New Peerless/Usher Build - noob questions

[ch83575]

This will be my third speaker building project. The first two I consider complete failures. I was never satisfied with them and I could never discover exactly why. So this time I am going to go more slowly, and ask for help along the way. I am bound and determined to get a speaker that I enjoy listening to out of this project!

I am planning on posting detailed build info to my blog here: http://www.cmhaudio.com/ but I will try to copy and paste any pertinent info here when asking specific questions so you don't have to flip back and forth.

Here is what I have so far!



Here is my first question: Is my box too small?



This is the basic tuning info. The box is .5 cubic feet and tuned to 42hz with a 2" precision port. I am worried that the x-max is crossed with only about 4 watts (the plot goes from 1-32 watts) under 30hz. I originally wanted a smallish speaker (I didn't realize how large .5 ft^3 is even) but now I am thinking that a larger box and lower tuning might be able to keep the woofer from unloading in the audio band. The scale of the group delay is not shown in the graph, but each line is 5ms. So the GD peaks at 15ms just under 30hz, but is down below 10ms by 40hz. Let me know if you think this tuning will sound ok, or if I should go a different direction before doing much more.

Also, on a related note. The end of my precision port is only about 2" from the back of my tweeter. I was thinking I had about 3", but that was to the baffle and when I was attaching everything I realized that the tweeter is really close to the inside port opening. Is this going to be a problem?



This is the modeled vs. measured impedance of the unit. My thought was that if they were drastically off then I would blame the port restriction, but they look pretty close to me. Does that mean I am in the clear as far as the port is concerned or will it start to make a bigger difference as volume goes up (the impedance was done at a pretty low level)?

And one more picture just because I cant resist:


Thanks in advance!
-Chad

[JonMarsh]

There are many different philosophies about how to tune a bass reflex ported enclosure. I don't think a typical QB3 alignment is a good choice for several reasons-

First, if you design the bottom end response for flat response anechoichly down to Fb, with boundary reinforcement in most room positions, the subjective response will be heavy somewhere in the 45-65 Hz area. I find a tapered response tuned to a lower frequency, down about -6 dB, produces the most pleasing response in room. OTOH, I'm not into Hip Hop, Rap, or death metal, and folks that are like an amped up sound, I guess. If you want a heavy bottom end in room, a six inch two way won't be your route to nirvanna anyway.

Second, that kind of tuning profile will have a lower Q and better step response, as well as more extension.

For some ideas about that sort of design approach, this page in the Modula MT thread may be helpful.


Modula MT first post

System is tuned to 34 Hz in a 0.75 cu ft. enclosure. Unibox 3 was used to evaluate the alignment, maximum output for xmax, and port velocity. Similar designs and tunings have been done with the Usher 7" and Scanspeak 7" drivers.

[ch83575]

There are many different philosophies about how to tune a bass reflex ported enclosure. I don't think a typical QB3 alignment is a good choice for several reasons-

First, if you design the bottom end response for flat response anechoichly down to Fb, with boundary reinforcement in most room positions, the subjective response will be heavy somewhere in the 45-65 Hz area. I find a tapered response tuned to a lower frequency, down about -6 dB, produces the most pleasing response in room. OTOH, I'm not into Hip Hop, Rap, or death metal, and folks that are like an amped up sound, I guess. If you want a heavy bottom end in room, a six inch two way won't be your route to nirvanna anyway.

Second, that kind of tuning profile will have a lower Q and better step response, as well as more extension.

For some ideas about that sort of design approach, this page in the Modula MT thread may be helpful.


Modula MT first post

System is tuned to 34 Hz in a 0.75 cu ft. enclosure. Unibox 3 was used to evaluate the alignment, maximum output for xmax, and port velocity. Similar designs and tunings have been done with the Usher 7" and Scanspeak 7" drivers.

I don't currently have excel on any of my machines so I cant use Unibox. Is this more like the tuning you are describing:



This is 25 ltrs. & 35hz. The old was 14 ltrs. & 42hz.

And an overlay with the current box and tuning:



I definitely want to avoid a "boomy" bass tuning as my room is not the smoothest to begin with. I think I would definitely prefer a "thin but tight" speaker but, like I mentioned, I originally wanted to fit it in a reasonably small enclosure. I guess maybe you just cant have both. Does the new tuning look definitively better to you?

-Chad

[Dave Bullet]

What driver is it? - an HDS? T/S specs?

Vented will mean larger for a given driver so yes - small enclosure and low extension are difficult without equalisation (assuming xmax to spare).

[JonMarsh]

That's definitely the direction I would choose; it will cover a wider range of program material, including organ and 5 string bass, with good power handling down to about 30 Hz. It's always a trade-off between reducing the driver excursion and getting output at the very low end, versus the driver excursion in the 40-60 Hz area, which will now bit a bit more like a sealed box. Note that the amount of enclosure damping will affect the Q of the tuning; lined walls is usually my preference.

In the same vein, here's another example, for a three way concept I'm working on with dual 6.5" woofers on low range, (either HiVi D6.8 or Jantzen JA-6012) in a 40 liter cabinet.






Same concept, just a larger enclosure and reasonable extension to the upper 20's. Note, these drivers wouldn't normally be used in a conventional ported system at all, but tuning below the driver FS produces useful extension and some interesting characteristics. Minimum cone motion is in the upper 20's, but maximum port output is at 24-25 Hz.

The real point is, that ignoring room boundary lift, just the anechoic output capability per cabinet approaches 100 dB or more down to ~26 Hz, with 25 watts per driver. With room boundary lift, you can expect a practical 6 dB more in most cases. That's pushing it to the limit.

[ch83575]

What driver is it? - an HDS? T/S specs?

Vented will mean larger for a given driver so yes - small enclosure and low extension are difficult without equalisation (assuming xmax to spare).

It is a Peerless 830875 6.5" Nomex. I also have the same size in the poly version and I plan on comparing the two once I get the design to the point where a comparison would be at all informative. The TS as I measure them:




Thanks for the suggestions Jon. I think I will actually build a larger box and compare the two for myself since I already have the smaller one. That way I will actually hear the difference with my ears instead of just looking at graphs.

-Chad

[ch83575]

Project updated: http://www.cmhaudio.com/2009/05/peerless_usher-post2/

I have a few more questions:

I am working on different boxes for these, but I thought I would take a stab at measuring them and setting up a basic crossover last weekend. I know it will change when I move the woofer closer to the tweeter, but practice makes perfect right? Can you guys take a look and give me a few pointers.







The thing I am most worried about is the phase matching. An octave above or below the crossover point the drivers are already almost 20 deg out of phase. Is this normal for a design with no effort made to align the acoustic centers of the drivers? I know there is a lot of talk about using asymmetric slopes to rectify this problem, and I always though I understood the idea. Until I actually try it! I just cant seem to make either of the drivers any steeper than this (without more components, and then they are way steeper) and still hit the actual crossover point. I would really appreciate any help you guys could offer on this.

-Chad

[JonMarsh]

Yes, this is pretty normal- it's due to the acoustic center offset of the drivers. A pretty typical way to do this is by relaxing the woofer to something closer to a 3rd oder alignment target, and keeping the tweeter at 4th order. Then you have nutballs like me that like to do sort of weird quasi series/parallel crossovers, as in the Modula MT, that have a much more benign impedance load and helps manage the phase transition. I don't use Sound Easy; they key though, is paying attention to the vertical dispersion polar amp, or the polar response plot, and the reverse null shape and depth. (when you have the crossover phase tracking really well between drivers, reversing the tweeter will put a deep wide null in the crossover region). You seem to already understand the concept.

Like the difference between this:




And this:



Take a look at the Modula MT thread in the Missions accomplished section. Obviously, I'm not afraid of a couple of extra parts in a crossover if they serve good purpose. What does your impedance plot look like so far?

[ch83575]

Take a look at the Modula MT thread in the Missions accomplished section. Obviously, I'm not afraid of a couple of extra parts in a crossover if they serve good purpose. What does your impedance plot look like so far?

You caught me... I always forget to look at the impedance. Here it is:



It probably isn't great to have it drop to 2 ohm between 3-4k is it? This is where I am completely out of my depth... how do I fix it and still get similar FR results?

-Chad

[JonMarsh]

It probably isn't great to have it drop to 2 ohm between 3-4k is it? This is where I am completely out of my depth... how do I fix it and still get similar FR results?

-Chad

Yup! Compare that with this....





And you see why I don't use standard parallel networks, especially for two ways like this, which usually need to work OK with more modest equipment, not my usual Ayre or Aragon amplifiers.

Even the NatalieP MTM is relatively easy for most gear to drive.



Now that I'm done showing off, :W, I'd suggest you go back to your optimizer, and set a minimum Z target, say, 4 ohms, and re-run, because your component impedances are too low; this is giving you the shape you want, but too low a net impedance. That's what I'd do in LspCAD with a parallel network. I'm presuming Sound Easy has that capability; LspCAD does.

[Curt C]

It probably isn't great to have it drop to 2 ohm between 3-4k is it? This is where I am completely out of my depth... how do I fix it and still get similar FR results?

-Chad

Try adding another pole to the tweeter network, making it a 3rd order electrical rather than 2nd. Then you can increase the value of the shunt inductor and re-optimize.

C

[ch83575]

Thanks Curt and Jon! With the extra cap on the tweeter I was able to get the impedance up to just under 4 ohm. I have an old Linn amp myself, it may not be as ballsy as an Ayre, but it was designed to drive old Linn speakers which ate current for breakfast. But I would still like to make the design as friendly as possible... who knows where they will end up.

I have also been playing with a LR2 crossover. I resorted to a ladder delay after trying a few times without it. Check it out here:

http://www.cmhaudio.com/2009/05/peerless_usher-post3/

Again, I can't get the phase to track more than 1/2 to 2/3 octave above and below the crossover. My question is do you think I follow the targets far enough? I have noted the areas I am concerned about on one of the images. Please let me know if you think there will be audible consequences. Also, are there any drawbacks (other than parts count) to using the ladder that I don't know about?







Thanks for all the help!
-Chad

[JonMarsh]

Chad,

Doing a good design is balancing a lot of interrelated parameters. Now, you're making progress on some of the details, but I have to wonder if you aren't seeing the forest for the trees.

A few thoughts to keep in mind:

Choose your target response considering the capabilities of the drivers. This is very important in both two ways and three ways or more. You have NO hope of getting phase tracking in the driver response above or below the working range of the driver. So, an LR2 target is a difficult choice for a two way system; more practical for a three way, especially as then you're trying to limit component account.

Extract the minimum phase response of the driver and look at that before choosing the crossover points- it will give you some idea about where the driver can be used with success. For example, an Accuton C89 midrange goes out to 6 kHz, but because it dies abruptly there, the phase reponse leading up to it shows lag, and makes doing a crossover network harder. Other midwoofers and tweeters can show similar behavior, which in all cases is related to the amplitude response of the drivers.

Remember that the driver roll off will add to the network roll off, unless you put a zero in the function, such as a resistor in series with the inductor. I don't recommend that, for the tweeter; basically, you're trying to run it lower than it's designed for, and feeding it more power to shift back to just using it's natural roll off curve.

Have you looked at the crossover filter transfer function? I'll bet you'll get some clues to what you're doing there...

My advice for a two way like this is to reconsider using an LR-2 network with these drivers. I think you'd be better off with an LR-4; if you don't want as much peak group delay, then try a flat delay 4th order.

Here's a trick that might be helpful, too- usually it's the tweeter crossover that's more critical, because of power handling and response limitations. Work it out first, get your LF network in the right configuration and the the general response target, then lock the tweeter component values and optimize the woofer values for a flat total response. Experiment with subtle variations in exact high pass on the tweeter, then re-optimize the total; save each one, and compare.

[ch83575]

Jon,

Thanks for the great reply. I am going to take your advise, back up, and try to pick a better informed crossover point. But, I have already bumped into a few questions. You have been extremely helpful so far, hopefully you don't mind me pestering you a little more.

First, I plotted the min. phase of each driver and I am afraid I don't understand what I am supposed to be looking for. Taking the woofer first, am I to stay away from where the rotation speeds up around 4k? The tweeter seems well behaved to me, but again I am not exactly sure what I am looking for.






Now, the tricky part. I did some distortion tests to help pick a crossover point. I started out with a mic distance of 10" and the measurements came out ok, but with lots of small reflections and the like. When I measured the tweeter the 2nd and 3rd harmonic looked good, but when I switched on the rest of the harmonics to make sure they were reasonably low the 6+ shot up way past the 3rd around 2k. This worried me, so I moved the mic into a nearfield position and remeasured without adjusting the power. The mic/pre did not clip, but the increased SPL sent the distortion numbers up so much it really got me worried about the validity of the measurements taken with my DIY mic / and pre. How much of my measurements are mic/pre distortion and how much is driver distortion? Can I even trust the relative patterns, or is the whole shooting match suspect? If the graphs are useful which is more-so, the nearfield or the 10"? These were not taken at ear splitting SPLs or anything, relatively low level (maybe 1w).









All of these measurements can be seen in my blog post in a nifty comparison script if you so desire.

So, as if all that were not enough. Armed with all of this I would love to hear everybody's opinion about what would be a good crossover point for these drivers and if there are any pitfalls that I should be watching out for.

Thanks for all the help!
-Chad

[JonMarsh]

Will be getting back to you on this, have started an answer, but I'm tied up for a while this evening!

[ch83575]

Well, after looking at the graphs for awhile I decided on 1800Hz LR4. I think this is the best compromise between a target that is easy to reach and will attenuate the two problems I see in the HD plots. The woofer looks like it has a 3rd harmonic peak around 1600 and the tweeter has the weird upper harmonic peak at 2k (probably just barely audible, but still there). 1800 should provide at least 4.5 - 5 db of attenuation of the fundamental at these frequencies, but I am not sure how much that affects the harmonics. Also, the actual crossover point drifted a little north (perhaps to 1900) when I relaxed the woofer slope to align the phases... is that going to cause any problems?





Also, I would like to be able to look at the vertical dispersion pattern, but when I measured the drivers I did both drivers without moving the mic. So in soundeasy, in order to get the phases accurate I have to leave the drivers coincident in the system layout tab. Is there any other way I can look at the vertical lobing? Or am I stuck re-measuring the drivers each on their own axises in order to do that?

Thanks,
-Chad