Measured the port length, according to that and unibox it should be tuned to ~36hz. Not sure why I tuned it so low in the first place or why it would up being 31hz. Will try removing some of the bracing and remeasure.
Why does the port (gold trace) output peak at ~31hz?
My recollection is that the dip in the NF cone output (purple, 35Hz) is the most accurate measure of tuning. The NF port (gold, 31Hz) is always slightly off from the true tuning frequency. I read this in Joe D's "Testing Loudspeakers" a while back. Of course, I could be remembering this backwards. Will double check when I get home.
Update: I dug out my copy of "Testing Loudspeakers" & looked up D'Appolito's comments about the best methods to double-check port tuning. (TL, page 69).
Here is my summary:
A) NF woofer cone notch: Very accurate measure of vented system tuning frequency .B) Impedance curve zero phase crossing: Very accurate also. C) Impedance curve minimum: Close to the above approaches, but not right on. D) NF port response peak: Least accurate method, peaks much more broadly.
So, don't look at the NF port peaking to determine system tuning. Look at the NF woofer cone notch or the impedance curve zero phase crossing point. Hope this helps. Bill
It looks like this build is going to take a large inductor. Not wanting to roll such a large one, I've been browsing PE's catalog. Iron cores around $12, air cores ~$40, and these odd donut shaped Jantzen ~$50.
Browsing eblay for bulk fishing hooks I came across 5pcs 4mH 9A Common Mode line filter Inductor 25mmx15mmx13mm.
Same concept? Will these work in place of a regular 4mh iron core? Not much data on their page.
I would bet their definition of mH is different from yours judging by the size. If this a second order you might see if you can trade some mH for uF...
It might be 4mH, with a large ferrite ring for the core. Likely would saturate early. If I use the chokes from surplus or eblay, usually they go in tweeter filters for better handling.
Wondering what happens when combining two smaller coils into a larger one. Anyhow, playing around with dats and placing the inductors into different positions in relation to each other. Coils are about 3"wide, 1½" tall
Individual coils: 1.9571/.38708 & 1.9734/.38349
Stacked on top of each other: 4.9358/.91474 reverse polarity: 2.8587/.95427
Flat on face side touching side: 4.0842/.92576 reverse polarity: 3.71/.87505
separated by 1/2": 4.038/.88294 reverse polarity: 3.7688/.89164
1": 3.9399/.86775 reverse polarity: 3.8199/.89079
2": 3.9061/.87559 reverse polarity: 3.8522/.88664
3": 3.8908/.88041 reverse polarity: 3.8656/.89294
Side touching side @90* with one coil standing up: 3.8726/.87902 reverse polarity: 3.8766/.87357
Just thought this was interesting. You can see a trend with both coils laying flat on their face in relation to separation and even at 3" there is still some influence. I assume adding more power will increase this mutual influence considerably?
This also shows how well placing the coils at 90* with one on its side works.
Anyhow, I can start proto'ing this crossover w/o having to wind a rather large inductor ... yet.
John, been using both a 2nd w/tank? and 3rd. At this point I'm wondering what the more gentle roll-off does for the low-end since I messed up on that part. I hate to start tearing into the enclosure unless I'm sure there is no other viable option.
Have to get new measurements with the felt padding in place before going any further with the crossover. As it currently stands, I think the tweeter section will get a lot simpler.
Remeasuring with the tweeter slots filled w/felt reduced total crossover parts count to 11pcs. May be able to reduce that to 8pcs. if the tweeter rez does not have to be addressed and if those peaks/valleys between 500 and 1.5k are not real. Bit better phase match to boot. May have a little too much dip around 2k. Time to start proto'ing.
Ben, it was previously noted that you dislike using inductors made from small gauge wire. Consider a couple small mh coils that reside in the tweeter network that need followed by a resistor; why or why not wind these from small gauge wire (repurposed motor winding) and omit the resistor? It would reduce the parts count to 9 ... but resistors are probably the cheapest part of a crossover so no real $ savings.
I had one occurrence where I tried the difference between larger gauge and a resistor and a smaller gauge with no resistor. FWIW, the lower DCR setup sounded better, even though the net resistance was the same.
Load your measurement of the two drivers playing together as an overlay. Draw a schematic out just like you wired them when you measured them together, and adjust delay value on one of the drivers until the response overlaps. For best results, it's important to turn off smoothing on all plots.
Isn't their a way to show the measured summed response along with the predicted summed response so they can be more easily matched? I seem to recall doing this but that brain cell must have been one of the weaker one in the herd and died off.
Something about get file, under the curves menu....That's what I was describing in my first sentence above, and the faint green "file" line shown in my screenshot.
When adjusting the mod delay setting, remember that + values of mod delay mean that the driver that you are adjusting is delayed in time (or further away) than the other driver (the anchor, or unadjusted driver). Generally, in a two way speaker, the woofer voice coil is further away than the tweeter voice coil. Therefore, if you apply the mod delay to the woofer, it will be a positive value.
If, on the other hand, you adjust the tweeter mod delay instead of the woofer, it will generally be a negative value, because the tweeter voice coil is closer to the microphone than the woofer voice coil. Hope this helps.
Wound a few more inductors and proto'd a crossover. It was not what xsim predicted. Too much of a hump around 1k and phase match looked like poop. Somehow, my sum of both drivers was incorrect, which in turn gave an incorrect acoustic offset. Offset wound up being 1.68" instead of 1.25".
It looks like I can still use all the newly wound inductors but the large one grew in size. No biggie as I was using two smaller coils for 3.8mh because I did not want to use up a large portion of 15gauge wire winding such a large air-coil. That inductor will more than likely be a purchased iron-core unless you guys think there is a better, and economical, alternative.
Progress has now come to a standstill as 60+hr. work weeks really cut into my free time.
Anyway, thanks for the help everybody. It is appreciated.
Comments
InDIYana Event Website
Wolfy's Photobucket
My recollection is that the dip in the NF cone output (purple, 35Hz) is the most accurate measure of tuning. The NF port (gold, 31Hz) is always slightly off from the true tuning frequency. I read this in Joe D's "Testing Loudspeakers" a while back. Of course, I could be remembering this backwards. Will double check when I get home.
Update: I dug out my copy of "Testing Loudspeakers" & looked up D'Appolito's comments about the best methods to double-check port tuning. (TL, page 69).
Here is my summary:
A) NF woofer cone notch: Very accurate measure of vented system tuning frequency
.B) Impedance curve zero phase crossing: Very accurate also.
C) Impedance curve minimum: Close to the above approaches, but not right on.
D) NF port response peak: Least accurate method, peaks much more broadly.
So, don't look at the NF port peaking to determine system tuning. Look at the NF woofer cone notch or the impedance curve zero phase crossing point. Hope this helps. Bill
5pcs 4mH 9A Common Mode line filter Inductor 25mmx15mmx13mm.
InDIYana Event Website
Wolfy's Photobucket
InDIYana Event Website
Wolfy's Photobucket
I think a second order with a tank is usually a good place to start, typically if I'm using a tank there's no resistor for a damped second order.
If you are still playing around, use the hold function in XSim and see if the smaller inductor large cap gets you any benefit.
When adjusting the mod delay setting, remember that + values of mod delay mean that the driver that you are adjusting is delayed in time (or further away) than the other driver (the anchor, or unadjusted driver). Generally, in a two way speaker, the woofer voice coil is further away than the tweeter voice coil. Therefore, if you apply the mod delay to the woofer, it will be a positive value.
If, on the other hand, you adjust the tweeter mod delay instead of the woofer, it will generally be a negative value, because the tweeter voice coil is closer to the microphone than the woofer voice coil. Hope this helps.