Reality Check

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
How concerned would you be with the following impedance graph?

An externally hosted image should be here but it was not working when we last tested it.

I'm not sure if this impedance would work well enough or if it would make an amp work too hard.
An externally hosted image should be here but it was not working when we last tested it.

I'm also not quite happy with the phase below 2K. But everything I try messes up the higher frequencies. So I'm afraid this might be the "best case scenario"

What do you think? Make it or break it and start over?
 
Hi,

It looks like the woofer is 8 R nominal. Is the tweeter 4R nominal because if it is this may be as good as your going to get. Although it looks a bit low even for a 4R tweeter.

If they are both 8R then Ithink you have a problem as the impedance is too low.

Do you have the impedance plots of the drivers on their own?

Regards,
Andrew
 
If I'm reading the charts correctly, it is a 3 Ohm load. Not really awful, not great either, but might be closer to many commercial speakers than you think. Does this 3 way design use multiple drivers in each part of the 3 way spectrum?

I also seem to see 95 db SPL, convention would assume that is with 1 watt of drive. So it won't necessarily take a monster amp since your design seems to be about 6 dB more efficient than the commercial average of closer to 89dB/watt

There is far less power in the 750Hz to 5KHz spectrum in music then there is at lower frequencies.

Any good grade solid state amp should not have a problem unless you want concert hall volume. Tube amps might be tougher, be sure to find one with a 4 Ohm tap on the OT.

Steeper slopes (especially on the woofer low pass and tweeter high pass) would help reduce the area where more than one driver is active, of course that does mean more complex crossovers and more money for xover parts. Low order crossovers are quite popular, I sometimes wonder if that's because without software help, the math was so complex that better sounding solutions were found with low order xover designs.
 
I went with 2 midranges in series to help things out, but the main problem was with the highpass, not the bandpass. I was going with a 2nd order filter to match phase with the midrange and used 2 notch filters to supress it below 2K. The midrange and the woofer both have wicked spikes at their breakup points so I'm using a 3rd order filter on the woofer and a 2nd order with a notch filter on the bandpass. As you can imagine, all these notch filters wreaked havok with the impedance.

Oddly enough, I added a notch filter on the woofers - that's how I'm getting such smooth low end response. And since there's a resistor in series with that huge coil, I can go with a cheap, high R coil and adjust the resistor accordingly. If you saw the response without it, I'm sure you would agree the expense is justified. If I'm in a silly mood, I might just add a second lpad or a button to bypass it for a 6db "bass boost".:rofl:

A single TB 25-1372SC tweeter (6 R)
2 Dayton RS52AN-8s in series
4 Dayton SD315-88s in series parallel isobaric chambers. (coils in series)

I'm redesigning it with a 3rd order highpass and an unusual bandpass - not sure what you would call it.
An externally hosted image should be here but it was not working when we last tested it.

I'm getting better impedance, but the phase matching is not as good; and the response is not as smooth. I think a few more days on it and I might have something worth looking at.

BTW, I summed up the impedance and averaged it out. It came up with 5.16 ohms. The new design is getting something closer to 6 atm. It still needs work.

One of the advantages of being a noob is not knowing what not to try. I've never heard of putting a coil in the lpads groundpath, but that certainly seems to help keep the tweeters response curve flat. Has anyone else done that before?
 
Hmmm ......

I'm not sure what is going on here but isobaric 12"ers crossing over
to 2" mid domes is never going to be a good idea in my book though
it is just about doable.

Twin 2" mid domes chucks away the major advantageof a middome
in the first place, which is dispersion, due to vertical lobing issues.

Isobaric bass drivers are fairly pointless when you can buy better
single drivers for twice as much with moreorless the same effective
parameters as the isobaric pair.

The crossover contains what I'd call "no-no's", take a look at :

http://www.zaphaudio.com/ZDT3.html

:)/sreten.
 
jdlech said:
I went with 2 midranges in series to help things out, but the main problem was with the highpass, not the bandpass. I was going with a 2nd order filter to match phase with the midrange and used 2 notch filters to suppress it below 2K. The midrange and the woofer both have wicked spikes at their breakup points so I'm using a 3rd order filter on the woofer and a 2nd order with a notch filter on the bandpass. As you can imagine, all these notch filters wreaked havok with the impedance.

Oddly enough, I added a notch filter on the woofers - that's how I'm getting such smooth low end response. And since there's a resistor in series with that huge coil, I can go with a cheap, high R coil and adjust the resistor accordingly. If you saw the response without it, I'm sure you would agree the expense is justified. If I'm in a silly mood, I might just add a second lpad or a button to bypass it for a 6db "bass boost".:rofl:<snip>
Wow a lot of territory to cover. First have you considered or modeled series notch traps instead of shunt? Put a parallel tank shunted by an R in series with the driver but behind any Zobel (to keep a good load on any high or low pass elements). Ditto with the tweeter L pad why not increase the series Rp1 allowing a larger shunt R and a higher total Z?

You describe second and third order designs in the text but the plots mostly show the ~6 dB/octave slopes of a first order design. The split high pass capacitors C3 (18uF) and C unlabeled (48uF) is unusual. Since this is one place where very expensive capacitors pay off, redesigning for only 1 C seems worth some effort,. My simulator shows both midrange shunt networks as having a substantial effect on impedance but little effect on shape. The mid notch seems to be tuned to about 20KHz(?).

Your crossover is complex enough that trying to comment or improve all of it at once by way of the forum is very hard. Maybe we could focus on a specific aspect?

Are you using a crossover design tool? Does it have an optimizer? Do your phase plots include phase errors from the drivers themselves or just the phase of the crossover?
 
It's hard to comment on your crossover, but looking at the polarity of drivers, it should be a LR2 design.

I'd not use the RS52A with 12" subwoofers in a LR2 design (not even in a LR4 design). Simply not a good match due to expected high distortions around the xover point. If you want to use large woofers with this mid, the RS270's are the best bet. The xover point should be as high as 500 Hz LR4 or 700 Hz LR2.
 
I recently switched over to speaker workshop as a simulator and soon to be testing software.

I used spltrace from the FRD consortium to trace both impedance and response off the mfg. datasheets for all drivers. It'll have to do until I put my jig together and figure out how to do a few tests.
The band pass notch filter tames that wicked 16K spike.


Yes, SW has an optimizer, but I don't know how to use it yet. It also has a "perturb" feature, but I'm doing it manually.

I'm starting over. I'm getting better impedance by cascading the tweeter. I'm still getting better results using two midrange drivers. I should be able to finish sunday.

How about an MTM configuration? Would that help vertical lobing?

Edit: I know what you mean, the huge coils are already out of the picture.
 
If you use the manufacturers reponse plot then you should add baffle step diffraction and the effect of the box on the bass. Both can be simulated and each can be added either before SPLtrace using a drawing app, or after using a text editor.

FWIW, you should find this compares surprisingly well with what you measure.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.