Usher 2 way 8948/9950 Simulation

[Steve H]

Hello all,

Ok, this is my first simulation and just about my first everything. I appreciate all the work that goes into this forum and those that share their amazing knowledge.

I have the above mentioned drivers in hand. I also have a box that I have modified to accept them. The enclosures are .7 cu ft, 16.5 tall with 9 inch baffles ans 3/4 roundovers. Baffles are 1 1/8 inch mdf and the rest is 3/4. Tweeters are not offset and their centers are 3.75 inches from the top. Using WinISD for the first time as well and have it tuned to 42 hz with an f3 of 43hz using a 2 inch port of 5.3 inches. My room is 14 by 22 - the speaks will be about 1 foot from the rear wall and firing down the long length. Ceiling is 9 feet in the centre - cathedral.

I traced the zma and frd files from the usher website. I modeled for full BSC in RM and extracted min phase into PCD. In the hopes I can upload my graphs, I humbly seek some feedback as I don't really know if I'm on the right track. My knowledge is thin at this stage and I don't have a handle on how to interpret all the graphs - especially the phase! Thanks for looking, Steve.

[kmibb]

Shawn A's Arpeggio design used those two drivers in a 0.75 cu ft cabinet and were very well received. You might look that design up for reference.

I have no input to give on your particular simulation.

[kmibb]

Here is a design with similar drivers (8945A and 9950) by Dennis Murphy. Here he decides to address the bump at 900Hz. I see you did not do that. Not sure if it's audible or not.

http://murphyblaster.com/content.php?f=usher2way.html

Jkim also did not address the bump in his design, which I'm building
http://web.archive.org/web/200910281.../Usher_701.htm

[martyh]

It looks very good! You may want to toss a 18-20 ohm resistor parallel to the tweeter to help with power handling a bit. It shouldn't affect the high end much at all.

You've done well!

M.

[Pete Schumacher ®]

Here is a design with similar drivers (8945A and 9950) by Dennis Murphy. Here he decides to address the bump at 900Hz. I see you did not do that. Not sure if it's audible or not.

http://murphyblaster.com/content.php?f=usher2way.html

Jkim also did not address the bump in his design, which I'm building
http://web.archive.org/web/200910281.../Usher_701.htm

Now, this is a pure sim, and the actual response may not exhibit that bump. But if it's there, it is audible and should be addressed. It reveals itself with female vocals, notably Nora Jones.

[kmibb]

---

Now, this is a pure sim, and the actual response may not exhibit that bump. But if it's there, it is audible and should be addressed. It reveals itself with female vocals, notably Nora Jones.

Dennis said he never thought it would be audible when he did his ScanSpeak designs, where the woofers have the same hump, but that he's convinced after hearing them that it should be addressed.

[Steve H]

Thanks everyone, some questions,

1. How does the phase look? What am I looking for in the graphs?

2. I have reviewed many of the designs using these drivers - that bump was noted, but admittedly I kinda forgot about it as I was just trying to get a basic simulation to work. So, lots of female vocals running through our system Pete, what might I hear if this bump is audible.

3. Martyh - thanks for the suggestion - what are the benefits of addressing the power handling issue with the parrellel resistor?

4. Given that this is a pure sim, am I at a decent enough place to start building an xo?

Thanks, steep curve here for me, appreciate the guidance,

Steve

[Pete Schumacher ®]

Thanks everyone, some questions,

1. How does the phase look? What am I looking for in the graphs?

2. I have reviewed many of the designs using these drivers - that bump was noted, but admittedly I kinda forgot about it as I was just trying to get a basic simulation to work. So, lots of female vocals running through our system Pete, what might I hear if this bump is audible.

3. Martyh - thanks for the suggestion - what are the benefits of addressing the power handling issue with the parrellel resistor?

4. Given that this is a pure sim, am I at a decent enough place to start building an xo?

Thanks, steep curve here for me, appreciate the guidance,

Steve

Phase could be better. You'd like to see the woofer and tweeter phase align at the crossover point. for LR4 slopes.

In the case of female vocals, she'll go from sitting on the stage to jumping in your lap when she hits those notes that excite the frequency of the bump.

[mdocod]

Hi Steve,

I'm in the process of learning this too so I took the liberty to work through a simulation with your drivers and box and current x-over components.

I did the box simulation in hornresp get the following:

Xmax limited to~:


Resulting in:


That's without port flares, if you can get flares on both ends of that 2" port, you can get chuffing under control. (~18m/s peak IIRC).. Still a bit too high, but livable. I would personally tune a touch lower. 2" flared with lower tuning would move more of the potential chuffing issues down below most musical content. The response is a bit "hooked" on the bottom end currently, I'd rather have a smoother rolloff on the bottom end personally- also solved with slightly lower tuning.

-----

I simulated the baffle and listening position in Edge (I chose to simulate in the effects of listening off-axis, which I normally do as I don't like the look of toed-in-speakers).
Here's the simulation of the tweeters relative position to the mic, on your baffle in the location described. This is simulated at roughly 30 degrees off axis, which, is where I find myself more often than not when in front of speakers.


The baffle simulations take into account the driver Dd, an off axis listening position, the effects of the baffle diffraction, etc etc. I trace those with +6dB set to 0dB, and -6dB set to -12dB. Then combine those traces with the driver traces to get an idea of what the driver and the baffle and the listening position are all doing together. There are some components to this simulation missing though... This simulation assumes "free space," not including room reflections that will tend to prevent a full ~6db baffle step loss. With that in mind, leaving a couple few dB of "droop" through the transition below the ~900hz step peak in this case would undoubtedly be fine. Also, your round-overs should technically smooth some of the upper frequency baffle related peaks and valleys, so we can assume a smoother response than predicted up top, (in other-words, don't "futz" too much with components to solve little issues on paper, they are probably already solved).

-----

Unless I missed something, the x-over is currently being simmed with the following components:


I noticed that you have the 8848 recessed on the Z axis by about 1", having looked at the side-view diagram of the driver, I would estimate the origin to be recessed closer to 1.5-1.75" deep. I have simulated based on this estimate rather than yours.

The resulting response with all of the above taken into consideration (including where I have done things differently to try to produce a more accurate simulation):


I see room for improvement

I'll see what I can come up with.

Regards,
Eric

[mdocod]

Hi Steve,

This looks better to me...







I'm using the zobel to aggressively knock down the cone break-up area without effecting the phase relationship necessary to hit our x-over in-phase. There are other ways to deal with this, this is just one way to skin this cat.

In reviewing what I have posted, I am noticing that I have "high C1" used as a label for 2 caps. The 3rd component part of the high pass should be labelled "high C2." In this case, "high C2," is not really being used for it's effect on slope, but rather, it's effect on phase.

Regards,
Eric

[Steve H]

Thanks Pete - wherever I end up on a first trial crossover, I will listen for that bump. Every time I brought the phase closer in alignment, I got a broad mound in the region of the crossover, so I left it for comment. Also, I only got a 17 db reverse null and from what I read, not so great.

Eric - also learnin are ya - sheesh? Seems to be coming along quite nicely I would say - thanks very much for your contribution. I know there are supposed to be no dumb questions, but it appears I might have some.

I don't even know how to ask this question. I never thought about it, but I guess sitting between two speakers not toed in is off axis? I need to digest your response later at home, but is this the major diff in our two sims? Your sim looks great - can I replicate this in PCD?

Regarding free space assumption - the speakers will be 3 feet from side walls and are about 8 feet apart. I used 1 inch for z offset as I had read that was reasonable default for 6.5 and 1 inch dome.

Thanks again, Steve

[mdocod]

Hi Steve,

Yes, if the speakers aren't pointed directly at the listener, then they are being listened to off axis. I've never liked the look of furniture or speakers or objects in a room that were not positioned parallel to walls, except in corners where 45 degrees across the corner looks alright to me. In most places I have ever had speakers, I have had them parallel to the walls, and as such, there is no listening position anywhere in the room that is "on-axis" for both speakers.

Here's where some of my ignorance will come into play... I like the idea of simulating in the effects of the listening position, however, I do wonder what effect that has on the programs ability to simulate phase tracking, and whether it is "accurate" or not. My instinct tells me that the simulated phase should still work well, but I just don't know for certain. I can't run the spreadsheet simulators so I can only guess based on looking at your screen-shots, that you would want to use "horizontal panning" to simulate an off axis response.

The 2 major differences in our simulations, are likely the driver offsets, and the off-axis baffle/driver simulation.

The difference of 0.6" driver offset in our simulations, accounts for about a 30 degree phase shift at 2,000hz, so it's something to keep in mind. You may want to take a closer look at that driver and try to determine where you believe the "origin" is compared to the tweeter. The ~1" "rule of thumb" may not apply in the case of this fancier driver. Also, keep in mind that those various controls in PCD should be used to compensate for the effects on phase tracking introduced by listening height compared to your intended driver heights.


-----

As far as the baffle response goes, there are some interesting differences between the on and off axis response....

Here's on axis with the tweeter...



notice the +2.5dB peak at ~1500hz, and ~3dB null at ~3000hz (from +6dB)? Granted, your round-overs may combat that to some extent; The issue as I see it is that, when you simulate based on this listening position (a position you may never listen from), you may be correcting for that peak and null, only to be actually adding a different null and hump in response at your actual listening position. Those hefty peaks/nulls, BTW, are caused by having 3 baffle edges almost the exact same distance from the driver. The corresponding ripple in response through the remainder of the upper octaves is caused by the same thing, as the distance to the edge of the baffle becomes an in-phase dominant or out of phase dominant reflection. My understanding is that the round-over has the most useful impact on those "ripples" in response above the first few primary transitions.

-----


Here's a more realistic listening position... (this is the simulation I used, just shy of ~30 degrees off axis)



Being off axis has the effect of reducing the 1500hz peak by 2dB, and the first significant lull is shifted to ~4000hz, and thereafter, there is a pretty steady and continuous loss in output due to being off axis.

To put things in perspective, the hump at 900hz, characteristic to the driver that has been pointed out as noticeable and annoying by some. It's only showing up as about a 2dB hump when not corrected for on the design axis, though, the difference between simulating on axis, or off-axis 30 degrees, could cause several more 2-4dB "problem areas" depending on the listening position. I venture to guess, that some of the "900hz hump" issues are actually related to both the driver, and the listening position merging, causing the 900hz hump in response to "stand out" a lot more just below a very nulled response through the x-over caused by a simulation that does not reflect the listening position as well as it could have. For reference, see my first simulation trying to "recreate" your simulation. I suspect the hefty suckout above the 900hz hump would cause that hump to stand out like a sore thumb, even to relatively untrained ears.

Regards,
Eric

[craigk]

built the murphy blaster 2 way using the 8945, same 800 - 900 hump. listened with and w/o the lcr and it does make a difference. tame the hump ! if you can.