I need help diagnosing a crossover problem

[JonW]

OK, my lack of knowledge of electronics is now staring me in the face. I made 3 crossovers for the Modula MT’s. Checked them with a volt meter. Two of them come out the same and the third looks different. I would appreciate any help in figuring out what is going wrong.

The details: The two ”good” crossovers measure 6.9 ohms, the “bad” one is 7.0. I played test tones through the crossovers and measured voltage at the woofer and the tweeter. (Putting the meter leads on the woofer or tweeter leads.) In all 3 crossovers, the woofer response at a given frequency was about the same. But for the tweeter, the “bad” crossover is reading about 20% low. I put a table below of typical values. (Sorry, I can't get the spacing to make it look easier to read.) Just one set of values for the woofer because they’re all about the same. And a typical “good” and “bad” set of values for the tweeters.

Any idea what’s wrong? I tested each solder joint with the meter and they all look OK. I double checked all the components. They look to be correct as far as which piece is where. I switched around the tweeters and woofers. And, basically, I have no idea what to do now. Pathetic, eh? Thanks in advance.

-Jon



frequency woofer “good” “bad”
(Hz) tweeter tweeter
----- ----- ----- -----
200 .035 0 0
500 .026 0 0
800 .024 .001 .001
1200 .019 .003 .002
1500 .016 .005 .004
1800 .013 .007 .006
2000 .011 .009 .007
2400 .008 .010 .008
2700 .007 .011 .008
3000 .005 .011 .009
3500 .003 .010 .008

[ThomasW]

You're measuring the assembled crossovers attached to the drivers?

If that's the case the difference between 6.9 and 7.0 ohms is not significant in this situation, given the normal tolerance variations in crossover components and drivers.

[JonW]

You're measuring the assembled crossovers attached to the drivers?

Yup.

If that's the case the difference between 6.9 and 7.0 ohms is not significant in this situation, given the normal tolerance variations in crossover components and drivers.

Right. It's not the 6.9 versus 7.0 that I'm concerned about. It's that the tweeter voltage is differing by about 20% low on one of the 3 crossovers. I'd think that's not normal?

Thanks.

[Brian Bunge]

You're measuring in the millivolt range here. Those differences look negligible to me. But then again, with all the crossovers I've built, I never bothered to measure voltages at the drivers.

[ThomasW]

I don't think you have a problem (it maybe just variations in the tweeters particularly of you're using the RS28's).

But you could mount one set of drivers in a box. Run leads out the port from each driver, then wire up the crossovers and listen to them individually.

[chasw98]

Jon:
Screw a set of woofer and tweeter into your enclosure and run the wires to the speakers through the port in the back. Then hook the 1 of the 3 crossovers up and play some music (not too loud at first) through the speaker. Then hook the next one up and so on and so forth. If you hear a difference between them then troubleshoot. As Brian said, I think you are seeing just normal production variances between the components and the drivers. It will also be helpful if you go to Radio Shack and buy a bag of alligator clip leads to use as jumpers to "clip" in the crossovers. Call me tonight if you have any questions.

Chuck

[Dennis H]

Jon, you need a bigger signal, like 1 volt, to get some more precision in your measurements. It's hard to tell anything meaningful from single digit numbers.

[JonW]

You guys are great. Thanks. :T

Jon, you need a bigger signal, like 1 volt, to get some more precision in your measurements. It's hard to tell anything meaningful from single digit numbers.

You're measuring in the millivolt range here. Those differences look negligible to me. But then again, with all the crossovers I've built, I never bothered to measure voltages at the drivers.

I can try that easily. When I turn up the volume, the voltage does increase. But I was keeping it low for fear of doing any damage to the drivers outside of the cabinets. So I guess 1 volt might be OK.

Also, two of the crossovers matched their low numbers pretty well. So the third looking lower definitely caught my attention.

[JonW]

I don't think you have a problem (it maybe just variations in the tweeters particularly of you're using the RS28's).

Interesting if it's OK. I would have thought a 20% difference in tweeter levels would be far from what is typically OK.

I tried switching the tweeters, actually, with no change. The difference seems to be in the crossover.

But you could mount one set of drivers in a box. Run leads out the port from each driver, then wire up the crossovers and listen to them individually.

I'll try that- thanks,

[JonW]

Jon:
Screw a set of woofer and tweeter into your enclosure and run the wires to the speakers through the port in the back. Then hook the 1 of the 3 crossovers up and play some music (not too loud at first) through the speaker. Then hook the next one up and so on and so forth. If you hear a difference between them then troubleshoot.

Thanks. I'll try that.

As Brian said, I think you are seeing just normal production variances between the components and the drivers.

Really? Wow. I would have thought the tolerances to be much lower. With all the design effort that goes into a crossover... to then have one tweeter voltage be 20% off another...

It will also be helpful if you go to Radio Shack and buy a bag of alligator clip leads to use as jumpers to "clip" in the crossovers.

I've got those little white screw terminals on there now, so that should make all this easy to do. I'll give you a ring.

[ThomasW]

Jon:
Screw a set of woofer and tweeter into your enclosure and run the wires to the speakers through the port in the back. Then hook the 1 of the 3 crossovers up and play some music (not too loud at first) through the speaker. Then hook the next one up and so on and so forth.

Oh sure steal my idea and phase it in language a person can understand.....

[chasw98]

Oh sure steal my idea and phase it in language a person can understand.....

I speak JonW-ease! :T

[JonW]

I speak JonW-ease! :T

I guess that's where a Ph.D. will get you- no working knowledge of electronics. And people have to speak to you in very small words. Slowly.
:P

[JonW]

OK, I did everything you guys suggested. Confusion persists.

-First, I measured all 3 crossovers with a much higher voltage. I set the voltage such that the woofer was about 1 volt at 200 Hz. The numbers are below. Two of the crossovers are very close to each other and the third is pretty far off for the tweeter. The woofer voltage for all 3 was pretty much the same, so I’m just showing the tweeter numbers.

I’ll try to provide the numbers like this…
Frequency
Voltages from crossover 1, 2, and 3

200 Hz:
0, 0, 0
500Hz:
.016, .017, .016
800 Hz:
.037, .038, .036
1200 Hz:
.089, .090, .079
1500 Hz:
.150, .151, .123
1800 Hz:
.218, .218, .163
2000 Hz:
.259, .259, .186
2400 Hz:
.312, .312, .220
2700 Hz:
.325, .325, .235
3000 Hz:
.326, .325, .244
3500 Hz:
.314, .314, .248

As you can see, crossovers 1 and 2 are really close to each other, but #3 is definitely different.

-Second, I mounted the drivers on a cabinet and ran wires from the drivers out the port. Then played 2 songs using crossover 3. Sounded fine to me. Switched to crossover 2. Also sounded fine. Switched back to 3 and then 2. I couldn’t tell a difference.

So I’m confused. Thanks again for all the help.

[ThomasW]

Jon,

Yours is a fundamentally flawed testing concept so don't fret with the results.

That fact that you heard no differences is your clue...

If you want to test, hook up an assembled system, run sine sweeps (test tones) , and measure the acoustical output (SPL) from the drivers.

[Dennis H]

While what Thomas says is true, it's the SPL that counts, you're measuring about a 2dB difference (20*log10(V1/V2)) and that's pretty big even allowing for normal component tolerances.

Have you triple checked that all components are the right values? For example, if R3 were too big, it could do what you are measuring -- tweeter padded down about 2dB across the band. Now that I think about it, what brand of resistors are you using? I've heard of the leads breaking loose internally on the Eagle resistors from Madisound if you bend them too close to the body.

Have you reheated all the solder joints just to be sure? Bad solder joints aren't always obvious.

Did you ever get an LCR meter? The only way you are really going to know for sure if you can't find anything obvious is to disassemble one or two crossovers and measure part values. You can't measure values with them assembled because the current can find another path around the component you're trying to measure. Start with the "bad" one and, if you don't find any obvious problems take a good one apart and compare values. The cheap meters may not be all that accurate but they should be repeatable comparing two components and measure somewhat close to the spec'd value.

[ThomasW]

I think he'll go insane trying to find this, since as Dennis says, the only way to do this properly is tear apart the crossovers. And doing that has the potential to do damage and create a worst situation.

Given there are 2 crossover that 'match', use those as the mains and the other as the center and sleep well tonight...

[JonW]

Thomas and Dennis-

Thanks _very_ much for the feedback. Much appreciated.

Have you triple checked that all components are the right values?

Well, I can only test the resistors with the borrowed meter I have (I think). I tested some (not all) of them before installing and they seemed OK (e.g., a 6 ohm resistor would be 5.9 or 6.0). But after having all 3 crossovers assembled, I put the meter leads on either side of the resistors and the values seemed odd to me. The 2 ohms all measured 1.9-2.1 but the 5 ohm resistors all measured 2.8-3.0, the 6 ohm ones measured 2.8-2.9. Odd, but consistent throughout all crossovers. This may be a different problem, but it’s not going to have one crossover different from the rest. Or it could be what you said here:

You can't measure values with them assembled because the current can find another path around the component you're trying to measure.

I used Mills resistors, perfect lay inductors, and Solen caps.

Have you reheated all the solder joints just to be sure? Bad solder joints aren't always obvious.

I used the meter for resistance to check every solder joint. I found one on the “bad” crossover that wasn’t right and fixed that. So they should all be OK now.

Did you ever get an LCR meter?

Not yet. I’m borrowing this regular DMM. The plan was to buy an LCR meter if this first foray into speaker building goes well and I’m itching to do some more. I recall there was a specific LCR meter that you guys recommended.

The only way you are really going to know for sure if you can't find anything obvious is to disassemble one or two crossovers and measure part values.

OK. That’s a real test of my dedication here. Hmmm… Might be easier to just build another from scratch. I do see your point, however.

If/when I make the next speakers, maybe I should get an LCR meter and measure all the caps, resistors, and inductors prior to building the crossovers. I didn’t do that this time.

[JonW]

Jon,
Yours is a fundamentally flawed testing concept so don't fret with the results.

Hey this is what Chuck [chasw98] told me to do.

That fact that you heard no differences is your clue...

Given there are 2 crossover that 'match', use those as the mains and the other as the center and sleep well tonight...

Super. :T

If you want to test, hook up an assembled system, run sine sweeps (test tones) , and measure the acoustical output (SPL) from the drivers.

Makes sense. Last night, as I was playing around with the crossovers and meter, my gal asked me if it wouldn’t be wiser to just assemble the speakers and use something to measure the actual sound they make. She’s smart. It’s probably time for me to pick up an SPL meter or microphone/amp/etc.



My main concern here, obviously, is just to get the crossovers OK (or know that they are already OK). So if all is well, I have nothing to worry about. Although I must admit that I’m curious as to why such difference would arise. Heck, one of the main reasons for undertaking this project was to learn some electronics- something I’ve never had a chance to learn before.

[ThomasW]

Hey this is what Chuck [chasw98] told me to do.

Remember he's so old he threatened to hit me with his walker..... 8O

So I'm not sure he's recommendations are reliable....:roflmao:

It’s probably time for me to pick up..... or microphone/amp/etc.

Those are much more fun to play with compared to a DMM...

[chasw98]

Remember he's so old he threatened to hit me with his walker..... 8O

So I'm not sure he's recommendations are reliable....:roflmao:

Those are much more fun to play with compared to a DMM...

Touche`, mon ami! ;x(
I vil zolve hiz problemo tonite by remote! ONWARD!

[BretH]

If Jon wasn't heck-bent on figuring out what's going on electronically, I would suggest a real A-B blind test. Otherwise I like this advice: "Given there are 2 crossover that 'match', use those as the mains and the other as the center and sleep well tonight..."

[JonW]

OK, Thomas, I’ll factor in Chuck’s senility the next time I build a speaker.
:P

Yeah, maybe I’ll pick up the Behringer microphone and that M-Audio preamp setup to be able to take measurements. Looks to be fun and interesting. It will further my speaker design education.


Chuck-
I’ll give you a ring tonight.


Brett-
Well, if I just put the speakers together with the crossovers as they are, I’ll then have the ability to do the A-B test. (Although I bought a shielded RS28 tweeter for the center channel.) So I can try that.

[augerpro]

Would this be a good all-in-one meter?:

http://www.partsexpress.com/pe/showdetl.cfm?&Partnumber=390-513

[PMazz]

Would this be a good all-in-one meter?:
http://www.partsexpress.com/pe/showd...number=390-513

I have one of those and it works fine. If you're like me you'll need to make up a set of test leads for measuring Inds/Caps.

I've had Mills resistors come in that were marked/pulled wrong.

Pete

[Brian Bunge]

Pete,

What kind of test leads? I've got that meter but have always just shoved the cap and coil leads into the little grooves in the front there. Did you basically must make up so test leads to put in there with alligator clips on the other end or something?

[JonW]

Hey Guys,

If you're looking for a meter, there was a thread recommending one:

http://www.htguide.com/forum/showthread.php4?t=19981&highlight=meter

I think I'll pick up the one augerpro linked to, but from the direct site, where it's a little cheaper ($60):
Go to http://www.cs-sales.net/
then click on test equipment, then multimeters, then LCM-1950
(the direct link doesn't alwayts work)

[Dennis H]

I've had Mills resistors come in that were marked/pulled wrong.

That's sure what it looks like to me. Jon, are you doing the 27TDFC version? R3 basically controls the gain of the tweeter and it's spec'd at 8 ohms.

[ThomasW]

As an aside, I've measured many samples of the Mills resistors, and none have met their tolerance specs. Oddly enough the cheaper PE (Dayton) resistors do meet their tolerance specs.

[JonW]

That's sure what it looks like to me. Jon, are you doing the 27TDFC version? R3 basically controls the gain of the tweeter and it's spec'd at 8 ohms.

I'm using the RS28 version. All the resistors seem to measure about the same between the different crossovers, now that they're in the crossover. Using the meter for resitivity (it's all I've got until I order that LCR meter), two of the caps seem different between the "good" and "bad" crossovers. One 12 and one 20 uF. I'm thinking about just picking up a couple replacement caps from a local shop, put them in there, and see if things improve. Sound OK? Thanks.

From measuring the resistors in the crossovers:

The 2 ohms all measured 1.9-2.1 but the 5 ohm resistors all measured 2.8-3.0, the 6 ohm ones measured 2.8-2.9. Consistent throughout all crossovers.

[JonW]

As an aside, I've measured many samples of the Mills resistors, and none have met their tolerance specs. Oddly enough the cheaper PE (Dayton) resistors do meet their tolerance specs.

Huh. I went with the Mills because someone here recommended them. I guess I'll use something different next time.

[ThomasW]

I've used the Mills for years and years without problems. One day I was messing around and slapped one on my LG True RMS meter and it was out of spec, so I started testing various resistors laying around the shop. Consistantly the Mills didn't meet spec and the Dayton's did. I thought that was strange given the price difference.

[JonW]

Thomas-

Yeah, that's strange. I'm not sure what my next project will be (Arvos?) but before starting it, I'll try and collect some recommendations for resistors, caps, inductors, etc. I did that for this project and thought the consensus was Solen caps, Mills resistors, and perfect lay 14 guage inductors.

It's good to know what to avoid and what tends to be good. The experience shared from folks like yourself is much appreciated.
(I'm passing through Denver mid January. If you're around, you can collect on the many beers I owe you. I'll drop you a line maybe next week when I return from a trip starting today.)

[mmoeller]

All the resistors seem to measure about the same between the different crossovers, now that they're in the crossover.

The thing is that your really not going to get an accurate reading measuring resistors, "in circuit". That has always been a no-no in my experience.

[chasw98]

The thing is that your really not going to get an accurate reading measuring resistors, "in circuit". That has always been a no-no in my experience.

Jon has 3 XO's built. 2 of them measure the same and 1 doesn't. He is not really looking to take them apart and see individual values, but he is comparing in circuit values for the good xo versus the "bad" xo. AFter doing that he has found different in circuit values for 2 capacitors, but he does not have a meter that will measure capacitance even if he took them out of circuit. I told him the best he can do right now is unsolder one end of the capacitor in the suspect xo and do the same for an end in the "good" xo and measure the resistance to see if they differ. If they differ, replace the one in the suspect xo. OTOH, he has said he can't hear a difference between any of the xo in testing. What a challenge.... remote troubleshooting!

Chuck

[PMazz]

Pete,

What kind of test leads? I've got that meter but have always just shoved the cap and coil leads into the little grooves in the front there. Did you basically must make up so test leads to put in there with alligator clips on the other end or something?

I always did the same until I started building the GE 10uF combos. I use test leads with alligator clips on one end and cut paper clips ops: on the other so they fit nicely into the slots on the meter. Measuring with and w/o the leads gives me the same values and they are a whole lot easier to use.

Pete

[Brian Bunge]

I always did the same until I started building the GE 10uF combos. I use test leads with alligator clips on one end and cut paper clips ops: on the other so they fit nicely into the slots on the meter. Measuring with and w/o the leads gives me the same values and they are a whole lot easier to use.

Pete

As the Guinness Beer guys say: "BRILLIANT!"

I don't really see anything wrong with the way you did it. You could easily strip back the wire, wrap it around the paper clip and solder it if the solder will stick to the paper clips.

Or better yet, do the same with the clipped component leads! :T

[JonW]

Jon has 3 XO's built. 2 of them measure the same and 1 doesn't. He is not really looking to take them apart and see individual values, but he is comparing in circuit values for the good xo versus the "bad" xo. AFter doing that he has found different in circuit values for 2 capacitors, but he does not have a meter that will measure capacitance even if he took them out of circuit. I told him the best he can do right now is unsolder one end of the capacitor in the suspect xo and do the same for an end in the "good" xo and measure the resistance to see if they differ. If they differ, replace the one in the suspect xo. OTOH, he has said he can't hear a difference between any of the xo in testing. What a challenge.... remote troubleshooting!

Chuck

Chuck-

Thanks soooo much for the help over the phone. :T I'm doubling the cost of these speakers with all the beers I owe you guys. :B I'm thinking a Keys scuba trip with a swing by your place would be nice.

Not sure how much weight I'd put in my not hearing a difference between the crossovers. It was some quick switching, so I don't really know.

Before leaving town today I stopped by the local electronics store and they didn't have caps with the correct values. So I placed a quick order with Parts Express. They should be waiting for me when I get back in town Friday. (Yay- hotel internet access. )

I was thinking of maybe not messing with a good crossover. Just unhook the 2 caps on the bad crossover. Measure them and compare to the new caps. Then replace whatver look bad. And hope that fixes it.

When I start on my next speaker project (not sure what that will be), I'm definitely going to buy an LCR meter and check all the parts before using them.

[JonW]

Pete,

What kind of test leads?

Can you not use regular leads with that meter? For the regular DMM I've borrowed from work, I'm using leads that have little clips on the ends, banana plugs on the other ends to put into the meter. Very handy. I think they're something like this:

http://www.radioshack.com/product/index.jsp?productId=2062624&cp=2032058.2032235.2032304&parentPage=family

[Dennis H]

Just unhook the 2 caps on the bad crossover. Measure them and compare to the new caps.

Jon, if you were just measuring across the caps with an ohmmeter (DMM), you aren't measuring capacitance. You're measuring the DC resistance of some circular path aroung the cap. The DC resistance of a cap is infinity. Please check R3 by unsoldering one end and checking the resistance.

[JonW]

Jon, if you were just measuring across the caps with an ohmmeter (DMM), you aren't measuring capacitance. You're measuring the DC resistance of some circular path aroung the cap. The DC resistance of a cap is infinity.

Right. Just trying to locate any differences in my primitive way.

Please check R3 by unsoldering one end and checking the resistance.

Will do. (When I get back in town late Thursday or Friday). Thanks for the help! ;x(

[PMazz]

Can you not use regular leads with that meter? For the regular DMM I've borrowed from work, I'm using leads that have little clips on the ends, banana plugs on the other ends to put into the meter. Very handy. I think they're something like this:
http://www.radioshack.com/product/in...entPage=family

To measure inductance or capacitance you can't use the regular leads. I really don't know why it's designed this way???? There are small slots in the meter just above where the leads plug in for inserting the ends of inductors or caps.

Pete

[chasw98]

To measure inductance or capacitance you can't use the regular leads. I really don't know why it's designed this way???? There are small slots in the meter just above where the leads plug in for inserting the ends of inductors or caps.

Pete

The reason you shouldn't or cannot use regular leads is that they will add to the capacitance or inductance of the component under test. Unless your meter has the ability to set a zero point or you remember to deduct the lead capacitance/inductance you will get readings offset by the amount induced by the test leads. On larger components this does not factor in greatly, but when trying to measure that .2 mH inductor, the results can be skewed by quite a bit. That is one reason for using the slots to stick the leads of the components into. I have also seen where the leads that come with some LCR meters give you very short 3 to 6 inch leads to use with them when it is impractical to stuff that 12 AWG inductor into the meter. Sort of like what Pete made up for himself.

Chuck

[mmoeller]

If your measuring a cap with the ohm meter and you ever hear "beeeeeep", that's bad.

[JonW]

Please check R3 by unsoldering one end and checking the resistance.

I did that lat night and the resistor looks fine (4.8 ohms). My replacement caps arrived and are now sitting here on my desk. So I'll try changing those tonight and see where that gets me.