Dayton 3-way Build Thread (RS225 / RS52FN / RST28F)

[crackyflipside]

Dry fit of the baffle before sanding and finishing.


Final crossover design with baffle diffraction compensation on mid and tweeter.


Crossover design before baffle diffraction compensation, to show original results


FR accounting for baffle diffraction and with corrected crossover padding on mid and tweeter.


The actual size is 11-1/4" wide. Depth of the box is 14", ported to rear.


2330 in^3 (38.20 L), tuned at 37hz, using 3" round port 8.08" long.


Those built in grills are so pretty.


Gotta design some grill protection for these WOOFS.


2x12 Poplar from the mill, it was about $10 per linear foot. Green, but some oxalic acid bleach gets the sucker brownish/orange.


Jasper Jig is a gift from the heavens.


1 down






Tee-nuts to make sure I can take the speaker apart! These will also go onto the back panel in case I want to adjust the crossover after taking measurements.


After sanding to 180-grit and oxalic acid bleaching.

End goal for the baffle to be high-gloss natural finish, very light 'natural' stain color, but first gets oxalic acid bleached to take the green out of the poplar. The rest of the enclosure to be 3/4" MDF, satin black finish.

Still in progress build. Just finished the rough cuts and router work. I built speakers from a CJD design (Khanspires) years ago and wanted to make my own from scratch. Tried to watch a ton of youtube videos and tutorials and cooked up a crossover, so it might be trash, but I'll just adjust after measuring. First was going to build a simple 2-way with 6" and tweter, but I wanted more bass extension so grabbed the 8" with mid/tweeter. VituixCAD gave me a ton of help after first designing the crossover in Xsim, both in that crossover wizard thing they have for figuring out values and the baffle diffraction simulator. I got the curve from the baffle diffraction sim, threw it back into Xsim, then drew up the padding circuit for mid and tweeter.

Was able to get a lot of work done in the past few days after taking some vacation days off from work.

[Dave Bullet]

Looks great! How much bsc have you allowed for? Where will these be placed in proximity to your back wall?

[crackyflipside]

I got the inductor and resistor values from some car audio calculator site. I wish vituix could apply the diffraction data easier to the crossover part. cause I had to generate an frd and import it like a house curve on Xsim to work around the values for it. If it doesn't work in gated measurements I'll just start messing with different values on the pads.



Stain conditioner applied, then natural stain after.



First coat of poly, gonna see what two coats looks like after sanding flat and run a polisher on it with cutting compound

[crackyflipside]

Sanded the first two coats of poly, was pretty wavy so I ended up burning through some of the finish while sanding it smooth.

This time I brushed on a much thinner coat and it was more even and smooth than the first coats. It's nice to brush on the sanded parts.

I used a 3M sanding pad (kind of like a sponge thing) and the results were real good for the roundover

Cut the MDF for the rest of the speakers, going to try to router out the rear port tomorrow and sand the poly again for another coat.

[5th element]

Great work I haven't seen anyone use the fabric dome mid yet! Do let us know what you think of it when you are done.

[Dave Bullet]

I still think you won't be satisfied with the bass response. The speaker will sound thin. See how your response curve climbs from 100Hz? You'll need more than LPad tweaks on the tweeter to restore bass balance.

I think your net sensitivity to allow for 3 - 4dB of baffle step compensation will end you up around 85dB. You're running around 3 - 4dB too hot through the upper bass / midrange and treble.

Note: Baffle step compensation is different to baffle diffraction.

[crackyflipside]

I still think you won't be satisfied with the bass response. The speaker will sound thin. See how your response curve climbs from 100Hz? You'll need more than LPad tweaks on the tweeter to restore bass balance.

I think your net sensitivity to allow for 3 - 4dB of baffle step compensation will end you up around 85dB. You're running around 3 - 4dB too hot through the upper bass / midrange and treble.

Note: Baffle step compensation is different to baffle diffraction.

Yeah, I planned making the back panel removable with t-nuts and run the crossover outside to tweak using mic and REW measurements. The front baffle is about two feet from rear wall in the room. You don't think higher value resistors will be enough to pad if I need it?

[Dave Bullet]

You're looking to flatten the response from around 100Hz. Adding resistors will only get you so far. Phase tracking can go out the window and may need topology changes.

For example - In most 3 way designs I've seen, at least a 4 mH inductor is used on the woofer to flatten the curve from 100Hz to 500Hz. You can use a parallel resistor (or several depending on the level of cut and inductor value and powerr handling) but a larger inductor often handles double duty for BSC.

Then the midrange will need padding.... and this may involve inductor and capacitor changes as well as resistor padding.

And further padding on the tweeter beyond 1 - 2 ohms in both legs will usually require crossover adjustment there too.

[crackyflipside]

You're looking to flatten the response from around 100Hz. Adding resistors will only get you so far. Phase tracking can go out the window and may need topology changes.

For example - In most 3 way designs I've seen, at least a 4 mH inductor is used on the woofer to flatten the curve from 100Hz to 500Hz. You can use a parallel resistor (or several depending on the level of cut and inductor value and powerr handling) but a larger inductor often handles double duty for BSC.

Then the midrange will need padding.... and this may involve inductor and capacitor changes as well as resistor padding.

And further padding on the tweeter beyond 1 - 2 ohms in both legs will usually require crossover adjustment there too.

I see what you mean now.

Xsim has it looking like the 4mh on the woofer will get me that rolloff to comp for that 700hz hump, change the resistor padding on tweeter, and it's the tweeter that will need another shunt to tame a new hump at ~7khz.

I'll definitely keep that design in mind once I get to start measuring the speaker.

Thanks!

[Dave Bullet]

Also, have you fixed your phase issue between 1.5 and 3Khz where the midrange is greater than the system response?

[Dave Bullet]

I'm actually rethinking your crossover points.

You have the RS52 crossed over way too high ~ 6.5KHz. The old RS52 (I'm assuming the new will be similar) had dome breakup at 15KHz. This will translate into a 3rd order harmonic "peak" at 5KHz.

I'd recommend crossing over no higher than 3 - 4 KHz - preferably 3KHz as the RST28 tweeter can take it. Phase alignment becomes even more critical as you go up the spectrum, so crossing over lower will help.

AS for the low end - Zaph (zaphaudio.com) recommends as low as 500Hz for an LR4 slope. You could take the RS52 down to ~700Hz no problem at LR4.

So try redoing your xo for 700 / 3KHz crossover points (or thereabouts), instead of 1.1Kz / 6.5Khz

[Bear]

See if this helps with BSC.

https://trueaudio.com/st_diff1.htm

TMM (Tim) posted measurements on his website of the RS52FN not too long ago.

http://feleppa.com.au/speakermeasmid.html

[crackyflipside]

The fabric RS52 didn't have that nasty 15khz peak people had issues with.





I will try out the crossover points /phase you suggested and see how it turns out.

On a side note, just got the ports cut, and all the MDF for the sides and bracing. Also sanded again and put a thin coat of poly. It's getting very easy to put on the poly, much smoother than the earlier coats.

Hopefully this weekend I'll get the back of the speaker put together with all the MDF.

[Dave Bullet]

Ahhhh - sorry the old RS52 was metal domed? My apologies

[Dave Bullet]

Hows it going buddy?

[crackyflipside]

Nothing much, might put another coat of poly on today, haven't been able to router out the brace yet cause the weather been crap.

I was modeling a new crossover to fix the phase things with the mid, but then it gives me phase issues with the woofer. I couldn't figure out how to squeeze out a lower xover point there either without the system FR looking really wonky. I'd have to completely redesign the whole thing versus swapping out a couple components, which is an issue because I'd already bought the original crossover parts.........

So I'll try assembling enclosure this weekend, put together the original crossover and diffraction padding, and run it. If I can't use simple padding to get diffraction under control and the sound makes me want to gouge my ears out after running an eq (Audyssey), I'll go with the xover redesign.

[Dave Bullet]

I don't mind having a go at the crossover phasing. Could you upload your FRD/ZMA files please? Just let me know if you have already included baffle diffraction or step in the files, so I can take that into account. If not - that's fine, I can apply myself. Just need to know my starting position.

If of course, you'd rather do it all yourself, that's ok too. It's a personal hobby and triumph after all when you've done it yourself. ;-)

[5th element]

With the off axis curves of the dome mid you really want to cross it at 2.5kHz.

[crackyflipside]

Here's the curve with really minor changes and the polarity fixed between mid-tweeter. It tracks the diffraction curve really well now +-1db most of the FR.

Best of all, it only changed values on the resistors and caps (only one very small inductor), so it won't destroy my wallet!

The only issue now is speaker impedance looks like 2ohms now.









See changes in yellow compared to original design below.



Edit: made another change, added 5.6ohm resistor parallel to the 68uF on the woofer, smooths a big hump there and I can recycle it from another resistor I already bought. Not perfect tracking at 100-500hz for diffraction but closer.

[Dave Bullet]

I think you might have diffraction ripple under control (>1Khz) but the baffle step isn't taken into account.

Have a read of this - it's great at showing a raw response (similar to yours) and how the rise in frequency response as the frequencies begin to project off the baffle (2 pi radiation) kicks in (meaning increasing response as you go up in frequency).

http://audiojudgement.com/baffle-step-calculator/

The guy uses a resistor in parallel with the woofer's inductor to reduce the output. Another alternative is to use a higher value inductor.

Another good article is here which shows baffle step occuring on all baffle types. The ripple is caused by placement of drivers on a baffle. but the rise (baffle step) is evident in all enclosures and is a function of enclosure size / shape and independent of the driver mounted.

https://www.trueaudio.com/st_diff1.htm

[5th element]

Why are you crossing so high between the midrange and the tweeter? You can see that the mid starts to become directional above 2.5khz, which is something you want to avoid. You've got this super capable tweeter that can cross as low as 1.5kHz, why not make more of it?

[Dave Bullet]

Why are you crossing so high between the midrange and the tweeter? You can see that the mid starts to become directional above 2.5khz, which is something you want to avoid. You've got this super capable tweeter that can cross as low as 1.5kHz, why not make more of it?

Depending on slope and how low he might cross the mid to the woofer wouldn't phase tracking across all 3 drivers become more of a headache if the crossover points are too close. Assuming he goes down to 500 for the mid, wouldn't it be simpler to phase integrate if the tweeter crossover is at least 2 octaves away? I'e 2000 Hz?

[crackyflipside]

Been a while, I will get more pictures up soon. Spent all day yesterday painting the back of the cabinets with black lacquer spray paint. Gonna sand and then apply a coat of clear tonight after work.

All the talk here has me going to return the crossover parts before the days pass too much, and starting the electronic design from scratch with different crossover points. Most of all I'm trying to focus on getting a better speaker impedance because everything I'm reading has me nervous of screwing up my amplifier haha. Also figured out how to import the baffle correction into Xsim and it makes it way easier to design filters to be flat relative to that diffraction/baffle step.

[Dave Bullet]

Been a while, I will get more pictures up soon. Spent all day yesterday painting the back of the cabinets with black lacquer spray paint. Gonna sand and then apply a coat of clear tonight after work.

All the talk here has me going to return the crossover parts before the days pass too much, and starting the electronic design from scratch with different crossover points. Most of all I'm trying to focus on getting a better speaker impedance because everything I'm reading has me nervous of screwing up my amplifier haha. Also figured out how to import the baffle correction into Xsim and it makes it way easier to design filters to be flat relative to that diffraction/baffle step.

If your amplifier has any safety circuitry - you should be ok. They'll cut out if impedance gets too low / output transistors get too hot.

Before you hook up your amplifier, make sure you use a multi-meter to check the DC resistance. just to make sure you don't have a dead short.

Of course DC resistance (nominal) will be higher than the minimum impedance in a system actually being used. If you can do an impedance sweep (using appropriate hardware or software) - all the better. Without that - you can largely rely on your simulation to show you the minimum.

From my experience, 2nd order electrical woofer to mid is higher impedance than higher order filters. For example my 3rd order filter on the woofer to get a 4th order acoustic target was dipping to 3.1 Ohms. this generated considerably more heat for my amplifier than the 2nd order electrical (with notch) to target a 2nd order acoustic slope. This now has a 4.3 ohm minimum in the midbass. My amplifier plays cooler as a result (only warm - not warm-hot).

I also think the speaker sounds more integrated / smoother with the larger in phase overlap between woofer and mid.

The downside (there is always one) is the mid now has more excursion with also 2nd order slopes. So at higher volume I'm potentially getting more non-linear distortion in the mid/upper bass than before.

[crackyflipside]

Without diffraction/baffle step looks like:



I'm working on this one, which gives way better impedance than the last revision. The woofer 5-6ohm, mid and Twitter around 4ohm with the only dip near 2ohm at 17k-20khz. I didn't have time to screenshot the impedance/FR before going to work, but I figured out how to format the calculated diffraction file to put into xsim.

Reversing the mid gives the best phase integration with the tweeter but there is a weird null of about 1dB in the woofer, but I won't sweat that. Xover point on the mid is 680-3khz, any lower and phase issues get real magnified on woofer and any higher messes with tweeter phase; I don't want to push a 2"mid much lower than that. Phase flips right in the middle of the mid FR, and the mid and tweeter individual phase track very close together.

It addresses diffraction ripple in the mid and there's a 4dB baffle step correction in the tweeter. I think the full 6dB would be a little aggressive in the room. I'll keep on working on it, but I really appreciate the input you guys are giving!

Edit: got all the graphs now and did a little change on some caps

[Dave Bullet]

I think your 2nd graph has too much bsc. You've got an 84dB system and you want to target around 86 (I originally said 85 but I think the sensitivity of the 225 is around the 86 mark after baffle step)

You also have a suck out where the woofer crosses with the mid. This explains the odd "only 1db null" when you flip polarity of the mid assuming you are targeting even order slopes which are by definition 180 degrees out of phase for 2nd order acoustic meaning absolute polarity is reversed between adjacent drivers or 360 degrees for 4th order meaning same absolute polarity.

Basically the black line should always be greater than or equal to all individual driver frequency responses or you are getting destructive interfere between drivers (assuming even order slopes here). To fix you want the mid slope to not be as aggressive. I.e shallower. Try increasing the 2.2mh inductor on the mid and adding a resistor to it.

[5th element]

Depending on slope and how low he might cross the mid to the woofer wouldn't phase tracking across all 3 drivers become more of a headache if the crossover points are too close. Assuming he goes down to 500 for the mid, wouldn't it be simpler to phase integrate if the tweeter crossover is at least 2 octaves away? I'e 2000 Hz?

Oh I wasn't advocating that he cross as low as 1500Hz, just that the tweeter is capable of doing so. A cross of 2.5kHz would make the most sense although realistically you'd be far better off using a small neo dome with a crossover that 'high'.

With respect to the phase tracking being harder, maybe? The mid would be more of a filler if you were using it between 500 and 1500Hz though.

[crackyflipside]

Still fooling with the crossover design, returned the old parts yesterday, but at least the enclosures are finished. Next time I'll find a cabinet guy to build finished boxes.

[Dave Bullet]

They look great Chris! Well done!

[BobEllis]

As a builder you always see even the most minor flaw a lot more readily than anyone else will. Take heart, nobody else notices whatever it is you see. Your next build will be better, even without outsourcing the cabinet. I agree, they look great. Thanks for sharing.