InDIYana build - SuperNova Minimus

[wogg]

I'm leaning toward the series still. Here's the results with a 6th component, a small inductor in the tweeter circuit to drop down that top end rise. This hits +-2dB with an easy 8+ ohm load.



That 6th component can't be used for inDIYana, but I can lay it out on the board and pop it in afterward for minimal cost.

[4thtry]

Excellent. Looking forward to hearing these next month.

[jhollander]

Just guessing...on the 5 component parallel, did you try a small cap (1.X uf) and big inductor (2.X mH) on the tweeter? I'm thinking you could get a bit more lift in the dip

[chrisn]

Since you're limited on XO components, maybe some felt on the baffle would smooth the tweeters response?

[wogg]

Just guessing...on the 5 component parallel, did you try a small cap (1.X uf) and big inductor (2.X mH) on the tweeter? I'm thinking you could get a bit more lift in the dip

Interesting idea. I did just go extreme with the values, and the tweeter output basically turned into a straight upward slope, no knee and not usable. I think the problem is that depression is above the crossover target, so any under damped knee I could induce just hits below that problem and makes it worse.

Here's the PCD screen for reference, with that LR4 target.

[wogg]

Since you're limited on XO components, maybe some felt on the baffle would smooth the tweeters response?

That could be a good idea. The problem may be baffle induced, perhaps I'll get a chance to try this.

[wogg]

I don't think my moderated post about the measurement process will ever come back at this point, so I've more or less re-created what I had said. In doing so, I just realized that even though I have my mic calibration data in ARTA, it wasn't actually being applied. So this afternoon I'll be re-doing my FRD files and checking the crossover sims.

This should have been on page 2, right before post #28:

Measurement Setup
My super budget setup includes only an iMM6 microphone and my trusty Microsoft Surface Pro 3. The iMM6 plugs right into the headset jack and becomes available as an audio input and output in Windows. For my application, I’m using ARTA for measurements.
Every iMM6 comes with a calibration file that you need to get from Dayton by following the included instructions, load that as FR compensation under the Setup – FR Compensation menu.



Once loaded, click “Use frequency response compensation” to apply the calibration to your measurements.
With the Surface Pro, since it’s essentially a tablet, I use a basic music stand to get the mic at tweeter height and set it up by eyeballing the tweeter axis and measuring the appropriate distance. For this project I used 1M, 39” but you could go closer provided you’re in the far field. I forget the calculation based on baffle dimensions for this, but Jeff Bagby’s white paper on quasi-anechoic measurements details this extremely well.

With this all set, you’re ready to make measurements. The Impulse response measurement / signal recording screen is where you’ll be.

Here are my recording settings, the sequence length is pushed up from default to 128k, but everything else is default. For distortion measurements, the “Center peak of impulse response” check box is required, but I found that changed the location of the gate settings so that was left unchecked for the crossover FRD files.



I setup the level by switching to the periodic noise tab and firing up a pink noise signal. I then used the Audio Tools app on the phone with the iMM6 to set the average level to 80dB for the woofer. Now the goal is to leave the amp level, mic position, and resulting gate settings alone for all measurements.
The measurements were made in this sequence:

• Tweeter alone
• Woofer alone
• Woofer + tweeter in parallel

It’s super important to leave all settings the same for these measurements, they will be used to find the Z axis offset later.

Once the impulse is taken, set your gates with a left click for the beginning, and a right click for the end of the impulse right before any visible reflections. Click the “DFT” button (Unsmoothed DFT Frequency Response / Spectrum) to get the results. This is where you’ll export an FRD file. Make sure you change the view menu to “Magn+Phase”, otherwise your FRD file will not contain any phase data.
Click File – Export – ASCII file to save your data. This will spit out a “txt” file, which will need a little editing and a rename to be useful. In Notepad, delete the top comment lines, leaving only the numbers. Then change the file extension from “txt” to “frd” for use in other programs.

Level calibration issues
Since I don’t have any level calibration with this setup, the output of the response measurements sit in the negative dB territory rather than at actual SPL. This doesn’t affect the response curves, but throws PCD for a loop and can’t be used. So the dB values need adjusted to fit in a program. This can be done with Excel.
Use the import text tool to bring in the FRD data into Excel, it is tab delimited text. Once you have the data in there, the first column is frequency, followed by amplitude and phase. I used formulas to mirror the 3 columns over, and added a fixed number to the amplitude of 80dB to bring them up to a useable level. This doesn’t relate to a real SPL level, but that doesn’t matter so much for the crossover design. To save back out of Excel, simply copy and paste your 3 updated columns into Notepad. By default, Excel data is pasted as tab delimited when it goes into text, so it will be ready to use. Just save the contents as a new adjusted FRD file.

[wogg]

Holy moly.... updated measurements made a big difference. Same measurement technique above, with the FR compensation actually turned on this time. The sequence was slightly different since I found the tweeter had the best source to set the gating correctly right off the bat, which makes sense because the sharpest edge of the impulse is the high frequencies:

• Tune SPL using the tweeter this time, pink noise at 80dB average
• Sweep tweeter first, set gates
• Sweep woofer + tweeter
• Sweep woofer

The Z offset actually changed a bit, which surprised me. This time I got 0.511" offset (0.013M in PCD) to align the system sum with the overlay. Now the parallel crossover looks pretty good, no more 5+dB dip at 4kHz. The tweeter padding was adjusted, but the LC portions of the crossover were able to remain the same and work well. Tweaking the values didn't look like it would get too much better.

This is the sim I'm going to build in real life now:



I find this response acceptable, with a 20+dB reverse null. The impedance does dip to 4 ohms at 3kHz, but that shouldn't be a problem for most amplifiers, the low frequencies stay in 8 ohm nominal range.

[chrisn]

0 component solution, very nice

[wogg]

Yay parts! Including some complementary resistor chips. I'll have to get customer service on that one, the big air cores beat the resistors up good.

[wogg]

For those that may not know this already...

PE Customer Service is excellent.

Replacement resistors are on their way. Also... they always ask if I bought through Amazon, which I never do. IMO... never buy PE stuff through Amazon, they don't need the money and dealing direct with PE is super easy.

[Mountainman Bob]

For those that may not know this already...

PE Customer Service is excellent.

Replacement resistors are on their way. Also... they always ask if I bought through Amazon, which I never do. IMO... never buy PE stuff through Amazon, they don't need the money and dealing direct with PE is super easy.

I agree with customer service being top notch, but shipping dept should step it up a notch, my last order was lucky to make it.....the inductors and caps were packed so loosely they ruptured the packing tape while ups played soccer with the pkg.....a whole side of the box was wide open.

Better tape and more packing are in order......they should know it’s in for a rough ride no matter who carries it.

[hittheskids]

These will be a vintage speaker re-build. I have 2 pairs of Realistic Nova 15's from the mid 80's to use, one pair will be my 3 way for MWAF, this pair will be a more direct homage to the Realistic original as a 2 way with an 8" (ish) paper woofer.

Behold... the original Realistic Nova 15:



The tweeters were already purchased last year at the MWAF tent sale: Peerless DX25BG60-04
The Fs is low on these, and they should be able to reach near a 2k crossover point. There's some wiggle at the top octave in response, however that wiggle disappears off axis so it doesn't worry me much.

After a lot of woofer modeling, starting at response curves, and some preliminary PCD work to see if a 5 part crossover would work, I landed on the Dayton RS180P-8 for the woofer choice. They won't get the full bottom end that I'd like (<50Hz, preferably 40), but they should reach a solid 60Hz, which conveniently makes them match the stated specification of the original Realistic design. The regular RS180 does better in this cabinet, but the breakup makes a simple crossover impractical.

Woofers are on their way this week, and after cutting out the baffle the experiments to add rigidity to these old cabinets will begin!

Hey... I recognize those cabinets . I really like the way you decided to do the baffles, looks like they turned out great. Good luck the rest of the way, I'll be excited to see your description of how they sound.

[wogg]

Beautiful day here in Columbus. On with the finishing!

A little chamfer action and some bondo for the port screwup earlier.

Hit the bondo with the router, then some sandpaper to get it OK again.

[wogg]

Now we have the first coat of Duratex.

While that's drying I added some damping to the back and upgraded the input terminals.