Jims 3 way speaker build

[jimbones]

there is some uncertainty regarding the files I am using. Here are the zma and frd for the woofer and the mid. I use a 4 inch set back for the woofer and 2.25 for the mid. tweeter in following post

[jimbones]

here are the tweeter files

[jimbones]

DeZZar I made slight modifications as I was getting a dip in response between mid and tweeter. couple of value changes as well as driver polarity. please see.

[DeZZar]

In my sim I have the tweeter and midrange on the same axis as you said you took your measurement for each at the midpoint between them but I have your woofer as roughly 20 inches below this point. I've added the acoustic offset you advised - woofer is 100mm behind the tweeter and the midrange 57mm behind...

I'm getting this from my sim:



​​

[jimbones]

OK so for my frd measurements the mic was midway between tweeter and midrange, I dont recall but in earlier posts someone said based on that the mid was 2+ inch and the woofer was 10.5. Then in later posts rpb stated he said the mid is recessed roughly 2 inch behind tweeter and the woofer 4 inch. I assume he is talking purely z axis where the other dimensions were between mic and source?? If I should measure again let me know

[rpb]

I used about .75", and 3". I think the woofer might be closer to 4". These are numbers I made up, not measured.

[DeZZar]

Generally for gathering your raw measurements to run a sim you would measure each driver from precisely on axis out to 90 degrees horizontally (every 10 or 15 degrees). You wouldn't really measure at any mid points for this purpose as the simulation software can be used to define your final desired listening axis and it will adjust the responses relative to the listening axis and distance you define allowing a significant amount of flexibility in the simulation - but to this it is expecting specific on axis measurements per driver.

It is a little odd that your tweeter measurement sort of drops off a cliff after 10K which may or may not have something to do with the off-axis midpoint measurement.

[rpb]

There's many ways to do this xo. I hope my sims have offered ideas, if not accuracy. DeZZar, and others know the fine points of sims more than I. I just like tweaking the values. Any xo you try can be changed if the measurements show that it's not working as planned. For a while, it was hard to find coils, but now I see there;s plenty of them available in both high, and low DCR. How critical is the DCR to the bass extension? I put DCR estimates in my sim, but didn't look them up. I like the idea of crossing at 300hz since the woofer is positioned low, but I don't know how that sounds vs 400hz. I haven't kept any 3-way I assembled, so my experience is limited. On my 2-ways, in my room, I tend to like a level response. Others like a tilt. Some others like a depression around 3k. With a custom xo, you can have virtually anything you want, but you have to listen, and decide.


My revised sim. (Remember, my offsets were made up, so this may not be accurate. It was fun though.) I like to round off numbers. Most of the values are not critical, so there's no reason to try exact values. Use as much, or as little of my sims as you need.

[jimbones]

I think the three of us are circling around the solution and are very close so it is a matter of ME getting my crap together lol. I need to make sure my info is accurate and that we have all the info needed, I posted physical offsets below. I do not know how that relates to offsets done by the original method of measurements

[jimbones]

So here are exact measurements
Tweeter: X=Y=Z=0
Mid X=0, Y=5.125, Z=1.75
Woofer X=0, Y=25.625, Z=3.5

But that means all my measurements need to be done on axis of tweeter not mid way between the 2. Correct?
​

[DeZZar]

That's the offsets we can put in the model if you intend to listen directly on tweeter axis (which you usually would). But to achieve a good model each driver gets measured directly on its own axis as the starting point. That way you can fine tune the listening axis - you could make it midway between tweeter and mid if you wanted. Each driver should be measured at the exact same distance from baffle. It'll be important to try and at least get the speaker up onto a chair or something for that woofer measurement. If you are measuring from 3ft away you'll want the woofer to at least be 3ft from the floor. I know it might be difficult but it's just a necessary part of custom speaker design.

rpb makes a good point. In my experience no matter how good the simulation is it's never given the final crossover first time. It just gives a very good starting point and helps provide a point of reference to confirm results and understand what's going on. Multiple iterations are usually needed. But getting the measurements and simulation as accurate as possible means you won't be chasing your tail with unexpected results.

[jimbones]

Agree Dezzar. I didnt know that I am supposed to start by measuring each driver on axis. I will do that and save the files for when needed. This isnt my first project and I cant say how many revisions it takes to get right but its a lot lol. Im with both of you on that. Its important that when I do something I understand the importance as you have so clearly explained.

[jimbones]

Update: I build crossover in post 235 minus the 3.3 up cap in the woofer LP and the 7.15mh is actually a 8.3mh, the 5.1 ohm is actually a 4 ohm and the only other change is I am using a 2mh coil instead of a 2.2mh as that was what I had in my parts bin. Its an OK starting point. Polarity for now is WMT, + - +. Ill look at the other suggestions and as rpb says sneak up on it. Oh and one other thing. Only one of the cabinets is stuffed so I have to do the other to match. I had done it on one only as a test.

[jimbones]

so this is my system zma. Good thing is left and right match lol. So I have a couple of questions
1) from the curve it looks like the port tuning is very low. Is that correct? seems very low. Fbox is 30 hz sounds right and Fwoofer is 50hz. Does anyone see any issues? Note that these speakers will be used with a tube amp so I am glad at least that it doesn't go super low.

On another note I use the generator in dats and the strongest output is at 50hz. This causes that "one note bass". Is there any way to address it? Thanks

[jimbones]

ok so i followed suggestions to measure each individual driver 24 inches out on axis of each driver. Not so pretty.

[DeZZar]

Is this with or without crossover parts?

I don't know what's going on with OmniMic but you can't get a sub 200Hz measurement gated to 3.5ms....so for the woofer we should not be seeing anything below 200hz...

[johnny5jz]

There is a selection in Omnimic for a blended response which will show below 200hz. Your woofer response really stands out as not looking right.

[rpb]

I no very little about tube amps, but I'd use a tough solid state amp till you decide everything is like it should be.

Have you listened to the speakers as a pair yet with this x-over? If so, how is the tonal balance, and bass? What music? I have several CDs that I usually start with, and if things are going well, I listen to quite a few tracks.

[jimbones]

bass is prominent with a hump in the 40's. the sound is very clear but mot balanced so I need to get the levels of each driver correct. I made some changes will test tomorrow. You any I are on the same page lol, I borrowed a SS amp for test purposes. It has promise I think this is where I sneak....

[rpb]

Will your mic work with a computer sound card, and computer software? HOLM is free, and I believe much easier to see what's going on, because the measurements are smoother.

[jimbones]

I no very little about tube amps, but I'd use a tough solid state amp till you decide everything is like it should be.

Have you listened to the speakers as a pair yet with this x-over? If so, how is the tonal balance, and bass? What music? I have several CDs that I usually start with, and if things are going well, I listen to quite a few tracks.

I dont think it will but I will look it up. Also I was told not to use smoothing. I could make the curves smoother.

[rpb]

I dont think it will but I will look it up. Also I was told not to use smoothing. I could make the curves smoother.

Smoother is not the goal, but getting a measurement of the driver without the room changing the response is.

Can you elaborate on the needed tonal balance changes? Or describe the new tonal balance after yesterdays changes?

[jimbones]

Smoother is not the goal, but getting a measurement of the driver without the room changing the response is.

Can you elaborate on the needed tonal balance changes? Or describe the new tonal balance after yesterdays changes?

what I am concentrating on is I take individuall response of each driver to see if there is excessive overlap and if they are playing in the range I expect them to. For example the mid was playing flat up to 4k before it even started to roll off. That is not what I wanted with a XO point of 2800hz. I made some value changes and they look now like they are are correct not I am playing with the padding on the tweeter and mid to make sure it is not too forward, dull, bright etc. Ill post XO soon

[rpb]

I would do the same, once I thought the padding was close. When you disconnect drivers, do so at the source of the signal, not at the driver. If the woofer is disconnected, and you take a gated measurement of the mid, and tweeter, it should be an easy measurement. Fine tune the mid to tweeter x-over if needed.

A close mic on the mid will show you the roll-off. Same with the woofer. Measure the mid and woofer together from about 12" from a point half way between centers. Look for a deep null when the polarity of one driver is swapped. If you get a deep null, that means the drivers were in phase with the other polarity. Being close to the floor may screw this measurement up, but maybe not. You might want to plug the port for this measurement.

[jimbones]

so this is the XO I have in play now. I had another listener come over to hear it and it sounds real nice. I think I need to pause and listen for a bit and pause making changes until I get real familiar with the speakers. Then I can refine further. One note the tweeter polarity should be shown as in phase. For some reason in the SIM the results are better reversed but in reality it is better In Phase

[Wolf]

If they get nasal in male vocals, pop a 20-30 ohm across the tweeter and see if it goes away.
Wolf