SevenSixTwo - InDIYana 2018 Coax Design

[Kevin K.]

Man you pulled that together fast! Considering what you started with, the old wise tale expression "forging diamonds from coal" comes to mind. Looking forward to seeing how the measurements come out with your crossover. Good luck!

[KEtheredge87]

Man you pulled that together fast! Considering what you started with, the old wise tale expression "forging diamonds from coal" comes to mind. Looking forward to seeing how the measurements come out with your crossover. Good luck!

Thanks Kevin and Ben! I might have obsessed over this a bit this week. I'm also excited to see what can be done here. Some of those filters on the tweeter modeled as being mighty sensitive to deviations. I'm tempted to grab some of those 1% tolerance caps from Dayton on those parts. I'll finish my shopping later tonight and see what the bill comes out to be. -- Sent from my SM-N900V using Tapatalk

[KEtheredge87]

Alright guys. I held off on ordering parts long enough to make a few tweaks to the XO. I increased the Y-axis scale until I got 1 dB per line so I could see the fine effects of changing values (FR graph attached). This allowed me to see exactly what I was changing. Unfortunately, I haven't been able to get rid of a little cancellation that PCD says is happening on my HF horn section between ~3K and 5K, but it's slight (~1/2 dB or so). I'm fine to leave this for now and measure to see what it really sounds like. (in reference to the next bit of info... I don't see the exact same amount of reverse null in XSim as I do in PCD... not sure what the difference may be. I included the acoustic offsets calculated from PCD in my XSim file)

In order to get a handle on layout and power dissipation, I recreated my final prototype XO in the XSim program. I like to do this in order to check expected power dissipation for resistors and the power levels seen by the drivers themselves. As expected, I am having to sink off a LOT of power to bring the 103 db @ 1W/1m high frequency horn driver down to the efficiency level of the Esoteric 7 woofer. I'm having trouble deciding the most practical way to address the 20 ohm notch filter resistor that goes before the XO circuit. If I used one 20 ohm resistor, Xsim says I'd be pushing over 35W of power through that single component with a 200W power source at 4 ohm, or 70W at 8 ohms.

I know I need to use some combination of higher power rated resistors and/or use them wired in parallel to distribute the load. The least exotic route I could think of is taking eight 40 ohm 10W dayton audio resistors and wiring them in series-parallel (two 40 ohm resistors in series x 4 banks in parallel) for an overall 20 ohm resistance. The diagram I attached shows this in the schematic, but this feels like I may be missing something.

So... XSim experts (read: anyone else besides me!) How am I really supposed to use the power amp tuning feature in XSim? Should I set it for 4 ohms instead of 8 ohms since the minimum system impedance of this design is 4.3 ohms at 1.8 kHz? I don't really expect I'd ever put 300+ watts of amplifer power to these speakers, but I want to make sure I've spec'ed the components out safely for an XO that will get installed in a sealed chamber and covered with polyfill. Overheating resistors is a non-starter.

Thanks in advance for any advice on using this program properly, or a smarter way to achieve the power handling and 20 ohm resistance I designed around.

[Paul K.]

Maybe this is kind of a stupid question, and I may have missed something you stated earlier, but when you're trying to determine how much power the 20-ohm resistor will be dissipating, what is the SPL level you're asking the system to generate and is it a level that's continuous or transient?
Paul

[KEtheredge87]

Maybe this is kind of a stupid question, and I may have missed something you stated earlier, but when you're trying to determine how much power the 20-ohm resistor will be dissipating, what is the SPL level you're asking the system to generate and is it a level that's continuous or transient?
Paul

Hi Paul! I don't believe in stupid questions... especially when it comes to speaker building and electronics!

I am honestly not sure about the SPL level, but I would assume the input would be kind of transient. I'll try to explain my thoughts a bit more. I know that Xsim shows me a response level in dB on the frequency response graph's y-axis, but I don't think those values really mean anything unless I took the measurements at 1W/1m (just like the sensitivity ratings are posted on manufacturer data sheets for each driver). Ignoring whether that's right or wrong for a moment, when I set the Xsim power amplifier to 300 watts at 8 ohms, it tells me the system output is around 96 dB. It also says my power dissipated through a single 20 ohm resistor is 108 watts at 1960 Hz.

If I think realistically, there are a few things that come to mind.
1) I don't currently own an amplifier that can drive 300W continuously, and most amplifiers don't go there unless we're talking sound reinforcement at a club rather than enthusiast home audio.
2) As I happen to like my ears and their ability to hear things, I wouldn't listen to it at 95 decibels all the time (or likely much ever)
3) Music is dynamic, so even IF there were 300W outputs from the amplifier, it wouldn't be continuous, thus I could likely get by with less power handling capability for this resistor than the math would suggest.

At the end of the day, I'm pretty sure what I have sketched up right now is gross overkill, but I don't know what the right amount of power handling and resistor load sharing is optimum either!

Again... If I'm doing something silly, please don't hesitate to point it out. I won't learn otherwise!

Thanks for helping me along!

[jhollander]

As program material is not constant I derate the wattage by 50%. If you adjust the xsim amp to a normal listening level you can derate further.

[augerpro]

Did you measure at 20 or 40 degrees as suggested? How does that look?

[KEtheredge87]

Did you measure at 20 or 40 degrees as suggested? How does that look?

Hi Brandon,

Yes. I took 20 degree off-axis measurements of the coaxial driver and ran them through the response blender. The 20 deg farfield mid-cone data got blended with the nearfield data, and the high frequency tweeter data had it's tail trimmed. I ended up with another blending anomaly though, as you can see by the odd spike in the data at 1548 Hz. The Fs of the tweeter was measured as 1196 Hz while installed in the cabinet with stuffing, so I don't think it has anything to do with that.

Overall I don't see anything that looks like a horrible shift. The grey line in the plot shows the 20 degree of-axis response I measured, where the black line is the on-axis response. Things only dip around 2 dB or so between 3k and 6k in the off-axis measurements, and the on-axis dip at 8k looks like it filled in and went above my woofer reference line when going off-axis. In my super limited experience, I think this looks passable. Hopefully it SOUNDS passable too!

Any other charts you'd like to see? Wasn't totally sure if there was other stuff that's common to inspect in off-axis vs. on-axis comparisons.

[jhollander]

It looks like you have extra energy above 10K both on and off axis. The 7,800 dip filled in nicely. Might try optimizing the off axis response with a slight downward tilt

[PWR RYD]

I look forward to hearing these at InDIYana.

And I'm sorry if I missed it but did you finish that other project (Super Bees's?) with the Dayton NE25FW tweeters? If yes, will they be there too?

[KEtheredge87]

I look forward to hearing these at InDIYana.

And I'm sorry if I missed it but did you finish that other project (Super Bees's?) with the Dayton NE25FW tweeters? If yes, will they be there too?

Hey Craig,

You haven't missed it. I got heavily side tracked into building these coaxes in time for the competition. When I last fiddled with the SuperBees, I had two versions of the XO built, but I wasn't entirely convinced I had it done. The first version had some fuzzy/boomy business going on at heavy acoustic guitar strings. The second version addressed that fairly well, but then I wasn't convinced that some tweeter distortion or some other artifacts hadn't been introduced in the process. I left some more details on what I thought I heard in that thread, as well as a youtube link to a video showing tweeter distortion that kind of sounds like what I was hearing. No one responded beyond that post (that I saw), so I haven't changed or finalized that bit yet.

That said... I do want to bring them to InDIYana as well. Once I get the coax XO going, I'll have to devote my attention to finishing both of these up concurrently. Nearly two months to go!

[PWR RYD]

Hi Keith,
Bring them regardless. Even if they aren't "done". After hours listening sessions are a great time to get multiple ears on them and most likely some constructive criticism and ideas.

[KEtheredge87]

Whew! Alright guys... after a bit more fiddling around I am calling the prototype XO finished until listening takes place. I did some more checks on my resistor power dissipation in Xsim and found a few more places that were predicted to eat loads of power. I decided to have some fun with this and try something a little different. In order to minimize the number of places where I would need a series / parallel arrangement of padding resistors, I ordered some of the "Zister" products that have 25W ratings. This should keep the XO board a bit less cluttered while providing the power dissipation headroom that's needed.

I have cleaned up my prototype Xsim arrangement and documented it here for grins. Once the parts come in, I also have 5 places where I needed to unwind some inductors to hit custom values. That'll be a tedious evening! I hope these sound good, since my introduction to 3-way XO's with these has been an exercise in finding a bigger hammer every time I need to change something!

Current XO Tally sits at 30 parts / $134.35 per speaker. I'd better figure out how to make my coax chamber act like the TARDIS, as it may need to be bigger on the inside to accommodate all the electronics!

[Kevin K.]

Keith, it's your baby but I'd suggest trying the crossover the first time without unwinding the inductors? Especially the .27 coils, you're talking about less than a 5% change in value. Might be worth a look just to see how it actually measures.

[KEtheredge87]

Keith, it's your baby but I'd suggest trying the crossover the first time without unwinding the inductors? Especially the .27 coils, you're talking about less than a 5% change in value. Might be worth a look just to see how it actually measures.

Hey Kevin, thanks for the suggestion. It certainly can't hurt to try it. The only reason I went this way was because of chasing very small predicted changes in PCD. We'll see how they pan out in reality before making any changes to inductor winding! I love a good science experiment.