Faux Pas

[DrewsBrews]

The tweet & mid phase look pretty "average".
The woofer phase looks pretty rough (esp. between about 3.5 and 7kHz).
You could try to regenerate phase for the woofer and see if it helps ... ?

Ok I found I can do that in response modeler, Just trying to figure out what to do in there.
Do I need to actually fill anything in or just use the "auto extract phase" from frd file option?.. I tried that, but has been sitting at 1% for 10 minutes .. I think I got the latest PCD, but I probably need to see if there is an updated version of the other tools (have Response Modeler 3.01 currently).

[PWR RYD]

I use "FRD Response Blender and Minimum Phase Extractor" written by Charlie Laub and Jeff Bagby back in 2013. It is an Excel spreadsheet similar to Jeff's programs. I like it because it can blend near field measurements with far field measurements, add upper and lower tails, and then extract minimum phase. Response Modeler and PCD stopped working for me years ago so I switch to their Blender and WinPCD and never looked back.

[DrewsBrews]

I use "FRD Response Blender and Minimum Phase Extractor" written by Charlie Laub and Jeff Bagby back in 2013. It is an Excel spreadsheet similar to Jeff's programs. I like it because it can blend near field measurements with far field measurements, add upper and lower tails, and then extract minimum phase. Response Modeler and PCD stopped working for me years ago so I switch to their Blender and WinPCD and never looked back.

I saw folks mentioning the blender when searching around. Am I required to do all the nearfield measuring and blending in order to get minimum phase? I don't much care about getting the LF data in there right now. Just trying to get where I can play around in PCD without it being totally inaccurate.

[wogg]

I'm moving toward VirtuixCAD these days, away from WinPCD and Blender. All the tools are in there, just have to figure out how to use them

[PWR RYD]

I saw folks mentioning the blender when searching around. Am I required to do all the nearfield measuring and blending in order to get minimum phase? I don't much care about getting the LF data in there right now. Just trying to get where I can play around in PCD without it being totally inaccurate.

No, you don't have to do nearfield measurements or blending. It will extract minimum phase from your .frd file.

[DrewsBrews]

Ok, thanks guys. The woofer minimum phase cleaned up some, but it was still pretty dirty in the HF region. It changed the blending of woofer to mid a bit. Though, doesn't seem to affect the mid to tweeter region since the woofer is probably 50db or so down by that point anyway.

I messed around some more and I think this is the best I'm going to get with this tweeter. Acoustically XO @ ~760hz and ~5700hz. That suckout at tweeter xo looks bad but I suspect after real world measurement and typical smoothing applied it would not look so bad? I mean HF seems to always jump around wildly anyway on raw measurements. I doubt Ill get much better due to acoustic centers being over 5" away on the baffle and the mid center calculating a full 7" back from the tweeter center. Nature of the beast it seems. Perfection was never an option really. I just have to have this fascination with old technology for some reason.


From here I want to shove the enclosure up near a wall and see what the room response looks like. I have some mild baffle step on the woofer low pass and want to get an idea if that might be reasonable enough to run with.

Then maybe play around with tweeter diaphragm options.

[DrewsBrews]

Tested the woofer with enclosure about an inch from the wall.



There is a lower mid-bass hole that I would guess is the baffle step off until the frequency is low enough to start using the wall. whatever the cause.. it is there. I had some more time so I tried dialing in more baffle step and dropping the rest down to match. Probably about 5db down in total by this point.

[DrewsBrews]

Getting the factory p-audio diaphragms by themselves would have been more expensive than just buying the tweeters again. So I have spares now. I did try one of the 4ohm klipsch titanium diaphragms to make sure I wasn't missing out on anything. They are super close in response to the paudio diaphragms. (klipsch blue, paudio green). 4ohm gives a db boost, but also seems to get a bit more of a distortion boost than you gain in db output (not much though, so rather moot). I decided to stick with the paudio diaphragms in the new tweeters.



I then re-measured all the drivers now that the tweeters would have changed. This time I put the enclosure in a more open spot in my living room to try and reduce reflections. Which did introduce some changes. I dropped the woofer down to a 2nd order, and the low pass on the mid I had to bump up to a 3rd order. I wanted to try with a little boost on the high end. I can easily drop it back down 1-2db by dropping the resistor across the tweeter 1-2ohm.





That calculated suckout between mid and tweeter is unnerving, but I imagine it will be very mic/ear level dependent and come out different in real world tests.. I saw an article on stereophile of the Forte iii. They measured them and posted the response. Seeing that gives me some confidence i'm not totally missing the mark.

And I think I am at the point of diminishing returns with putting time into the XO software. Push one thing into line and it pushes something else out of line. I need to sift through my xo parts stock and put in an order for the rest.

[DrewsBrews]

Made a PE run for XO parts over lunch. The wife just headed out on a road trip with some family... Perfect time for a living room takeover

First go with the values straight out of the latest sim. I left the measurements with no gate for now to get an idea of room response.





Tweeter response was a little more tapered as frequency went down than the sim lead me to believe. I swapped out a resistor and then a cap to boost and flatten it a bit. Then I checked off-axis. I folded a sheet of paper in 3rds at a corner to give me a rough ~30degree angle to sight down. Violet is the on-axis response after the resistor and cap change. Green is 30deg off axis from there.



So the interference is pretty much what the sim said, except for the woofer-mid area (700hz). Swapping leads on the mid just shifts it higher up and gets rid of the nice boost the woofer gives the bottom end of the mid to flatten it out

Resistor may go back to what it was originally to drop the tweeter level back down a bit, but I might change the cap to an even larger value to see if I can flatten it more. Not sure what I will try for the mid-woofer region yet.

That's enough excitement for tonight though.

[djg]

How do they sound? What are you driving them with?

[DrewsBrews]

Running a Pioneer vsx-521 I got new ~12yr ago.

I'm only running one speaker at the moment to get the measurements. I don't have enough test leads to duplicate the setup.I obviously miss out on all the details the stereo effect would bring, but the lack of all that seems to highlight the response shortcomings. I could hear the dip in the lower midrange as everything sounded thin without the lower resonances in voices, guitar, etc. I couldn't outright tell the hole in the mid-tweet region, just something didn't sound quite right.

I focused on the upper XO first. It was odd.. at times the tweeter would be slightly subdued and no bad peaks on the measurement but still sounded a bit bright. Other times I got the dips to shrink a bit and looked nicer on the measurement but sounded dull. Eventually landed on a combo that seems to work.

For the lower XO I found the original (non minimum phase adjusted) woofer .frd showed the 700hz hole in the simulation. But, it still didn't correlate to real world measurements when I modified it from there. So I just relied on the simulation to show the filter curve differences, and tested out combos to see what would reduce the cancelation in reality. I had to go back to a 3rd order on the woofer, but I think I've found the best I'm going to get.

At some point I switched over to gated response to play around with that. So ignore below ~250hz on these.





Then there is the boominess from the bass hump (visible in previous posts). Sometimes it works ok for a recording, other times it is a bit annoying. I mic'd the port and it seems to peak around 70-80hz. I'm a bit confused. It would make sense to me that the 2nd impedance peak in the woofer would be the Fb point. But I read somewhere when researching (don't remember where) it is the midpoint between the two impedance peaks? I'm wondering if there is some bad standing waves in the enclosure, or if the resulting enclosure volume is way less than I had estimated and the port/box is indeed tuned to 70-80hz (instead of ~50hz I was shooting for). I wonder if it might flatten a bit when the second speaker is in-play due to no longer being point source?

(Port response)

[DrewsBrews]

Empty enclosure volume with no bracing calculates to ~2.26cf. I guesstimated 2cf final volume. 50hz in 2cf with a 4" port calculates a port length of 3.85" which seemed within the ballpark of a 4" long port with a single small flare. A real result a little above or below would have been fine to me.

But I throw a 75hz tune into the calculator and try different enclosure volumes. I find a little under 1cf would be the volume.

I measured with the mic flush with the port opening as close to parallel with the port as I could eyeball. Probably wasn't perfectly centered if that makes a difference. Not quite sure what is going on. Unless the foam is dropping the apparent volume by that much. My best guess is there is some other resonance that is taking over.

I might experiment with port extensions to see what effect that has.

[Chris Roemer]

MY WinISD (using T/S parms from the newark site) shows a 2.26cf box w/a 4"id x 4" long port tunes to 46Hz.
YOUR (inbox) z-plot shows a tuning (Fb) of 44? Hz. By definition, the Fb IS the "valley" between the 2 peaks.
This should correlate to a bottom-end rolloff (F3) in the mid to upper 40s - just about all the bass you'd need (for music).

Also, (WinISD shows) this rather short port tube has the 1st (main) port resonance near 1.7kHz - NOT the "boom" you seem to be getting near 80Hz or so. It's just a coincidence that your upper z-peak is also right in that range. Maybe a room mode?

[DrewsBrews]

MY WinISD (using T/S parms from the newark site) shows a 2.26cf box w/a 4"id x 4" long port tunes to 46Hz.
YOUR (inbox) z-plot shows a tuning (Fb) of 44? Hz. By definition, the Fb IS the "valley" between the 2 peaks.
This should correlate to a bottom-end rolloff (F3) in the mid to upper 40s - just about all the bass you'd need (for music).

Also, (WinISD shows) this rather short port tube has the 1st (main) port resonance near 1.7kHz - NOT the "boom" you seem to be getting near 80Hz or so. It's just a coincidence that your upper z-peak is also right in that range. Maybe a room mode?

Thank you Chris! I didn't expect anyone to track down the specs and do all that. Though your input does help immensely in confirming and correcting my understanding of things.

I started off in WinISD before ever making sawdust and got the same results that you see. I was hoping the specs weren't too far off as I don't have a DATS yet.

From what I've gathered in the past: The closer the impedance peaks are in amplitude, the better matched the box is to the woofer. In this case the peaks seem pretty dang close. That would correlate to the WinISD suggested box specs being very close to my real box volume and tune. So, I suppose I can infer the published specs aren't too far off.

​I haven't had a problem with this from other builds. Though this is the first time having a woofer and port so low to the floor, so it could be finding a new mode. I imagine it must be pretty potent if it influences the nearfield port measurement so much. Entertaining the idea of taking the laptop and a little amp outside to get some rough measurements to compare to.. at a time when there isn't a dang train on the tracks 1/4mi away lol.

---

Last night I rolled up a piece of cereal box to create a ~7.5" long tube and shoved it in the existing port, which should drop the tune to about 39hz factoring in its slightly smaller diameter.

Blue is the first port measurement, Green is the "extended" port. Not sure if I had the volume up higher on the amp the first go than I normally have it set. Orange is room response before, Red is after extending the port. Dramatic boost to the bottom end but nothing else drops down.

[PWR RYD]

From what I've gathered in the past: The closer the impedance peaks are in amplitude, the better matched the box is to the woofer.

​

Where did this information come from? It may give the lowest possible anechoic F3 for a given woofer (in sims) but there are other important things to consider.

[DrewsBrews]

I really don't know. Most of what I learned was over the past 15+ years poking around the internet trying to piece together information into something that seemed to make sense. I only tend to get a build in every few years due to the cost and, well, life so a lot of that "knowledge" is probably pretty stale.

I still learn a ton from each build. And that's kinda how I like it.