Crossover - physical layout/construction help

[scottvalentin]

Haven't soldered the terminals, just clips in. but I will check those before testing again. This is frustrating!! I will try removing the resistor and just seeing what the coil only does. john h I didn't sand the coil leads but they had a lot of the copper coloured coating removed and the ends were all silver coloured.

[GTPlus]

Can you use your jig to check the coils? When I tried your circuit in Xsim, you can mimic your output by shorting the coil.

[jhollander]

Not likely, but your note about "clips in" reminded me about a bunch of people who had bad x-o testing results due to bad alligator clip to wire connections.

[Wolf]

Yeah- that looks like cap only to me as well.

Wolf

[scottvalentin]

Thanks Wolf and jhollander .



Ok, I did run this filter in Xsim (I like the free-form!). Here's what it comes up with:




Then, because I have some Sprite and Isetta parts lying around, I stuck a 0.9mH coil on the filter in place of the .35 and also took out the resistor. PCD shows a large spike at 1k or so, measured response again much different than would be predicted:

BLUE = raw tweeter
GREEN = just the caps and resistor in series
YELLOW = add the .35mH coil and 2R resistor to ground
RED = 0.9mH coil, no resistor



So, I don't know what to try now. I've measured continuity through all the points, resistors and coils I've measured, and although they are 0.2 to 0.4 above spec'd value, they aren't measuring out of whack.

Any more ideas? I was hoping for a much better outcome than to be guessing at parts after about 15 versions of the simulation!

Thanks

[Wolf]

Was the circuit all connected when you tested continuity? and you get continuity from the T junction to the far end of R1? As long as you do, then something else is amiss. Test it at both cap leads at the junction.
Try shorting R1 out. with a wire to see what that does.

Later,
Wolf

[scottvalentin]

Was the circuit all connected when you tested continuity? and you get continuity from the T junction to the far end of R1? As long as you do, then something else is amiss. Test it at both cap leads at the junction.
Try shorting R1 out. with a wire to see what that does.

Later,
Wolf

Thanks Wolf, I have now tested continuity between all points of the circuit AND from both sides of the T junction and it all is connected. I tested continuity at all the input leads and the tweeter leads and the resistance as well, and everything checks out.

I had a thought though, and that is with my original impedance measurement. I used the Behringer UCA222 and if you look at the tweeter impedence, you will see it rises starting below 80hz or so due to the crosstalk issue that I didn't know about when I bought it! I originally figured this wouldn't affect the frequencies way up above 1000 hz, but would this be wrong? I will do an fprawn trace of the SB impedence profile and redo the xsim models and PCD models with just the manufacturer Z-plot.

[rpb]

Is the coil wire twisted tightly around the cap lead, or is it just laying against it and soldered? Try connecting a jumper wire to make an additional contact between the cap, and coil. If the solder joint is good, nothing should change.

If that fails, I'd revise the sim to use a 3 ohm in parallel with the tweeter, and a 6 ohm after the x-over, then redo to cap, coil values. Try 8uf, .35mH, and 16uf as a starting point, and adjust the sim from there.

[scottvalentin]

I have re-simmed with a traced impedance plot for the SB26 and the results don't change in the sim.

rpb thanks I will try a jumper tomorrow morning and see what happens. The cap and coil have a good solder joint, but I'm out of guesses at this stage so I will give it a go.

I did sim your values but had the impedence drop a bit too far near the crossover region. I upped the 16uf cap to 30uf to match phase better, but your values were a good starting point.

[rpb]

I have re-simmed with a traced impedance plot for the SB26 and the results don't change in the sim.

rpb thanks I will try a jumper tomorrow morning and see what happens. The cap and coil have a good solder joint, but I'm out of guesses at this stage so I will give it a go.

I did sim your values but had the impedence drop a bit too far near the crossover region. I upped the 16uf cap to 30uf to match phase better, but your values were a good starting point.

Maybe add a 3 ohm padding resistor before the x-over, and reduce the size of the resistor that is after the x-over. The resistor that is in parallel with the tweeter may need to be increased slightly too, so that the other two resistors can be larger. The woofer x-over might be causing part of the impedance dip too.

[scottvalentin]

Good news!! I removed the 2 ohm resistor after the 0.35mH coil and this looks much closer to what I was aiming for. I didn't change anything else, didn't add the jumper suggested either. I am in the process of wiring it up with the woofer section - looks like I may need to adjust padding a bit, but looks much much better.

I'm still not sure what the main issue was. With the 2 ohm resistor after the coil, the network was giving a response as if the coil value was close to zero. Additionally, the predicted response of just the coil by itself without a resistor would show a peak which I didn't see with either the 0.9mH or the 0.35mH.

Thanks all for the help so far, I will update once I have one fully wired up and measured. Might also need to add a notch on the woofer, but will see.

[scottvalentin]

Success - fully wired and tested in and out of phase. Thanks all for the input! These are ungated, smoothed at 1/48th.

[scottvalentin]

Ok, I have both up and running now with the following response:


Both speakers measure very very close, so all in all, I am pretty happy with the result for my first from the ground design.


Here is the current state. I ended up veneering the facets and then using a black dye and finishing with Poly. I am in the process of doing up solid maple baffles. I will finish with just OSMO Poly-x, I ended up liking just the natural colour more than the grey/brown and wish I'd left it alone!

Now, the 5db dip through the upper mids (between 2k and 7k) seems to be part of the natural tweeter response. Outside of just some simple EQ, any ideas on how to address in the crossover?

Current crossover schematic:


As mentioned, I am pretty happy with how they sound, but would like snare drums to have more snap! Any suggestions appreciated!

thanks

[scottvalentin]

A bump for my previous question - what would be the best way to address the dip between 2k and 7?

Should I bring down the woofer and bit like go to 2.75 or 3 MH on the coil or reduce the tweeter padding more? If I do that, how would I roll off the top end a bit?

Any input appreciated thanks

[Wolf]

How high is the breakup magnitude on the woofer? in your previous plot it was pretty high, and likely audible.

To flatten the response, you could reduce the first coil on the woofer a smidge as well as maybe reduce the cap value a touch. This will bring up it's end of the xover area. On the tweeter, I usually try increasing the coil or cap, and then even try to change the Q of the filter by increase cap and decrease coil or vice versa to fiddle with the knee at rolloff. If you want to flatten the response, a CR or tweaked zobel across it can flatten out the rise in the top-end, and then reduce the padding resistor value to bring up the level and match the woofer. You'll likely have to fiddle with the phase alignment and values, but you'll get there.

Later,
Wolf