Faux Pas

[DrewsBrews]

The day has come!

Assembled crossovers for a couple days. Had to make a PE run yesterday for a couple resistors that I only had 2 of but I needed 4 total. My new DATS came in handy measuring all the parts. Turns out two of my 33uF electrolytic caps were 39uF while the other two were 34uF. I opted to save them for future use and picked up some 68uF caps instead.for C5 in the above schematic (measured 69uF).

MY phone had 0 clue what to focus on..







Measured both to make sure I didn't wire any one driver backwards. They don't match up perfect, but one set of drivers has more break-in from all the measuring and listening. Initially everything sounded super wide....In fact no imaging... and a rather muted bass. Hmm. Flipped the cables on one speaker, the image and bass appeared. DOH! somehow manged to wire one entirely backwards. I opened up one of them and it seems wired correctly. Will check the other later.

[DrewsBrews]

Circuit

[DrewsBrews]

Changed my mic position to better show response at my listening position. Farther away, so more reflections creeping in, but thought it might show a bit more what the woofer is doing since it is less off-axis vertically. Also the cancellations around the crossover frequencies change.

Unfortunately my house is small and I just don't think I have the space to get a reliable gated response. I check the impulse response and it is a blur..

I should also note, I changed the receiver/amp I'm using. Old one started getting crackling/ pops on the hdmi inputs.
Db dropped on the measurements due to more distance and getting to know the new amp.




Made more tweaks. After reducing the impedance swing on the tweeter it sounds less harsh, even was able to bump it back up a few db without hurting the ears.



Made some changes on both sides of the woofer-mid XO. After I started packing up I remembered to check distortion. Measurement is spotty above 2k, but I already know the stuff up high is fine. I dunno' if it is a fluke or dumb luck, but the distortion peak dropped out. Woo!



Need to button this up and move on to the inDIYana theme build.

[DrewsBrews]

Took some digging, but sounds like a "conjugate filter" is basically a zobel circuit but on a tweeter instead.
I don't know that it would help in this case. The impedance swing on the tweeter is primarily from the high-pass circuit itself. I doubt putting the circuit before the high-pass would be advisable due to it's parallel orientation. Unless I misunderstood what you were saying?

Current tweeter circuit : 2.56uF, .2mH, 1uf with 3ohm (series) before it and 20ohm (parallel) after. The mid and tweeter crossover electrically around 5k where the impedance wiggles.

[Edit]
'Played around in WinPCD some more. I might have some alternate options for the tweeter circuit that should be more reasonable on the impedance. Just need more time to do real testing.

While I'm at it I might try filling the top end of the enclosure with more foam and see if that gets me anywhere with cutting down the TL upper bass null. Have my doubts, but I've got plenty of foam.​

[DrewsBrews]

A few things to consider:

1. Consider a conjugate filter to smooth out the top end of the impedance curve and bring it down to a value between 4 and 8 ohms. This would make it a better load for class D amps that are not load-invariant.

3. The null in the upper bass response caused by the transmission line effect can be reduced or eliminated entirely by changing the vent's entrance inside the box (which in turn would require curving the vent). I have an Excel workbook on my website that can help to determine the best location for the vent to address it.

Thanks for the input!

I'm still not up on all the circuit options. Is there a circuit that would normalise the impedance the amp sees without affecting the response of what is already there? Something like a zobel?

Unfortunately the port is what it is by this point. It is two precision port flares mated back to back for roughly 5" port length. Maybe I should have just gone rear port so I'd be able to locate it farther up the enclosure.

[Brian Steele]

A few things to consider:

1. Consider a conjugate filter to smooth out the top end of the impedance curve and bring it down to a value between 4 and 8 ohms. This would make it a better load for class D amps that are not load-invariant.

3. The null in the upper bass response caused by the transmission line effect can be reduced or eliminated entirely by changing the vent's entrance inside the box (which in turn would require curving the vent). I have an Excel workbook on my website that can help to determine the best location for the vent to address it.

[Steve Lee]

After months and months of effort and copious time-off from adjusting the XO's you will finally arrive at a passable destination but you will always wonder if you could make it better sounding and then the weather will change and the humidity as well and then its off to the races and parts bin again . . .

Keep chasing that dragon.

:D

[DrewsBrews]

Another living room takeover today. Spent about 5hrs tweaking, measuring and listening (to just one speaker, like before). About 30 sweeps. The woofer response, in the enclosure at ear height, begins dropping off past 500hz so I wasn't able to do much for raising that XO point. Once I came to terms with that, I moved on to doing more tweaks for voicing. Normally I seem to enjoy a general flat (unshaped) response, but in this case I decided the flat response comes off quite bright. I wonder if timing has something to do with it, as the tweeter signal would likely arrive first due to closest physical distance.

I did a bunch of tweaking with the resistors, adding a parallel resistor to the tweeter to help out. I managed to get a slight downward slope out of it. Male voices like Cash and Chris Jones getting bonus points on this one (Roadhouses and automobiles was quite soul haunting) Which at least told me I was going in a positive direction. Female voices being so-so. Seeming simultaneously too much and too little at times. Theorizing the higher octave voices may be reaching for the mid-tweeter cancelation zones. But I may have been suffering listening fatigue before I got the tweeter under better control. Finally I eyeballed roughly 30deg to get a quick off-axis measurement (green).



Then I hooked up my new DATS to check if I threw the impedance out the window. Doesn't go below woofer impedance and doesn't shoot for the moon. So that is something, I guess.

[PWR RYD]

Thanks for the tips.

[DrewsBrews]

You made those ports look SO much better! I'm going to copy what you did with them. One thing I have done with my Precision Ports in the past is add 4 more holes for twice the mounting screws. Four just didn't look or feel like enough.

More screws would definitely give a beefy industrial vibe and match with 8x screw hole drivers. Good idea!

For the sanding...I just did it "lap" style to keep it nice n flat: face down on a whole sheet of sand paper (the grippy back stuff was pretty helpful). I started with a higher grit and quickly found that would take hours, even just being plastic. So went back to 80grit, then 150 and finished with 220.

[PWR RYD]

You made those ports look SO much better! I'm going to copy what you did with them. One thing I have done with my Precision Ports in the past is add 4 more holes for twice the mounting screws. Four just didn't look or feel like enough.

[DrewsBrews]

Looks like 95.7db @850hz. So pretty much -38db at that point.

I know I'm stretching the capabilities of $20 drivers. I'm not surprised they aren't perfect. I'll stew on it a bit. Maybe I'll muster up enough effort to try bumping up the XO point a bit.

[Wolf]

I'm seeing 57dB for the 850Hz HD peak, what output was the fundamental at? 95dB?
if you are -25dB, that is the minimum attenuation for distortion modes while -40dB is considered standard allowance. -50dB is where it is then completely inaudible.

Wolf

[DrewsBrews]

Finding some time in the holiday whirlwind...

I still wasn't convinced this project was worthy of the PA-310. So I tried some of the dirt cheap MCM 12s in 4ohm flavor and took over the living room this afternoon. Think I may be near ready to finish them up.

I was able to go back to 2nd order on woofer-mid XO. Total component count of 10+4 resistors overall.



Green=On axis, Blue=~30deg off axis. Big woofer dips I suspect are room interactions/floor bounce... Or because they are pieces of crap. I'll stick with room interactions for now to make me feel better

Next up, distortion



Distortion from the mid shoots up quite a bit down low, showing it really aught to be crossed above 1k to live its best life. But I suppose it is still within reasonable levels? That tweeter is certainly holding its end of the bargain!

And finally the resulting impedance...



Not exactly stellar but I've seen some pretty funky curves that were deemed acceptable so I really don't know what I'm talking about.

Anything I should look into fixing before getting out the soldering iron?

[Steve Lee]

Glad you had this little epiphany and decided to go for a sanity check with the PA310 driver, Drew.

Interested in your further findings as I am working a PA310 into a cabinet vol of 1.3 cu ft net right now . . .