Help with 1st design

[ch83575]

Ok... I just got what looks to me to be a descent crossover design for my first speaker project. This being my first time I would really apreciate it if the more experienced people here could take a look at my design to see if I have done anything stupid before I order crossover parts. Here is what I have:

This is a simple 2 way using the SEAS 27TBFC/G tweeter and the Aurum Cantus AC-130MK2 5.5" mid/woofer. Here are the in box frequency responces:



I am fairly sure that I properly added the baffle step into the woofer responce (which acounts for the drooping bass responce). Also shown in the graph is the 2000hz LR4 targets.

Here is the crossover:



And here are some graphs:












And here is a spl with the LR4 targets overlayed:



Ok... here are my concerns:

* I could make the driver responces hit the targets perfectly, but then the phase tracking was horrible. So my crossover point ended up more like -7.5dB instead of -6. Is this ok?

* I am not sure that I accounted for BSC properly or not... does it look ok to you? Or is there something else I needed to do?

* The main breakup node is hit directly by the notch, but the secondary nodes are only 27dB down... is this enough?

Does anybody else have any other questions or coincerns? I would really apreciate a critique before ordering the parts. One more question: would you recomend ordering multiple parts for tweeking the crossover? If so, which ones are the most important for tweeking by ear?


Thanks a lot in advance!
Chad

[augerpro]

I'm too new to give any advice about the XO, but I'll be using this woofer soon in a bookshelf so I'm curious what you think of it? Did you measure the T/S and/or do any HD plots? I'd be curious to see you're data...

[Curmudgeon]

My impression is that it's a very impressive beginning. I'd suggest plugging in the inductor resistances, and seeing if any further modifications are needed as a result.

As for parts that you might want to tweak as you do the final voicing, R7 and R8 seem like first choice. I've had good luck with Mills resistors. C3 should be as high a quality as you can afford. For future reference, you might want to put the driver polarity markings on the schematic.

Your questions are good ones, but my feeling is that you will have to answer them yourself, by listening. You might disconnect the tweeter branch and listen for roughness in the upper range of the woofer. Without the tweeter to mask, it's a reasonably sensitive test for uglies out of band.

Some items I'd comment on a bit; a rule of thumb for xo component breakin is 40 hours. (ouch). I don't really find it worth evaluating changes before breakin. This also means using a few recordings with which you are very familiar. Don't try a quick and dirty breadboard, and especially, don't use clip leads. If you build one channel first, you'll need to build a Y adapter (if you don't have a mono switch on your equipment) as you need both channels combined to evaluate the sound. And the speaker needs to be in a decent position for evaluation.... but you knew that anyway

Good luck, it looks very promising.

[ch83575]

augerpro: I am not really in a position to comment on the quality of this woofer yet. Perhapse after a few other projects I will be able to have some other experiences to compare it to (I have some Peerless 6.5 poly woofers on the way now). As for distortion graphs, i dont have any yet. I am thinking of taking some when I get my new amp for measurements (my previous amp could only produce a few watts and those werent exactly "clean" so the FR graphs will be fine because soundeasy basicaly filters out any contribution from the amp, but with distortion measuremtents at even moderate levels I think I would have been getting confusing measurements). Here is my measured data (note that some of the slots have have been left in the default positions ie. I do not know if the maginet is alnico or not (i kinda doubt it?)):



Curmudgeon:
Thanks for the encuragement... i hadent even realized yet that I should go through and see if these are component values that I can buy! I should probably create a crossover that can be built dont you think? A project for this weekend.

Thanks and keep it comming ,
Chad

[infinia]

Hi
BSC looks good to me. It's flat down to 100Hz, not drooping as shown in the driver alone. Might want to order some extra resistors to adjust the tweeter pad for taste i.e. in final listening environment.

One main concern.
The mid-woofer response looks very wide band before filtering to be sure the final response will be the target 4th order at the crossover region, but its probably OK further out. (because the on axis breakup is obscuring the woofers real roll off?) The polar plot seems to show it's fine tho. If it was me I would add another pole to the woofer's filter just by looking at your pre crossover plot (ignoring the polar plot for now) without having access to other information.

My only other concern is that the inductor's loss terms are not present (place holders only) or are they just realistically too low. This will effect your design in a few ways, one example being the Qts term of the woofer will be effectively higher than your model, as well as the filters response will be altered somewhat. Not to mention the mid- woofer level will be a bit lower than you think.
I don't think these minor points should stop your progress to your ordering the values in your 1st crossover design at this time. That being said I have not really spent the time to critique this properly so all my off handed thoughts might be taken with a grain of sand.

[Curmudgeon]

Originally posted by ch83575
[BHere is my measured data (note that some of the slots have have been left in the default positions ie. I do not know if the maginet is alnico or not (i kinda doubt it?)): Yes, Not Alnico.

Curmudgeon:
Thanks for the encuragement... i hadent even realized yet that I should go through and see if these are component values that I can buy! I should probably create a crossover that can be built dont you think? A project for this weekend.

Thanks and keep it comming ,
Chad [/B]

Yes, using purchasable component values is definitely something worth considering. FWIW I use Michael Percy.

Another thought; I may have missed it, but what is the Z axis relationship between the tweeter and bass/mid? Are the calculations based on the vertical alignment of the acoustic centers? Or offset as will occur if both are mounted on a flat, vertical baffle? Is the vertical offset taken into account also?

[infinia]

Hi
One more point.
Think about deleting L10 & L13 as you model some more.
IMO they are not really necessary in the larger scheme of things.
Tweeter peaking (breakup) cannot be heard that high.
Don't decouple the woofers Low Pass cap at HF, not good right.

Left handed complement... I don't think this is your first design is it?

[lndm]

Originally posted by ch83575
my first speaker project.

Your results are impressive.

I could make the driver responces hit the targets perfectly, but then the phase tracking was horrible. So my crossover point ended up more like -7.5dB instead of -6. Is this ok?

I think so. It's always hard to get phase and frequency spot on at the same time. Your compromise is reasonable.

Are you using auto optimisation, or doing it by hand?

The main breakup node is hit directly by the notch, but the secondary nodes are only 27dB down... is this enough?.....Does anybody else have any other questions or concerns?

The following statements are minor concerns. I feel your design is already good, so feel free to ignore this section.

27dB is a fair way down, barely worth worrying about but as a future tweak, I would look at it, and also the phase of the woofer at the notch frequency which again, is not worth worrying about just now in my opinion.

Your notch is sharp. You appear to be using the notch to achieve your slope, and that's OK. If you were to ease up on it and increase you slope, you could possibly take out the upper resonances and smooth the phase at the same time.

would you recomend ordering multiple parts for tweeking the crossover?

It is very handy to have parts lying about. They don't need to be expensive parts; I have found small inductors and electrolytic capacitors have helped me get on track while tweaking.

If so, which ones are the most important for tweeking by ear?

With regards to the woofer, naturally the inductor. Having another inductor (or pair) of two thirds of the value will allow you to try a little larger and a little smaller (as you'll probably want to do regardless).

For tweaking the notch filter, have resistors handy. Due to the tuning and hence the capacitor and inductor needing to be changed simultaneously, simply changing the damping via the resistor will permit you to hear the effect of the filter.

With regards to the tweeter, I would pay more attention to C3, L4 and R11. BTW, I agree with Infinia on the tweeter notch filter. Acceptable as it is, I wouldn't do it that way myself.

[ch83575]

Thanks for all of the compliments (left handed or otherwise )! It seems to me that people are mainly coincerned with the notch filter in the HP and LP circuits. The HP notch was not very high (5500hZ) and was used to flaten the exagerated diffraction peak there before the crossover components took over. This peak has been plagueing me since the start of this project... I have another post somewhere asking how to handle it and most people seem to recomend ignoring it, but in my measuremets (not shown) it seems to be just as large up to 30 degrees off-axis! I am thinking about building a IB to test my drivers on to rule out any strange problems with the drivers being the sourse of this bump. As for the LP I was bothered by the harshness of the main cone breakup of the woofer in initial listening tests conducted with the soundeasy crossover emulator (when I had an amp... right now I dont, but I should again soon). The notch was not really my idea... I just copied how Zaph handles the breakup of the L18 in his all metal design . I see now that perhapse my driver has too much energy above its breakup for this to be a good choice.

I tried again this morning without either notch, and I got much closer than in my original attempts (not ever posted because they sucked!), but the polar plot is a deal breaker in my book!

















So now my inexperience gets the best of me... to me the first design (with the notches) looks better, but dont really know for sure. Is the peak at 5500 going to be a problem? How do I correct the polar plot?

Oh... by the way, the drivers are centered in the baffle and all of the graphs do take into account their relative positions and are taken at 1m.

Thanks for the help!
Chad

[infinia]

I like your idea about measuring your midbass concentrating at the upper end (don't need IB for HF). Try some off axis at 30 deg to get to get a handle on the drivers natural roll off response and use that to incorporate into your crossover. If you move your crossover point higher for this then your crossover might get simpler.
I think if you used loss terms ie DCR on your inductors then your notches would tend to go away.