1/4" shims inserted). I did the measurements twice, once with my dual channel mic setup to create Impulse files for conversion and merging in VituixCAD. And then again using my USB OmniMic and OmniMic software to add and display the FRD's. The results are similar, but can you can see significant differences in the individual and summation graphs below. I believe that the differences are due to the fact than OmniMic phase data is not truely minimum phase and this creates a "time of flight" sync problem with OmniMic NF port measurements. I think the dual channel summation is more accurate.
Before clamping it up:
All clamped up and ready to measure:
Here is the VituixCAD merger of the front and rear ports, driver in the 7.5" offset position, no stuffing, and both slot ports in the wide open 1.25" position:
Here is the same summation, but this time using the OmniMic microphone and software to add the front and rear FRD files together. This was done using "Add curves" to load the front port FRD and then using the "Sum" function to add the rear port FRD to this file.
Before taking the clamps off, I used DATS V2 to run an impedance curve on this unstuffed line. Looks like the rear port is tuned to about 31Hz. The front port looks like it is tuned to about 65Hz. The glitches at 110, 130, & 220 appear to be the beginning of line harmonics. Looks like the unstuffed tuning is roughly where I would like it to be. I am looking to move the rear port tuning down to about 25Hz. And the front port will be lowered to somewhere in the 50 to 55Hz area by the time I get done. This looks very promising so far. I will have to come up with some method to organize my testing process, maybe put the data in a spreadsheet or something.
Have you additionally tried any measurement's say at distance (mic aimed btwn slots) - let the slots combine as one unit? Good to know what each port is doing, but also as a unit.
I tried placing the mic half way between the two ports, close to the baffle. The FR is similar to the summed responses above, except that the low frequencies at 30Hz and below roll off much faster. This is probably because the mic is simply too far away from either of the two ports. You have to keep the mic close to the mouth of each port to get an accurate NF measurement. If I try to move the mic back further but keep it centered between the ports, I just end up measuring the room and not the speaker.
If you look at the VituixCAD Merger screen that I posted above, the red curve is the combined response, and this should be very accurate up to almost 1kHz. The two green curves on this graph are the individual port responses. The green curve with the dip at 31Hz is the front port; the green curve with the dip at 75Hz is the rear port.
If I try to move the mic back further but keep it centered between the ports, I just end up measuring the room and not the speaker.
Yes, at some point it’s all about ‘in room’ response, often measured at ~ 1meter and listening location +/- 12 feet. I had just suggested ‘at distance’ measure as the ‘at port’ summations, while critical in evaluation, seemed to have its own issues. I’m really interested in your experiment here - kudos on design and execution.
Thanks, tajanes. I'll try to include some "at distance'' measurements as well. Initially, my plan will be to perform near field measurements on each port individually, varying the woofer offset incrementally by 1/8" per measurement, and noting how the measured line harmonics change in magnitude and frequency. I am going to do all my initial measurements without stuffing, so the FRD curves will look bad and be very peaky in the 100Hz to 1kHz range. As I understand it, this is the design procedure recommended by experienced TL builders (Bob Brines, Brian Steele, @rjj45 , etc.). You find the best woofer position without stuffing, then add stuffing to suppress the remaining line harmonics.
I will also be attempting to match the stuffing density recommendations of my LA-TL and Hornresp models to the amount of stuffing actually used in my line. Don't know if this will be possible, but I plan to give this a try.
I put together a spreadsheet to keep track of all the variables for each set of near field (NF) port measurements. I've only made two measurements so far and I've discovered that it is extremely important to keep the amplifier volume and Steinberg UR22 settings the same for all measurements. Otherwise I will not be able to compare one set of measurements with another.
The following will be my measurement variables:
1) Woofer position offset in the TL, varying from 3" to 10" down the line in 1/2" increments.
2) Slot port tuning dimensions, varying from a 0.625" wide slot up to a 1.25" wide slot.
3) Stuffing material physical locations, broken out by line section (2 front & 8 rear sections).
4) Stuffing density, measured in ounces, broken out by line section.
5) Stuffing type, broken out line section.
I'll be using my calibrated Sonarworks SoundID mic and Steinberg UR22 MKII in dual channel loopback mode to capture NF front and rear port IR files. These IR files will be converted to frd's in VituixCAD and then summed together into a single frd using OmniMic's "AddedCurves" and "Sum with" functions. I will not be applying a time delay between the front and rear port locations during the summation. I will be using DATS V2 to capture individual zma files of each test. When I get more data collected, I'll post some overlay graph comparisons, together with my interpretation or guess of what I think is going on.
I"m going to something a bit similar, playing with lots of stuffing variables in a folded TL, and figured that I should also run a DATS with every iteration to compare. I'm sure you know, if you over stuff a TL, it turns into more of a sealed box alignment.
Also, at one time I was looking to make a personal air filter (late wife had COPD), and found that I could get an air flow meter fairly cheaply. It might be possible to measure port air speed - dunno.
The air is moving/oscillating both directions rather quickly for flow measurements directly at the port but if you introduce a second same size port and place a piece of paper over both ports such that air can flow in to the box from one port and out the other port you will create an air-flow pulse train and the peaks should be readable so that you can calculate the airflow from the driver's excursion and Sd.
Sound crazy?
All this talk about port oscillation got me thinking that I should change my slot port tuning method. As I was testing, I reduced the 1.25" slot port dimension down to 0.75" and noticed that the tuning went lower (from an FB of about 32Hz down to about 28Hz or so) but I also lost some output at the tuning frequency. Not sure if this is due to side wall frictional losses, but it might be. 7x0.75 is 9.33 to 1! So, I decided to shift gears and abandon the idea of adjusting the ports along the 1.25" dimension. Instead, I sliced up a bunch of 5/8" and 3/4" particle board sticks, each 17" long. I plan to stick them along the edges of the slot ports with carpet tape for testing. It makes much more sense to adjust the tuning this way because the final aspect ratio, when I hit my target Fb of about 25 or 25Hz, will be close to 4" x 1.25" (or 3.2 to 1). Much better in terms of frictional losses.
Great observation and decision ^
