Data from my FIRST DIY

[TurboFC3S]

This whole measurement setup thing is pretty cool It's also very reassuring when the data collected matches what I'm hearing ...

Measured are my floorstanding 2.5 ways using GR Research M-165x drivers and GR T2 tweeters. Cabinet is 1.65 cu/ft vented, tuned to 34hz.

Today I changed the crossover design for the tweeter, to bring the level up a bit based on yesterdays measurements. It helped somewhat, but now I'm thinking the tweeter just might not be worth keeping - and I should switch to something more universally approved :T Although I'm sure I can get the response better using what I already have.

I haven't figured out a way to make nice looking schematics yet, so let me try and word it out.

Tweeter - 2 ohm resistor, then 3rd order electrical - 5.6uf, .39mh, 20uf giving approx 4th order LR slope at 1900hz (according to Passive Cross Designer)

Upper Mid - 2nd order electrical - .82mh, 20uf - then Zobel of 23.3uf and 4.99 ohm

.5 Woofer - Right now, just a 5.6mh Erse steel core

First off response curves of Left Speaker:

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

And the Right Speaker:

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

Now Nearfield Measure FR of Left side Tweeter:

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

Nearfield FR of Left side Midwoofer:

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

Nearfied FR of Left .5 Woofer:

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

Measured Port Response:

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

CONTINUED ...

 

[TurboFC3S]

My first thought is that these tweeters just can easily be beat at $30 each ... an that it's not worth putting all the notch filters and such in place, that I'd be better off with a 27TDFC, RS28a, Peerless 810921, etc. But once I use my Yamaha Receiver to equalize the tweeter levels, they do sound nice. So I think it CAN be a nice sounding tweeter. The question is, be happy with what I have + equalization - or change the tweeter to something that might sound better without equalization? Hmmm

Obviously I will still make some more crossover changes. I'm thinking change the tweeter resistor to 1 ohm ... and play with the modeling a bit more to see how I can bring up the top end level just a touch and bring down the 3-4K hz hump a bit. Maybe an RC circuit can do this.

Also the midwoofer definately needs to be crossed later. Modeling showed a 2000hz cross point, but that's apparently not what's happening in the real world!

And lastly, since I'm bi-amping anyway - I think I'm going to go active x-over on the .5 woofer. The impulse and group delay measured for it was HORRIBLE. I'm assuming the big inductor just introduces tons of delay, could that also be part of the reason for the big 500hz hump in port response? Or is that some sort of internal resonance point?

Anyway, this stuff is fun. Can you believe this is my first ever DIY project, and less than 2 months ago I know next to nothing about this stuff! It's because of great websites like this (and others), and having many helpful people that one can get into such a complicated thing so quickly and easily. Thanks to all!

 

[RJ]

The tweeter frequency response is rising in the area that causes it to be shrill or sibilant and then it falls. Try flattening the 3 - 8 K region and get the 8 - 20K to rise a bit by using an RC circuit just in front of the tweeter. Something like 2 ohm and a 1.5 uF cap to start with.

 

[infinia]

Hi
An impressive 1st design indeed. Look forward to the crossover schematic when you post it. Some observations.

Noticed on the final response it peaks around 500-1K and a dip centered at 3KHz.
You could probably ignore the dip at 3K for now (BBC dip). Possibly further minimized it if you end up padding the tweeters down.
But you might want to tweek some on the 700 Hz peak. Could you move the port to the rear of the enclosure as a first try? If it's not already, then if not you might add a shunt cap or heavy zobel on the 0.5 woofer.

 

[Geoff H]

That's pretty good for your first project.

Looking at the response curves, it appears the mid and tweeter are not meeting, resulting in the dip from 1.5 to 3 K. A bit more tweeking and your there.

Measurements are great for confirming audible problems, but at the end of the day, let your ears be the judge.

Well done.

Geoff

 

[TurboFC3S]

Hi
An impressive 1st design indeed. Look forward to the crossover schematic when you post it. Some observations.

Noticed on the final response it peaks around 500-1K and a dip centered at 3KHz.
You could probably ignore the dip at 3K for now (BBC dip). Possibly further minimized it if you end up padding the tweeters down.
But you might want to tweek some on the 700 Hz peak. Could you move the port to the rear of the enclosure as a first try? If it's not already, then if not you might add a shunt cap or heavy zobel on the 0.5 woofer.

Once I get it all done I'll post a full writeup with all the gory details. Since I'm still voicing things out, it would be premature now.

The broad depression from 1500-3000hz is 100% because of my getting off the midwoofer too early, look at the measurement for the midwoofer. It's starting it's LR4 slope at pretty much 1000hz ... what I find so interesting is how I modeled this all out RIGHT (or so I thought) by tracing the frequency responce, impedance, and phase graphs for both the woofers and tweeter and making .frd and .zma's out of the data. Then imported that into Frequency Response Combiner to model the baffle. And then used that data in Passive Crossover Designer ... and still my actual real world crossover point is 1000hz off!

I think I have a pretty good handle on how to fix it:

1) Get the midwoofer slope starting at the right spot, ~2000hz
2) Raise the overall tweeter level about 1db
3) Figure out what's up with the left side tweeter resonance at about 12khz - either it's a defective tweeter, or I wired something wrong. I'm thinking it's a defective tweet, some good CSD data should tell me.

I think once I get it all sorted out I'll have a nice HT setup, at a good price. I just can't imagine how I would have figured this out without a measurement setup, I had an idea of what was off from listening, but could have just as easily made a wrong adjustment without the data as a right one.

That's pretty good for your first project.

Looking at the response curves, it appears the mid and tweeter are not meeting, resulting in the dip from 1.5 to 3 K. A bit more tweeking and your there.

Measurements are great for confirming audible problems, but at the end of the day, let your ears be the judge.

Well done.

Geoff

Exactly, I just don't know how I could have done all I did to model things right, and still end up with the midwoofer x-over being so far off.

My ears tell me these GR M-165x woofers are better than Zaph's test data says. And by the way, it was one of my woofers that Zaph tested! Sure they may have some high F3 and F5 distortion numbers compared to others in the field, but they also have tons of excursion and a very linear XBL motor ... and my ears tell me a well behaved cone. They have tons of bass, enough for sans-sub use - and they seem to be able to handle more power then they're rated for. I'm bi-amping with a 150x5 amp, and haven't found their limit yet.

 

[infinia]

Hi
Some thoughts.
Seems in IMO shorter peaks are more audible than broad dips. As far as linear distortion is concerned. I would tend to try and resolve those first. In your reply the peak around 500-1K was not acknowledged for some reason?
Regarding this peak - In your port responce plot I noticed a secondary port resonance near this area. So thinking it may be a contributor, as I noted in my first reply. I assume your final response is a sum of the 3 near field measurements. So if the port is mounted to the rear it should not be as much as a concern here.
You may be closer than you think to a pretty good design, so maybe some far field measurements 500Hz - 20KHz taken from 1 to 2 meters away would be justified before you tweek the crossover circuit anymore.

 

[Geoff H]

TurboFC3S, have you tried listening to a pan flute across that dip?
Sorry. I can't stop laughing either.

Edit: Just a note on the power ratings. Some drivers are mechanically ok at power levels higher than the voice coil thermal ratings. I did it once. I smelt the coils about the same time the sound went horribly wrong, and I was 6 metres away.

Geoff.

 

[TurboFC3S]

Hi
Some thoughts.
Seems in IMO shorter peaks are more audible than broad dips. As far as linear distortion is concerned. I would tend to try and resolve those first. In your reply the peak around 500-1K was not acknowledged for some reason?
Regarding this peak - In your port responce plot I noticed a secondary port resonance near this area. So thinking it may be a contributor, as I noted in my first reply. I assume your final response is a sum of the 3 near field measurements. So if the port is mounted to the rear it should not be as much as a concern here.
You may be closer than you think to a pretty good design, so maybe some far field measurements 500Hz - 20KHz taken from 1 to 2 meters away would be justified before you tweek the crossover circuit anymore.

The Left and Right full FR measurements are 'far' field ... so they take port response into account. So the 250-450hz dip is real, but it could be floor cancelation. These measurements were taken indoors, with the speaker on a wood floor. To get a better idea of what is really going on below 500hz, I need to measure these outside elevated a bit. Assuming it's still there on those measurements I don't know how to fix it It could be a phase cancellation because of the huge delay from the 5.6mh inductor on the .5 woofer. In which case, going active on the .5 woofer would theoretically fix that.

In either case I'm worried more about the stuff where I have more confidence in the measurements, mainly where the actual crossing over is happening ... and the tweeter response.

This does make me wish I had tried a rear port though.

I thought I had read many times that high Q narrow peaks and dips aren't as audible as the lower Q broad stuff. I could be wrong, but that's what I've been lead to believe.

 

[TurboFC3S]

TurboFC3S, have you tried listening to a pan flute across that dip?
Sorry. I can't stop laughing either.

I'll see if Arta or Speaker Workshop has a pan flute signal generator

 

[TurboFC3S]

FYI, next up is the MTM center and 2 way rear surround channels. Both will be sealed, since the M165x gives pretty decent F3 in a sealed box. 7.8mm xmax will come in handy there

The .75cu/ft surrounds model out at a F3 of 52hz ... not bad.