Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Jeff Schoonover1]

I am glad you found the wiring error, I tried like crazy to find it in the picture for you.

To truly understand though you need to stop using the word resistance.  It's impedance.  This is an AC network not DC.  You are building a frequency dividing network and part of the math is figuring out how it will perform at the extents.  That includes component tolerances and frequency curves. 

This multi-series analysis is called the Monte Carlo Method.  You will find some good spreadsheet templates to help with the task if you google that phrase (is it OK to use Google as a verb?0.

Good luck

Look closely.  Do you see how the last inductor from the low side is connected to the output of the first cap in the high side?  Soon as I saw it, I promised myself not to drink beer while wiring again.  DOH 
Yes - impedance.  Resistance varies with frequency...
As far as Monte Carlo - variables like component tolerances, specs vs. real and so on... Without lots of fancy equipment, there's a certain amount of trial and error isn't there?  I have calibrated mic and software to analyze the finished product, but the design is kind of guessing and refining, isn't it?

[Scott Holtzman]

Look closely.  Do you see how the last inductor from the low side is connected to the output of the first cap in the high side?  Soon as I saw it, I promised myself not to drink beer while wiring again.  DOH 
Yes - impedance.  Resistance varies with frequency...
As far as Monte Carlo - variables like component tolerances, specs vs. real and so on... Without lots of fancy equipment, there's a certain amount of trial and error isn't there?  I have calibrated mic and software to analyze the finished product, but the design is kind of guessing and refining, isn't it?

No it's just math. Resistors at 10% caps, whatever they are.  The analysis will wrangle those series for you. 

The drivers should have published impedance curves too.  You do need a test rig for that.

 

[Chris Grimshaw]

I have calibrated mic and software to analyze the finished product, but the design is kind of guessing and refining, isn't it?

To some extent, yes. Simulations can help, though. I've just completed my own passive crossover design for a pair of small speakers, where bi-amping didn't make sense. 18Sound 6ND430 and a Das M3 compression driver.

What I did was build the boxes and measure the drivers (impedance and frequency response) in those boxes. Then I fired up XSim and played around until I got something good, that also met the criteria of keeping the HF driver safe and the off-axis response reasonably smooth.

You can see what happened here: http://www.diyaudio.com/forums/pa-systems/271132-chris661s-pa-system-thread-9.html#post5359495

Chris

[Art Welter]

As far as Monte Carlo - variables like component tolerances, specs vs. real and so on... Without lots of fancy equipment, there's a certain amount of trial and error isn't there?  I have calibrated mic and software to analyze the finished product, but the design is kind of guessing and refining, isn't it?

Using a program as Chris does like XSim will get you close, but the amount of trial and error after is still pretty uncertain, and can't be determined without measurement.
You may be able to change only one or two values and get "close enough", but the more poles you add, the more complicated the interactions.
Both frequency response and impedance measurements are needed- "flat" response might end up with a narrow band impedance dip that is problematic with a certain amplifier (damn it..)...

Art



 

[Chris Grimshaw]

Using a program as Chris does like XSim will get you close, but the amount of trial and error after is still pretty uncertain, and can't be determined without measurement.

Art

Very true. XSim and REW measurements didn't quite see eye-to-eye, despite my best efforts. It got me close enough that a little trial and error got me where I wanted to go, though.

Chris

[Ivan Beaver]

Using a program as Chris does like XSim will get you close, but the amount of trial and error after is still pretty uncertain, and can't be determined without measurement.
You may be able to change only one or two values and get "close enough", but the more poles you add, the more complicated the interactions.
Both frequency response and impedance measurements are needed- "flat" response might end up with a narrow band impedance dip that is problematic with a certain amplifier (damn it..)...

Art

Agreed, impedance dips are real easy to have.

Or part of the impedance curve in which the phase of the impedance is to inductive or capacitive, putting additional strain on the amp.

There are a number of things that you must pay attention to, or you can put a strain on various parts of the system.

As usual, there is a lot more than simple amplitude to be concerned with.

[Jeff Schoonover1]

Darn you guys, LOL always doing things correctly... 

I checked a couple vids on X-Sim.  I've worked on preamps with LTSPICE, so it looks pretty easy.  I'm also pretty good with REW (which I was going to use to measure anyway) and, like the guy in the vid, I have boxes of wire, caps, resistors, etc.  I've also thought of adding an L-Pad, and maybe an impedance smoother.  Why not do it correctly...  It'll be interesting to see how far off the online calc is.  This is just the top end of my bass guitar rig, but why not make it optimal.
(As an aside, this is really silly.  For my last three gigs, one of which was a semi-known touring band - it's all been in-ear monitoring. i.e. no backline at all.  I don't even bring speakers or amps.  Even in the clubs, noisemakers on stage are becoming very passe for bands who actually want to sound good.  The new gig I have is also with IEM's only. This is kinda for my own amusement I guess.)

One thing I saw which raised concern.  In the vid, the guy had his speaker set up in the garage with his mic a meter away.  I know this would result in a horribly skewed response with a sub but will room variances really not be an issue with a mid/high?  Was thinking I had to wait for a nice day and a friend with a big backyard to measure accurately.  I'd love it if I can measure here at home.

Another thing is finding the driver simulations/measurements for X-Sim. The gentleman seemed to have a couple on dropbox but not what I need. Anyone know where I can find them for an Eminence Delta 10A and a JBL 2426?  The horn makes a huge difference too.  In the interests of space, I had to use a non-optimal one.  Maybe I should just measure it myself outside the box.

[Mark Wilkinson]

The best indoor measurement technique I've found, is to put both speaker and mic on a hard surface floor, away from other objects if possible.
1 meter works.  Measurements are of course more suspect the lower they go, but I'm often pleasantly surprised how well the on-the-floor-indoor hold up above 100Hz, compared to outdoors.

[Jeff Schoonover1]

I'm often pleasantly surprised how well the on-the-floor-indoor hold up above 100Hz, compared to outdoors.

Have you compared?  How much difference do you get?  Because yeah, you're right - with lows, measuring indoors is completely unusable, unless it's anechoic somehow...
How far away from walls and objects was the speaker when you measured?  My place is pretty small.

[Mark Wilkinson]

Have you compared?  How much difference do you get?  Because yeah, you're right - with lows, measuring indoors is completely unusable, unless it's anechoic somehow...
How far away from walls and objects was the speaker when you measured?  My place is pretty small.

Yep, made lot's of comparisons.  Below is a mid section box, green trace is on the floor, blue is outside.  1/6 oct.
By outside, I mean off a deck putting speaker 15ft off ground  pointing away from house.  Mic is on a boom 9ft out from speaker.

I find the floor method works with stuff fairly nearby.  Just get in the center of the room if you can. 
Oh, nearby drivers, even shorted out....are often worse than fixed objects.

I lay speakers on their side, and block the back side up to have speaker point perpendicular to mic on floor.  Makes for good on-axis.
It pays to just move mic around to make sure measurements aren't too mic position sensitive, needing averaging.

FWIW, with REW, I'm willing to eq per floor measurements, using frequency dependent widowing of FDW 5 or 6. 
Any finer resolution is probably (but not necessarily) letting too much room in, or too mic position dependent.

Hope this helps!

[ProSoundWeb Community]