[wushuliu]

To answer your questions, the speaker was on a table outside, with no walls within 6 meters, the mic at the same level as the speaker, at 1m away from the speaker.
I did not measure them both at once. I hooked up the woofer only, with no filter at all, used REW and umm6 mic to measure. Then, did the same with the tweeter only. Didn't change volume on computer or volume on amp. Saved those two frequency responses as txt files, imported them in vituixcad and that is what you are looking at.

Your images are still too small to see the db scale so I'm guessing what it's showing is the almost 10db difference between the woofer and tweeter. If you look at the original specs for both of them on PE website you'll see the tweeter goes as high as 96-97db, the woofer only 87 or so through most of its response.

When you measured did you keep the mic at the same spot for both woofer and tweeter? You want to make sure to do that. Don't move the mic height from one to the other. Usually for 2ways people keep the mic at a point between the tweeter and the woofer.

It's also easier to analyze if you just show one image of the woofer response and then one of the tweeter. It's really hard to see details in this images.

Hopefully you measured with the speaker at the very edge of the table. Otherwise it sounds like you did okay. It's a start!

[wushuliu]

Also you can do these measurements inside, as long as you have a general idea where they are going to be. Although it's more standard to do them outside with wide space and height, this isn't realistic for most people and is really just best practice if you want your design to be built by other people.

For a beginner who knows where the speakers are going to be I think it's just easier to do them inside as long as you keep the measurements at 1m or less. Even Kirby admits that even though, again, he only gives you part of the process (he doesn't tell you how to position the mic at all which is *important* as I mentioned above). He is doing his outside because he has to for the kits he sells.

[JLM]

Not uncommon for "matching" drivers to not match (measure the same), even high-end drivers.  Don't know how many manufacturers measure each driver they install (or to allow for 20 hours or so of break-in first before measuring). 

[wushuliu]

Not uncommon for "matching" drivers to not match (measure the same), even high-end drivers.  Don't know how many manufacturers measure each driver they install (or to allow for 20 hours or so of break-in first before measuring).

True, but in this case it's not the matching between two of the same drivers he's referring to, it's the disparity between the woofer spl and tweeter spl. They are about 8db apart. Which is fine, just means padding the tweeter down with some resistors. It looks particularly bad though in his measurement because of the baffle step drop which causes a massive peak where the sound waves start to wrap around the baffle around the 1.5k mark before plummeting. This is why I was pushing for using the baffle diffraction simulation tools in his software. Then he would have seen it coming.

[S Clark]

And it's worth noting that baffle step compensation usually means lowering woofer output another 3 db or so across most of it's range.  These puppies are going to need a lot of watts to get some volume. 
This is another example of why I always try to steer first time DIY'ers toward a kit.  Building speakers on your own requires a rudimentary knowledge and a pretty decent measuring system... and even then it doesn't always work well. 

[Tyson]

Re: baffle step, I always wonder why 2.5 way speakers aren't more popular - it's a nice elegant solution to the baffle step loss and doesn't cost you anything in efficiency from the midwoofer.

[wushuliu]

Re: baffle step, I always wonder why 2.5 way speakers aren't more popular - it's a nice elegant solution to the baffle step loss and doesn't cost you anything in efficiency from the midwoofer.

Agreed. But there are the trade-offs IIRC:
lower impedance (8ohm-ish builds are still more popular then 4ohm).
Parts cost increase: the .5way woofer usually needs a hefty inductor
Frequency response: possible lobing issues since one woofer is closer to the floor (more theoretical than audible probably)

Usually what happens is someone asks this question, we go yeah why not, then someone convinces that you might as well just build a 3 way. Sometimes the most convenient option is less appealing I guess.

Lastly, I think the 2.5way concept is relatively 'new'. I believe Jeff Bagby said he is the originator only some 20yrs ago.

[mick wolfe]

I could be wrong, but I believe my JSE Infinite Slope Model 1's were defined as 2.5 ways. This is pushing roughly 40 years ago. Great speaker that I thought bested the rave of that time period, the Vandersteen 2c.

[Florent]

I'm working hard on this ! Just bought some more crossover parts so i can try multiple arrangements and measure them, just found a store nearby that sells all i need and this is cheap stuff.
By wiring them in parallel or series i have plenty of room to experiment and it gives me the opportunity to learn a little more about how those values actually impact the frequency response. (and it's fun)
But I'm worried about metling the amp, so i go sometimes back to the computer and enter the values I'm about to try, so I'm sure about the minimum impedance.

I think i am getting closer to an acceptable result. But I was wondering, in REW, just how much smoothing is OK ?

Is a flat frequency response the best target or "house curve" should be preferred for a more "easy listening experience" ? Not that i actually am going to achieve one or the other with success but for knowledge, which would you choose ?


I will try to come back this week end with plenty of details. Pictures and measurements.

In the meantime, here is a picture of the last measurement i made. In my workshop, mic at 1 meter, speaker at the edge of the table, psychoacoustic smoothing, and the difference in db from the highest peak to the lowest dip is approx 10dB. So yes, improvements are yet to be made and that is also why I am asking about smoothing because unless it is Var or Psycho smoothing, it looks like a wild roller coaster.

Critics are welcome !

Sorry for the poor quality of the picture.

[Florent]

Here you can see the inside of the enclosure and the port.

[wushuliu]

I think i am getting closer to an acceptable result. But I was wondering, in REW, just how much smoothing is OK ?

Is a flat frequency response the best target or "house curve" should be preferred for a more "easy listening experience" ? Not that i actually am going to achieve one or the other with success but for knowledge, which would you choose ?

I will try to come back this week end with plenty of details. Pictures and measurements.

In the meantime, here is a picture of the last measurement i made. In my workshop, mic at 1 meter, speaker at the edge of the table, psychoacoustic smoothing, and the difference in db from the highest peak to the lowest dip is approx 10dB. So yes, improvements are yet to be made and that is also why I am asking about smoothing because unless it is Var or Psycho smoothing, it looks like a wild roller coaster.

What does it look like with 1/24 or 1/12 Octave smoothing?

[Florent]

What does it look like with 1/24 or 1/12 Octave smoothing?

Each horizontal line is 5dB.

I think I mislabled this measurement. It seems to be the previous version of the crossover shown earlier, there is a differnce...

1/12

1/24

[wushuliu]

Each horizontal line is 5dB.

What you are seeing is based on the speaker and mic placement. Below 800hz you are getting reflections from somewhere - the floor or a wall that's too close. That's why you have the dips at precise intervals from the woofer. The 1/24 is the more accurate and you can see the dips are really narrow, so not that bad. So you can either move the mic 6in-12in closer, raise the speaker off the table even higher, or move it away from any walls closer than 3ft/1m. You'll see the response change.

What's more important right now is the overall response, which looks really good! You seemed to have handled the tweeter well. Now you can start playing with tweeter level values depending on your taste. Some people like very flat, some people like a slight tilt down like yours. Up to you.

Can you show the current crossover?

[Florent]

What you are seeing is based on the speaker and mic placement. Below 800hz you are getting reflections from somewhere - the floor or a wall that's too close. That's why you have the dips at precise intervals from the woofer. The 1/24 is the more accurate and you can see the dips are really narrow, so not that bad. So you can either move the mic 6in-12in closer, raise the speaker off the table even higher, or move it away from any walls closer than 3ft/1m. You'll see the response change.

What's more important right now is the overall response, which looks really good! You seemed to have handled the tweeter well. Now you can start playing with tweeter level values depending on your taste. Some people like very flat, some people like a slight tilt down like yours. Up to you.

Can you show the current crossover?

[wushuliu]

Okay, that seems more in line with a proper 2-way crossover, good work!

1. Looks like you have the tweeter L-pad resistors (3.9 and 1ohm) before the 22uf. Never seen that before. L-pads are almost always placed last. So you might want to have them switch places. Ignore that. I'll rephrase that shortly.

Edit: You put an L-pad first for the tweeter crossover. I have never seen that done before. Typically L-pads are placed last. Typically what people do is have a resistor value first, then another resistor AFTER the cap and inductor to adjust tweeter levels. So try removing the 1ohm and instead add a resistor after the cap to get the desired result.

2. 1.8mh on a 4ohm woofer seems excessive and your crossover does look a bit boomy in the low frequencies. So you might want to lower that value and see how it sounds. Your simulation had too little BSC, now you might have too much.

Could you show the individual woofer/tweeter crossover results in REW?

[wushuliu]

I should also say, please make sure to listen to both speakers as you make all these changes. That way you get a feel for what each change does and whether you like what you hear.

[Florent]

About the lpad i Will try it.

For the 1.8mH, i tried 1.2, 1.6, 2, and 2.2mH
<1.8 there were big bumps in the 600~1200hz but less boomy
>1.8 no difference. It will probably sound boomy but i think i was at less than 2ft from the wall. On the other hand, it Will give me the oportunity to cut those peaks in the low end with dsp without sacrifying to much volume. Does that seem correct ?
But i will try ! I am far from done with all this :-p

[Florent]

I found that "equalizer apo" software on windows that let you import REW filters. Seems to work prety well

[Florent]

I improved my setup for taking measurements, did another round of tests by switching values one the caps, inductors and resistors,

I tried to lower the inductor's value but couldn't get the 500-1000Hz to stay flat so went back to 1.8mH
As for the Lpad, tried to put it before and after the cap, one resistor before, one after, but it didn't change anything by measurement or ear (wich are not really great in my case) So I kept the same values and set it after the cap, as suggested.
I also changed the value on the cap, looks better.
After "noodleing around" I measured the drivers separately and was kinda surprised that they cross at only 1500Hz.

As you can see, I have a bump in the <100Hz, up to 8dB, I kinda like that so I'll leave it that way but if I wanted to damp that, what can I do ? Add a capacitor ? Lower the enclosure's volume ?
Each horizontal line is 5dB

Did some paint job, not satisfied, cheap spray can... I will sand it again and paint it with better quality paint.

What do you think ?

[wushuliu]

I improved my setup for taking measurements, did another round of tests by switching values one the caps, inductors and resistors,

I tried to lower the inductor's value but couldn't get the 500-1000Hz to stay flat so went back to 1.8mH
As for the Lpad, tried to put it before and after the cap, one resistor before, one after, but it didn't change anything by measurement or ear (wich are not really great in my case) So I kept the same values and set it after the cap, as suggested.
I also changed the value on the cap, looks better.
After "noodleing around" I measured the drivers separately and was kinda surprised that they cross at only 1500Hz.

As you can see, I have a bump in the <100Hz, up to 8dB, I kinda like that so I'll leave it that way but if I wanted to damp that, what can I do ? Add a capacitor ? Lower the enclosure's volume ?
Each horizontal line is 5dB

Did some paint job, not satisfied, cheap spray can... I will sand it again and paint it with better quality paint.

What do you think ?

Since the goal (IMO) is learning the basics, I think you are off to a solid start, especially considering how little information has been provided.

1. Below, say 300hz, the room will dominate your response measurements at 1m. So what you are seeing is I assume the effect of the speakers being closer to a wall/boundary. If you want to reduce that the easiest way is to stuff the port. You also want to make sure to have stuffing inside the speaker like polyfill or cotton batting or wool (I prefer wool). Adjusting all of that will impact the bass response.

2. I didn't realize you were crossing the tweeter that low. I don't think that tweeter can handle down that far. You never want to cross down to the tweeter's Fs. You want to cross no lower than double the Fs. In this case the Fs of the ND25 is 1350. So you need to be crossing no lower than 2.5k - which is what PE recommends in their specs. It's possible you may damage the tweeter if you cross that low at high volumes.

So again there are some really critical basic steps/info you are missing. I encourage you, again, to go to the PE Techtalk forum and get help there. They are really helpful to newbies and you will get a ton of responses and even crossover schematic recommendations. It's been interesting but I'm not going to offer any more help here because I think it's doing more harm than good.