Anyone using LTspice to simulate and REW to test crossovers?

[mboxler]

Finally got REW to work, I think.  Discovered I can use my computers Line In to capture the voltage across the crossover outputs. 

 

Line Out-->Amplifier with Volume control-->Crossover Input-->Crossover Output (with dummy resistor across the output as well)-->Line In.

 

I had success with a simple 47uf capacitor/8 ohm resistor circuit combined with a 2.5mh inductor/6 ohm resistor circuit...Pretty close!

 

Unfortunately, my test "jig" broke (two wires soldered to a pcb mount 1/8" jack's pins).  Ordered a breakout board that I hope works better.

 

I had to be careful and use an amplifier with single ended outputs.  I assume connecting an amp with bridged outputs to the Line In jack would be a disaster.

I also don't know what the input impedance of my computer's line in jack is.  I assume it's impedance, in parallel with the test resistor, might affect the REW test.

 

Curious how others test crossovers before installing them in their speakers.  

 

Mike

 

 

[mboxler]

Here are my LTspice and REW plots for the high pass of the ALK ES5800t crossover.  Can someone tell me what's going on with REW below the 70db level?

Is there a measure setting I'm getting wrong, or is this noise on the TRS connection from the crossover to the Line In jack?

 

Thanks, Mike

 

 

[mboxler]

I thought I'd use this technique to see what, if any, effect the inductance of an autoformer (in this case a T2A) had on the frequency response.

 

I ran two measurements.  

 

First (red) was a 27.2uh capacitor in series with a 14.62 ohm resistor.

 

The second (green) was a 6.8uf capacitor connected to tap 5 of a T2A. I then connected tap 3 to the same resistor.  Since the load on the capacitor now 4 times the resistor, I used a 27.2 / 4 capacitor.  I then added 6db to the green plot.

 

As I thought, the inductance between taps 0 and 3 of the T2A does contribute to the frequency response.   

Trying to figure out how to run this same test on a K-55M/K-400 to get a real world plot.  

 

Mike 

 

[mboxler]

I realize I'm one of very few that finds this interesting, but at least I have a handy way of documenting my results.  Apologies to all that view this, only to immediately press the back button.

 

My experiments have revealed some interesting things (to me anyway) regarding low pass filters, in my case, to a K-33-E installed in my Khorns.

I never realized how much the voice coil influences a passive crossover.

 

On the attached plot, the yellow represents the voltage across the K33 in series with a 2.5mh inductor.  Notice how the voltage starts to drop very early and continues to drop, eventually at a 3db per octave rate.

 

I've never used a zobel network, and decided to give it a try.  I settled on a 4 ohm resistor in series with an 80uf capacitor, in parallel with the K-33.  Since this gives me a pretty constant impedance of around 4 ohms from 200hz on up, I chose a 1.6mh inductor for the next test.  A 1.6mh inductor is around 4 ohms at 400hz.  The purple represents the K33 voltage.  As you can see, it looks more like a first order low pass at 400hz.

 

I'm curious why zobels aren't used more often.  With one, it's easier to shape the woofer voltage knowing that the voice coil inductance isn't messing with the frequency response of the crossover.

 

 

[Edgar]

25 minutes ago, mboxler said:

I never realized how much the voice coil influences a passive crossover.

https://community.klipsch.com/index.php?/topic/204821-kp-3023002-high-frequency-speaker-intermittent-scratchy-sound-resolved-bad-cap-connection/&do=findComment&comment=2699406

 

https://community.klipsch.com/index.php?/topic/187697-measurement-systems/&do=findComment&comment=2436802

 

[mboxler]

51 minutes ago, Edgar said:

https://community.klipsch.com/index.php?/topic/204821-kp-3023002-high-frequency-speaker-intermittent-scratchy-sound-resolved-bad-cap-connection/&do=findComment&comment=2699406

 

https://community.klipsch.com/index.php?/topic/187697-measurement-systems/&do=findComment&comment=2436802

 

 Thanks for the links!  My searches didn't help much.

[StevePaul]

6 hours ago, mboxler said:

I realize I'm one of very few that finds this interesting, but at least I have a handy way of documenting my results.  Apologies to all that view this, only to immediately press the back button.

 

My experiments have revealed some interesting things (to me anyway) regarding low pass filters, in my case, to a K-33-E installed in my Khorns.

I never realized how much the voice coil influences a passive crossover.

 

On the attached plot, the yellow represents the voltage across the K33 in series with a 2.5mh inductor.  Notice how the voltage starts to drop very early and continues to drop, eventually at a 3db per octave rate.

 

I've never used a zobel network, and decided to give it a try.  I settled on a 4 ohm resistor in series with an 80uf capacitor, in parallel with the K-33.  Since this gives me a pretty constant impedance of around 4 ohms from 200hz on up, I chose a 1.6mh inductor for the next test.  A 1.6mh inductor is around 4 ohms at 400hz.  The purple represents the K33 voltage.  As you can see, it looks more like a first order low pass at 400hz.

 

I'm curious why zobels aren't used more often.  With one, it's easier to shape the woofer voltage knowing that the voice coil inductance isn't messing with the frequency response of the crossover.

 

 

Voice coil inductance also affects how well the lower sensitivity woofers match up to the higher sensitivitysquawkers in the 400hz-800hz range.  Take a woofer with an Le of 1.6 compared to a woofer with an Le of 0.6, ....and the 1.6 Le woofer is going to be general sloping down of several db's in that 400-800 zone compared to the 0.6 Le woofer.  The presence region becomes more difficult to control/maintain with a high inductance woofer voice coil, without a zobel. Or...use a woofer that has a low voice coil inductance..mabye 1.0mH or less.  Of course once you add in baffle step and other factors, it is more involved; but voice coil inductance is an important aspect to consider.  

[DrWho]

On 12/3/2021 at 8:58 AM, mboxler said:

Here are my LTspice and REW plots for the high pass of the ALK ES5800t crossover.  Can someone tell me what's going on with REW below the 70db level?

 

You're hitting the noise floor of your measurement rig. You can probably play with your gain structure to improve this a bit, but you'll always run into this limitation when measuring filters. At some point the filter attenuates the signal so much, that it's below the noise floor of your measurement gear.

 

[DrWho]

On 2/9/2022 at 7:55 AM, mboxler said:

I'm curious why zobels aren't used more often.  With one, it's easier to shape the woofer voltage knowing that the voice coil inductance isn't messing with the frequency response of the crossover.

I think there are a few reasons...

 

1) The goal of a xover design includes more than the voltage response at the transducer. You're trying to blend the acoustic output of two drivers together, AND you're trying to offset the power response of the speaker. The ideal voltage response will therefore be different than the ideal filter calculators compute.

 

2) A Zobel requires extra parts, which adds cost. For the same price, you could go to a higher order filter and get more frequency shaping.... Whereas a Zobel doesn't necessarily move you in the right direction (although sometimes it does). The Zobel is just one of many filter tools to achieving a desired result.

 

3) A Zobel consumes power, although it depends on how it was designed. High power dissipation tends to be a recipe for injecting distortion.

[Edgar]

7 hours ago, DrWho said:

2) A Zobel requires extra parts, which adds cost.

 

I suspect that this is the dominant factor, by far. It's a simple fact that parts cost gets multiplied by 5 to 10 in the final retail price. Sure, an added resistor and capacitor may only cost a few dollars, but manufacturers cut costs wherever they can.

 

7 hours ago, DrWho said:

3) A Zobel consumes power, although it depends on how it was designed. High power dissipation tends to be a recipe for injecting distortion.

 

True, but the voice coil still sees approximately constant voltage. Having a nearly constant, nearly resistive load makes amplifiers happy.

[mboxler]

On 2/11/2022 at 10:24 PM, DrWho said:

I think there are a few reasons...

 

1) The goal of a xover design includes more than the voltage response at the transducer. You're trying to blend the acoustic output of two drivers together, AND you're trying to offset the power response of the speaker. The ideal voltage response will therefore be different than the ideal filter calculators compute.

 

2) A Zobel requires extra parts, which adds cost. For the same price, you could go to a higher order filter and get more frequency shaping.... Whereas a Zobel doesn't necessarily move you in the right direction (although sometimes it does). The Zobel is just one of many filter tools to achieving a desired result.

 

3) A Zobel consumes power, although it depends on how it was designed. High power dissipation tends to be a recipe for injecting distortion.

 

I agree with all your points.  In my case, much of what I'm doing is educational.  When I first started this "hobby", I knew nothing about crossovers.  I then built ALK's Universal kit, and had no idea what the components were doing.  The schematic called for a 1.3mh woofer inductor.  Al's point was that the 1mh voice coil should be subtracted from the normal 2.5mh inductor used in, say, the Type AA.  A couple of years ago I replaced the 1.3mh with a 2.5mh.  I soon began to wonder what was really going on.

 

Attached are plots of various lowpass filters (voltages are across a K-33-E in my Khorn).

 

Blue is 1.3mh
Green is 2.5mH
Black is AK-3
Red is AK-4 (I only have a 5.4mH inductor, so I used 200uf to arrive at the same resonance frequency)

 

Next step is to measure SPL using these filters.

 

[Wirrunna]

Looking forward to seeing the results of the SPL measurements.

 

 

[Wirrunna]

Mike, when you start SPL measurements you will need to settle on a microphone position.

 

I always set up so that the tip of the UMIC is 1 m from the center of the Eliptrac horn and 118 cm off the floor.

 

 

[mboxler]

Well, measuring voltages is a lot easier than SPL 🙁.  This is what I get measuring my Khorn bass bin only, no crossover.  I assume the big suck out at 60hz is my room?  1/24 smoothing.

 

 

[captainbeefheart]

48 minutes ago, mboxler said:

I assume the big suck out at 60hz is my room?

 

Certainly could be.

 

What amp are you using to drive the bass bin? Amps with a high output impedance will track the impedance plot of a driver and 60Hz is about where the impedance dips down to it's lowest spot around 4 ohms. It continues to track along with the high impedance resonance area below it.

[mboxler]

13 hours ago, captainbeefheart said:

 

Certainly could be.

 

What amp are you using to drive the bass bin? Amps with a high output impedance will track the impedance plot of a driver and 60Hz is about where the impedance dips down to it's lowest spot around 4 ohms. It continues to track along with the high impedance resonance area below it.

 

Nice idea, but I decided to measure my other Khorn, and came up with this plot.  Red is the left speaker, green the right.  Way different up to 75hz.  When my back recovers from shoveling the snow coming in later, I think I'll swap the two speakers and re-measure.  This will tell me if it's the room or the speakers???

 

 

[captainbeefheart]

We can rule the amp out. I am curious what you find swapping things around. Must be room interaction or possibly something off with the cabinet/speaker.

[Wirrunna]

Mike, Deang posted this anechoic  response

 

 

[mboxler]

On 2/9/2022 at 1:45 PM, StevePaul said:

Voice coil inductance also affects how well the lower sensitivity woofers match up to the higher sensitivitysquawkers in the 400hz-800hz range.  Take a woofer with an Le of 1.6 compared to a woofer with an Le of 0.6, ....and the 1.6 Le woofer is going to be general sloping down of several db's in that 400-800 zone compared to the 0.6 Le woofer.  The presence region becomes more difficult to control/maintain with a high inductance woofer voice coil, without a zobel. Or...use a woofer that has a low voice coil inductance..mabye 1.0mH or less.  Of course once you add in baffle step and other factors, it is more involved; but voice coil inductance is an important aspect to consider.  

 

I measured Le on my Hersey K-22-R's (Rola, I believe), and got 2.2mH @ 1000hz 😮.  I realize voice coils are not true inductors, but is it okay to measure them this way?  Using a DER EE DE-5000 LCR meter.

[mboxler]

During my crossover tests, I noticed something interesting.  Connecting a tweeter circuit to tap 5 of the autoformer attenuates the squawker by another 1.5db or so.  I'm sure this is not news to some of you, but to others like me...

 

I don't have a Type A crossover, so I simulated one.  I don't have a 13uf capacitor, but was able to parallel a couple to reach 12.7uf...close enough.  This was connected to tap 5 of a T2A, and I connected tap 4 to the K55M in my Khorn.  The purple plot shows the voltage across the K55M.  It's down around 3db.

 

I then connected the tweeter circuit to tap 5 and noticed the extra 1.5db drop across the K55M (brown plot).  That attenuation is pretty flat throughout the useful frequency response of the K55M.

 

Was the reason Klipsch used a T4A due to the fact that the AK-3 tweeter filter is not connected to tap 5?