The Joudas project

[trevordj]

I built a prototype crossover. The speaker has plenty of bass and male vocals sound good. The tweeter sounded a little shrill particular with higher female vocals. I wasn't sure how accurate my ears were so I took a measurement. It turns out pretty accurate in this case. I am going to build the other two crossovers and see which one sounds the best out of the gate and then start tweaking it.

The first one:



Impedance predicted vs measured:





FR predicted vs measured:

[rpb]

Don't discard this one just yet! I don't know if you are using jumpers, or if you soldered things in place. First thing to do is measure the woofer with the xo, but disconect the tweeter xo section. Make sure the woofer is doing what you want. Then, alter the padding on the tweeter, and drop its level about 10dB. Measure the tweeter section with the woofer section disconnected.

I would suggest taking a raw measurement, and then a measurement with a x-over, and put them on the same graph like this previous example. Try to have the graph extend 40dB or more below the desired response level.

[trevordj]

Don't discard this one just yet! I don't know if you are using jumpers, or if you soldered things in place. First thing to do is measure the woofer with the xo, but disconect the tweeter xo section. Make sure the woofer is doing what you want. Then, alter the padding on the tweeter, and drop its level about 10dB. Measure the tweeter section with the woofer section disconnected.

I would suggest taking a raw measurement, and then a measurement with a x-over, and put them on the same graph like this previous example. Try to have the graph extend 40dB or more below the desired response level.

Cool, thank you. I will do that. I quickly soldered the leads in place but it can be undone easily.

Before I saw this message I built crossover #2. Immediately, it sounds much better. That harshness is not there and, overall it just sounded nice. As it is this is a speaker I would be happy with. Its an expensive crossover but I like it. I may try to remove the notch filter and see how it sounds as the measurements reveal a 5-6dB dip at about 1300hz.

[rpb]

Odds are the dip is a summation issue. My advice is pretty much the same as the previous post. Make sure the woofer response is where it should be, and then make sure the tweeter response is where it should be. If they don't quite add correctly, some tweaking can probably resolve that.

[trevordj]

Odds are the dip is a summation issue. My advice is pretty much the same as the previous post. Make sure the woofer response is where it should be, and then make sure the tweeter response is where it should be. If they don't quite add correctly, some tweaking can probably resolve that.

Right on, thank you for your feedback. After the kids go to bed I am going to take the measurements as you suggest. -Trevor

[trevordj]

Ok, here we go. This is still crossover #2

Woofer:




Tweeter:



System:

[rpb]

The dip is a summation issue. You could listen to the pair, and just enjoy them for a day, and get a feel for the overall tonal balance. When you are ready to continue, you can alter the slopes to see if you can get a better sum. It's hard to tell by looking, but I think I'd try a 4th order on the tweeter, and see if that works better. Before you do that, measure a few degrees above, and below the listening axis, The dip will probably fill it in one position, or the other. If it fills in above the axis, then the 4th order on the tweeter might help. As an option, you could let the woofer go a little bit higher. Small changes to the woofer filter. At this point, it may be frustrating, or it may fall in line quickly. It's not always easy, but sometimes it is! To me, it's always fun!

Second guessing myself, I'm thinking a sharper knee on the woofer response might work. Try to get the woofer response up about 2dB at 1k, and then make it bend downward a bit less gradual. (ie, a hard turn downward.)

I'm just thinking out loud. There will be a way to make them blend. It's not always obvious.

[trevordj]

The dip is a summation issue. You could listen to the pair, and just enjoy them for a day, and get a feel for the overall tonal balance. When you are ready to continue, you can alter the slopes to see if you can get a better sum. It's hard to tell by looking, but I think I'd try a 4th order on the tweeter, and see if that works better. Before you do that, measure a few degrees above, and below the listening axis, The dip will probably fill it in one position, or the other. If it fills in above the axis, then the 4th order on the tweeter might help. As an option, you could let the woofer go a little bit higher. Small changes to the woofer filter. At this point, it may be frustrating, or it may fall in line quickly. It's not always easy, but sometimes it is! To me, it's always fun!

Oh ya, this is great! I’m loving every minute of it. I’m excited to see and hear how well these tweeters sound even with such a low crossover point. I’ll take some more measurements tomorrow and maybe try that fourth order on the tweeter and see how it goes. -Trevor

[trevordj]

I just wanted to update this. I had a blast designing this test speaker and I think I have it dialed in to the point where it sounds fantastic. Don't tell TVRGeek that I am crossing the XT25 at 1300Hz because this speaker will sound horrible crossed anywhere below 3500Hz :D.

I ended up tweaking the crossover a bunch. Sim, build, tweak, measure, repeat. I ended up building 7 prototypes and have massaged it to this:

[rpb]

My favorite part of this hobby is the x-over. You make choices, and listen to the results. Then later, you try something slightly different. Along the way, your tastes in tonal balance may change, or you find a gem of a CD that you've had forever, that just really shines on a new design. I try to keep x-over notes, but they usually are just the layout, and how much I like it. It seems each design does something better than the previous, but sometimes is lacking in some other way. It soon becomes a toss up, then I swap out a driver, and continue. On top of that, if you swap rooms, or position in the room, your favorite can change. It becomes addictive!

[trevordj]

My favorite part of this hobby is the x-over. You make choices, and listen to the results. Then later, you try something slightly different. Along the way, your tastes in tonal balance may change, or you find a gem of a CD that you've had forever, that just really shines on a new design. I try to keep x-over notes, but they usually are just the layout, and how much I like it. It seems each design does something better than the previous, but sometimes is lacking in some other way. It soon becomes a toss up, then I swap out a driver, and continue. On top of that, if you swap rooms, or position in the room, your favorite can change. It becomes addictive!

Totally, I am still very new to this but I very much enjoy it for the same reasons. I will say my favorite part of the process remains the 3D design. I love drawing 3D models of enclosures, its so satisfying. Just as you are saying often I re-design over and over again, every iteration a little better than the last. With the crossover though there is immediate gratification. It helps that I have a huge cache of crossover parts now thanks to a timely purchase in the classifieds. For example, in the last couple days I decided I wanted to tame a peak at about 4300 Hz, I had all the parts to toss together at notch filter and see how it sounded. This changed the required resistor value on the tweeter padding so I was able to swap out the series resistor and it sounded great!

Tonight I may try to add an additional parallel resistor after the crossover on the tweeter but before the notch filter and see if I can get a downward sloping high end. It looks like about 3 ohms will do the trick. Its so satisfying... try it, if it works great, if not on to the next one.

[rpb]

Try a notch on the tweeter centered at about 11k. Use a small coil to create a wider notch. Add enough resistance to limit the notch to about 3dB.

[trevordj]

I didn’t have any coils small enough so I ordered a few last week that I can try the broad notch at 11k and see how it sounds. I tried a couple parallel resistor values in parallel after the crossover and before the notch and it really sounds good. I haven’t measured it yet, I have just been listening for the last week and it sounds great. I will take some additional measurements after I get the smaller coils and compare a couple more times.

I have finished the construction for the Diversion project and I am probably going to wrap them up after this last go with the crossovers. I have to put together the subs and then paint everything and then I have another couple secret speakers I have been working on that will go in my new business. I will have to take a break with the speakers for a bit after that as I need to build some doors, a desk, and a small cabinet for my business.

Here is what the completed Diversion cabinet looks like.



These won’t be living in my home theater, I am putting them in my 7 year old son’s room and he is pumped. The stand bases are filled with sand for added weight and stability and there is a speakon connector coming out the back of the bottom of the stand for a nice clean wiring setup.

[rpb]

He's a lucky kid. Hope he catches the DIY bug too!

[trevordj]

I added a notch at 11k and I didn't like the way it sounded, it kinda took the life out of the speaker. I am almost set on version 7 of the crossover posted above but I am waiting for parts for one more version that looks promising. I didn't have all the parts for it, but I actually don't mind waiting for them as it gives me time to listen to the current version. I still really like crossover 7.