Low impedance -- ideas to fix?

arc2v

Member
2007-05-09 7:15 pm
I will post plots later when I finally get them off my other computer.

I recently finished my prototype open-baffle center channel speaker. It ended up being a three way using Dayton RS 8" woofers, Neo3 PDR dipole tweeter, and 3" Tangband bamboo midranges.

Crossover points are 500Hz and 3.5kHz, 2nd and 4th order L-R acoustic slopes.

The ONLY thing I don't like about the finished design is two low points in the impedance plot at 1kHz and 8kHz. With the exception of crossover point peaks, the average is around 4 Ohms, which was the target.

Those two aforementioned dips go down to 2.5 and 2.2 Ohms, respectively. It is a pretty steep dip, it doesn't linger down there.

My first thought is panic and redesign the crossover so I don't cook my amp. But then I'm thinking because it's at higher frequencies, even a low impedance won't demand a lot of current (and it's the current and heat that kills the amps)

Any advice on this? Am I being too paranoid? Ideas to fix it? Would a Zobel on the woofers or mids help?

Thanks,
Anthony
 

arc2v

Member
2007-05-09 7:15 pm
Here's the impedance plot in question:
 

Attachments

  • impedance.jpg
    impedance.jpg
    56 KB · Views: 335

arc2v

Member
2007-05-09 7:15 pm
I've been studying this for a while today, the only thing I can think of is that I wired the Mid low-pass ahead of the Mid high-pass.

In a band-pass filter, does it matter what order you put the elements in? I thought the resonances being so far apart would mean it would not matter.

If it does make a difference, could I just switch them, or would I have to redesign since it would mess up both sides of the crossover?
 
Well, I'll be the first to admit, I'm over my head here, but I would be curious to see the frequency response graphs. I think that would give the Impedance graphs more perspective.

Also, the precise model numbers of the speakers would help.

At best I'll be an observer, but I'll probably learn a bunch.

steve/bluewizard
 

arc2v

Member
2007-05-09 7:15 pm
Sure thing, here's the REW measurement (non-gated, in room). 2m, on axis (blue), 30 degree off axis (red). BTW, 30 degrees puts this well outside the seats of my HT.

There's a room anomaly at 200 Hz. It shows up on measurements of my Magnepans as well.
 

Attachments

  • adht_2m.jpg
    adht_2m.jpg
    79.6 KB · Views: 261

arc2v

Member
2007-05-09 7:15 pm
I isolated this phenomenon to the tweeter crossover. The notch filter is doing the higher drop and the crossover itself is causing the lower one. I am attempting a redesign now to fix it.

My first cut got the lowest "theoretical" (PCD from the FRD tools) impedance at 3 Ohms. I believe this because my old circuit modeled spot on at 2 Ohms at the same frequency.

I'll keep you all posted.
 

Norcad

Member
2006-06-13 4:20 pm
Have you done this in Speaker Workshop?
Should be pretty easy to play along with the simulation, and look at the simulated impedance plot. Then you will find out why this is happening.
Most amps will survive an impedance dip down to 2 ohm, but at your plot you also have a big negative phase. Then the amp will stroggle more.
At about 750hz, you have 3ohm/-50 degrees. That shouldnt be the goal for an xo :)

Are you sure your measuring setup is 100% correct?
The SW setup and calibration can be hard to get right........
 

arc2v

Member
2007-05-09 7:15 pm
My measurement setup is non-ideal.

I can only measure in-room, with a pretty narrow gate. Plus I found out my test amp (MOSFET) shows a nice rising response in the top octave (as compared to the Outlaw amp). And I forgot to put my ECM8k correction file (also problems in the top octave).

so I was sort of chasing a ghost with that notch filter, since some of that rising response was amp+mic. However, I know some of it is real, since the Adire DDR design and other measurements of the dipole Neo3 show a large hump in the top two octaves.

I'm getting closer. The good news is that I think I'm at a reasonable baffle design and all subsequent problems can be fixed in the crossover.

Oh, and I finally figured out how to do an impedance measurement in SW. Stupid checkbox!!! I had been going back and forth in passive crossover designer (PCD) to get my estimated impedance measurements.

Thanks for the help all.
 

arc2v

Member
2007-05-09 7:15 pm
Okay, I made some headway in PCD and SW today.

First off, the Neo3 in dipole is a real bear to dial in. The notch filter that makes it work ends up dropping the impedance below 2 Ohms. That was for the 4th order crossover.

I did, however, design a crossover that was 2nd order for the low and high pass (500 and 3.5kHz). It's impedance drops only to 3 Ohms at two points and the impedance phase minimum is at 6 Ohms at 450Hz, otherwise, it looks very well behaved.

I had to use some resistors on the midrange crossover to help adjust the slope and keep the impedance up. Thankfully, I had efficiency to spare in the mid.

I'll build it tonight and see what I come up with.

I may end up ditching the Neo3 and go with something more traditional, like 2 dome tweeters instead. That might make for an easier crossover design with fewer problems.
 
I may end up ditching the Neo3 and go with something more traditional, like 2 dome tweeters instead. That might make for an easier crossover design with fewer problems

Better off finding a higher efficiency dome if you need it rather than 2 domes. Interference pattern would be very destructive if 2 domes are used per speaker causing all sorts of peak / cancellation issues on and off axis.
 

Norcad

Member
2006-06-13 4:20 pm
arc2v said:
Oh, and I finally figured out how to do an impedance measurement in SW. Stupid checkbox!!! I had been going back and forth in passive crossover designer (PCD) to get my estimated impedance measurements.

Thanks for the help all.


I'm still wandering if you actually measured the impedance in SW?
Do the impedance plot of each driver looks right?
Have you tested with an accurate resistor, to check your set-up and calibration?
And if so, do you use an amp when you measure impedance, or just the soundcard?

Im asking because I had trouble with this my self, and the problem was my soundcard. I had to use an amp on the imp measurements also, to get it right.

Good luck!:)
 

arc2v

Member
2007-05-09 7:15 pm
First, the two domes would be for dipole, so the interference is on purpose :)

Second, I use SW for all the measurements except the non-gated in-room (the one posted).

My setup is as follows:

Laptop --> M-Audio Mobile Pre sound card --> California Profile Amp (car sound w/ 12V,10A power supply) --> Jig.

Jig has voltage dividers built in to allow me up to 1.2V measurements at the lowest gain setting on the amp. It's similar to the Wallin jig, except it has no means to do microphone testing. For those, I plug into the MobilePre directly, using its phantom power.

The jig uses a 10 Ohm resistor (measured several times to be 9.98 Ohms with different meters). Sound card was calibrated independently of the jig. I also soldered in 5 Ohm and 20 Ohm resistors for recalibration when necessary.

For setup procedure, I follow the standard calibrations, set latency, and measure channel difference. For gates, I use a pulse measurement and try to balance the smooth response with the longest gate. My room is not ideal, but it's what I got. :(

Two big mistakes I believe I made:
One, I did not put my soundcard calibration file in. ECM8000 mic. People say it's flat, but mine is definitely one of the models that has an 8dB hump in the top octave. What measures flat, sounds dead, and what measures bright sounds flat. Some newer cal files for it show this exact phenomenon. I need to remeasure the drivers with this imported.

Two, I did not calibrate the amp. With this current iteration, I ran two 2m in-room tests. Both were in Room EQ Wizard. One used my SW setup with the Profile amp, the other was through my home system with the Outlaw amp. The results are telling: the profile amp adds significant brightness in the top two octaves (this was with the mic calibration file).

So I need a new testing amp, or at the very least, a full calibration for it.

Any insight you may add would definitely help. I know I'm close to a good design, but it's like the balloon animals, squeeze in one spot and the problem moves down the line.

Thanks,
Anthony
 

arc2v

Member
2007-05-09 7:15 pm
Also, within a few Hz, the impedance plots of the woofers and mids look to be on spec.

Ideal Dayton RS8's have an impedance peak at 27 Hz and a low of 7 Ohms.

My measurements show it at 30 Hz and the low (in parallel) of 3.4 Ohms (6.8 individual).

So bottom line: I trust my impedance measurements, I do not trust my acoustic measurements (completely).
 
I've already admitted this problem is over my skill level, but I have to ask, is 2 ohms really that bad?

From the discussion, this is functionally a 4 ohm speaker, is it that odd for impedance to drop that far for a 4 ohms system? I recall a commercial speaker system rated at 8 ohms with a side note that impedance could drop as low as 3.2 ohms.

I'm speculating that most Amp manufacturers know that rated or nominal impedance is something of an average; the speaker, in every case, will drop below the rated impedance. Further, this low impedance occurs twice, once at 1k and once at about 6khz for a very short span of frequencies. But I don't see any corresponding aberrations in the frequency response graphs. So, it may affect your amps, but it doesn't seem to affect your sound, and I think the affect on the amps would be tolerated unless you are abusing you amp in other ways.

As to the Zobel networks, they would certainly smooth out the peaks, but I don't think they would raise the 'dips'.

It will be interesting to find out what the final resolution to this problem is.

steve/bluewizard