InDIYana build - SuperNova Minimus

[jhollander]

Why not brace the front baffle? A post brace would help. The bracing can be flush with the old baffle and add the same gasket material you plan for the new baffle. BTW nice build!

[skatz]

So what will you use to fix the round-over gouge? Bondo? Durham water putty? wood filler? I ask because I've been there myself.

[wogg]

Why not brace the front baffle? A post brace would help. The bracing can be flush with the old baffle and add the same gasket material you plan for the new baffle. BTW nice build!

Not a bad idea, there's room between the tweeter and woofer for a post. I may just float it in there with tension on some rubber weather stripping and easy to remove glue on one side, like a hot melt.

[wogg]

So what will you use to fix the round-over gouge? Bondo? Durham water putty? wood filler? I ask because I've been there myself.

I'm going to try Bondo. I'm not confident a regular wood filler would get hard enough to smooth back out on the router re-run.

[4thtry]

Great project. And nice baffle retro-fit. Was the original crossover a single, NPE capacitor in series with the tweeter? With the woofer running free on the top? I can't believe RS would do such a thing!

[wogg]

Great project. And nice baffle retro-fit. Was the original crossover a single, NPE capacitor in series with the tweeter? With the woofer running free on the top? I can't believe RS would do such a thing!

LOL, sure was Bill! If you stepped up to a three way you got 2 NPE caps!

This one did sound pretty good in it's day. I believe they ran around $70 a pair, and were a consumer reports top pick. Can't find that online, but I think that's true.

[wogg]

More progress! Spent some time with the old hand drill today. I was a little nervous without a drill press to help with verticle and depth guiding, but decided to just go with what I have.

Rather than hacking with a rasp for an hour, I pulled out the old scroll saw to make some clearance for the port assemblies. I did a whole pinewood derby car with this scroll saw once. This was definitely much faster than a rasp. Can you tell I don't have too much concern with the internal look of these?


A few minutes later I could drop the baffles on with the port and elbow in place. Here's how they align in the box. This is about 6" of effective port length.

[wogg]

Next up all the mounting holes were drilled. I have 10-32 insert nuts to mate with the large head mounting bolts for the baffle. To recess the screw heads a bit I used a 3/4" forstner bit. To keep it from going badly by hand, I used the slow speed on the drill and just took my time.


Using a socket driver on the drill I set the inserts in, and final torqued them with an allen wrench.

[wogg]

For the driver hardware, I didn't like anything I could get at the big box store, so I went to Amazon and got 8-32 pan head hex screws. I went for the not easy, but reliable method of going all the way through the baffle and mating with a washer and lock nut.




The drivers pulled in pretty close to dead flush, and the screws look fantastic.

[wogg]

Now I added a weather stripping line around the edge, and mounted the sucker.


Here's where we have the downside of hand tools. The centers of the mounting bolts and insert nuts for the baffle walked around a little, making them not so center in my 3/4" recess spot. This is only a problem if I can't reel in my OCD. Once this is all finished, I think this will be a really minor issue, and really just give a good hint that these are not OEM.

This is one of the better ones.


This is the worst one.


I soldered a couple bare wires in there and shoved them through the port, and now I'm ready for testing.

[wogg]

I tested without any additional port length past the elbow first. The DATS sweep confirmed a 45Hz tuning, right where I was planning it anyway.


I pulled the actual tweeter ZMA file as well, and used X-Sim to determine what I can use for protection that won't affect the crossover area for measurement. I figured that about 300uF will block very low frequencies with less than 1dB of affect at 1kHz and above.

Straight ZMA


And with the 300uF enabled


I've got the woofer breaking in at the moment, running some 30Hz tones at pretty good excursion. The predicted F3 is 60, but just doing a sweep as it is and listening, this thing will do the 50Hz range just fine, should be a pretty decent speaker!

I'll re-run DATS in a while and see if the tuning has changed. Then it's time to setup measurements for crossover design.

[Jake]

Coming along nice. I love the retro fit.

[wogg]

I managed to make some measurements and get started on a crossover sim. I think this part of the process is generally poorly documented on here, with a lot more attention going to some of the fantastic woodworking and sometimes tweaks with resulting measurements but not really the measurement process itself. Since this is my first design that I haven't used manufacturer data to get a crossover going first, I figure I'll expand this area a bit.

For equipment I'm using an ultra budget setup, an iMM6 mini microphone with calibration data from Dayton that plugs right into the headset jack on my trusty Surface Pro 3. I use a cheap music stand to prop the surface pro up to the correct height and distance for measurement, and ARTA for the capture. This is a wicked cheap setup, but fully capable of single channel measurements with pretty good accuracy.

In ARTA, all I've done is load the Dayton calibration data into the program as frequency response compensation (under the setup menu).


On axis measurements methods for my space...

• Use 1M distance, about 80dB loudness with pink noise before the sweep
• Lots of obstacles, gates will end up about 3.5ms

I used a tape measure to get 39" distance, and eyeballed the tweeter axis to the mic. For the 80dB check, I turned on the pink noise from ARTA and used the iMM6 mic with audio tools to read 80dB average from the woofer to set the sweep volume. Then the mic went back into the Surface for the sweeps.

In lieu of the cap on the tweeter I figured I'd use Equalize APO to high pass at 300. This turned out to be a bad idea, there was more data in the config file that was boosting the highs for my crappy Yamaha computer speakers. After going through the whole process once and attempting to figure out a crossover I finally realized I wasn't ever going to get flat due to a 10dB rise above about 8k, at which point I connected the dots and figured it was the EQ. I disabled the EQ and everything went much better.

For sweep settings I bumped up to 128k sequence length, but I think everything else is default. The "Center peak of impulse response" checkbox is required to get a distortion measurement, but I found that would change the gate position when switching between drivers so I left if off to get FRD files.


The sequence will be run without any changes to mic, speaker position, gates, or amplifier settings.

• Measure tweeter, save FRD
• Measure woofer, save FRD
• Measure both together, save FRD

Once the sweep is made and my gates are set (left click right before the impulse, right click right before the first reflection blip). I used the "Unsmoothed DFT frequency response / spectrum" analysis to get the response. First, under the View menu make sure you have it set to "Magn+Phase" or your FRD will be saved without phase data. Then you hit File - Export - ASCII file to save the response. They'll need to be edited to remove the top comment lines, and then changed to a file extension of "frd" to be useful.

One snag I ran into with this setup is the complete lack of level calibration with this mic. The output from ARTA is accurate for the frequency response, but shows up in the -10dB range rather than the actual SPL. That caused WinPCD to puke on import with out of range data, so that had to be adjusted.

Adjusting required no extra software, just Excel.

• Import data from text file, tab delimited
• Columns are frequency, amplitude, phase
• Use a formula to add a fixed number to the center column
• Make sure you have 3 columns in the same order
• Copy and paste into Notepad, the default for this is tab delimiters so no work needed
• Save as a new adjusted FRD file

I added 80dB to the files to get them into a useful range. The SPL displayed in PCD or XSim won't be correct, but it's the response that matters for this.

[wogg]

Now the files went into WinPCD to find the Z offset, and start on the crossover. Note that I'm using the gated results, and did not bother adding in close mic data below 300Hz. So the graph under 300 is invalid.

Finding the X offset requires you to compare the measured combined response with the sum generated by software. First add your woofer and tweeter FRD files and let them run full range without any crossover selection. Then in the System Response Graphs window, click Import Overlay and pull in your combined measurement FRD file.

This is my driver set example. Note the little notches in there, those coincide with funny phase blips in the FRD file, I'm ruling those out as not real.



Next, make sure the "polars active" button is on and start moving the woofer Z value. You can do this with the up and down arrow keys, leaving the system response graph on the screen so you can watch it change real time. Once you hit the right Z value the system response will align with the measured combination. In my case it looked like this:


That shows a Z value of 0.028 for my drivers, about 1.1 inch. That seemed reasonable for a flush mount dome to a RS180P woofer. And now I'm into crossover design, limited to 5 components. For targets I'm trying for 2nd order butterworth, but really that flies out the window when I'm avoiding extra components. So I'm in a trial and error loop, sticking with second order electrical on the woofer and tweeter with a single series pad resistor on the tweeter.

So far, I've come to this result. This hits my +-3dB target, but I'm not sure I like the aberration in the 4-6k range. That looks like it would be audible under a comparison like InDIYana. I'm going to keep plugging at it. That dip at 5.5 / 6k is a stubborn bugger, it doesn't seem to react to filter changes much. Also, I've ended up with my tweeter pad resistor before the crossover network, I found the system impedance would end up more like a 4 ohm speaker with the pad after the network.




Next up... getting this crossover as good as I can with only these 5 parts... and trying the close mic measurements + port combination to get the real bass response.

[Mountainman Bob]

What happened to the part about the imm 6 ?

i went to look for the downloads you were using and now it’s gone!