Crossover redesign - how sensitive are tweeters at lower frequencies

My problem is this: I've been equalizing my crossover curve using an excel spreadsheet I devised (five-way and with different sensitivities on each driver, thus the need for all the magic) until I realized I wasn't taking into account that the tweeter would be less sensitive at lower frequencies.

Obviously this will very from tweeter to tweeter, but to what extent can I assume that the small amount (one or two milliamps) of current running through the tweeter will drive the tweeter with the same sensitivity as at higher frequencies/normal current?

If it matters, the tweeter in question is a ribbon, the HiVi RT1C-A. The electrical curve for the tweeter is otherwise very sharp, it just happens to "level off" for a bit around 1.5khz, and happens to fill in a hole in the mid-range at 1khz.

I would attach the bode plot, but I can't figure out how to export it from LTSPICE (haven't gotten my Gentoo box up and running yet, unfortunately, and regular SPICE crashes for me in windows).

Also, if you guys know any good passive all-pass filters, I'd appreciate it, because the phasing on my crossover sucks something sour.
 

Pano

Administrator
Paid Member
2004-10-07 6:05 am
Panama
Seems like a simple problem to me, if I understand correctly. The tweeter has a usable bandwidth, below and above that it will not give you the same SPL for a given voltage applied.
You have to know the tweeter's frequency response if you want to build the proper crossover. If you use it well within its bandwidth, then maybe just the sensitivity number would do.

Have a look at its FR plot
[IMGDEAD]http://www.swanspeaker.com/product/curve/rt1c-a-sm.jpg[/IMGDEAD]
 
The exact power level was 1.4V, 315mA through the tweeter at 1khz with 5.34 A being drawn from a 25V source and about 4A going through the other drivers (these are all peak ratings, and are total overkill, since my amp only goes up to like, 60 watts rms).

I should clarify (so you don't think I'm a total moron for using a 5-way crossover) that I'm renovating a speaker that was orignally 4-way with five drivers, so a 5-way crossover isn't really much of a stretch, and it's really more like 4.5 way because there's no strict cutoff on the lower tweeter, and there's only a 2khz difference in the crossover frequencies.
 

Pano

Administrator
Paid Member
2004-10-07 6:05 am
Panama
it just happens to "level off" for a bit around 1.5khz, and happens to fill in a hole in the mid-range at 1khz..
Looking again at the plot, I see what you are talking about. But I wouldn't want to use that part of its response, distortion is going to be very high. And it's going to be way, way done after you apply a high pass filter, right?
 
Looking again at the plot, I see what you are talking about. But I wouldn't want to use that part of its response, distortion is going to be very high. And it's going to be way, way done after you apply a high pass filter, right?

No, I was just an idiot. I was using a special high-pass filter that gave a kink at 1khz. Actually, should be simple enough incorporating it into my spreadsheet. It'd just require me to copy down numbers from every point on the chart, but I'd only have to do it once.
 
The funny thing is I was actually deriving the inductances by hand so I could model their impedance. I actually put a zobel on one of the tweeters because its inductance was so high. I don't know why I ignored the frc.

My high-pass filter doesn't have an order per se because it doesn't fall into the traditional crossover paradigm. I designed the crossover so the HIVI is sitting behind the other tweeter in series (as per the original design) with an additional capacitor in front of it and an inductor in parallel. The first tweeter has a second-order filter on it, so you might call it a third-order filter. But then I have a small capacitor coupling the hivi directly to the amplifier (or actually to a small resistor and than to a shelving low-pass filter) that boosts the high frequency response in the HIVI at the expense of the slave tweeter. Originally the goal was for an 18db/octave slope from somewhere around 4.5khz for the hivi, and something lower and shallower for the other tweeter, which is less sensitive, but eventually I realized I needed to bring high frequency response up and that the manufacturer recommended a 12db/octave curve from 3-3.5khz, so my average 6db/ octave curve from 4.5khz with a 17db/octave kink at the beginning was more than adequate and what's more flattened the frequency response curve. Now obviously I have to go back to the drawing board, but what I'm planning on doing will actually be quite similar, although the bode plot will look quite different. Will it sound good? I won't know until I try. But it looks like it will work on paper, and since Russian capacitors are cheap as dirt (and air core inductors are cheap as long as you keep them small) it doesn't really hurt me to try.
 

Pano

Administrator
Paid Member
2004-10-07 6:05 am
Panama
Originally the goal was for an 18db/octave slope from somewhere around 4.5khz for the hivi,
OK, it seems like you could hit that with 2.2uF in series and 0.6mH in parallel. That would be close to 3rd order at 4500Hz and would flatten the top end.
You'd lose about 3dB overall, but that's the trade-off for the flatter response.

It's something to try, anyway...
 
Ooops, sorry. I forgot that's a 5 ohm driver, right? The values above are wrong. :(

The other problem with that is that there have to be two tweeters or there will be a big gaping hole in the original cabinet (although maybe a drone...but no, too hard to design). The other tweeter is 3.5 ohms minimum impedance (3 ohm dc + I guess half an ohm from mechanical driving), with the rest coming from the inductance of the voice coil. The HiVi is actually more like 4.75 ohms if you look at Swan Speakers' plot. It turns out it's a lot more efficient to run these in series than in parallel if you want to keep the nominal impedance of the speaker around 8 ohms, since otherwise you'd have to put a ton of resistance in front of them.