How to measure a 3 way passive crossover

I'm trying to track down a subtle problem with a pair of 3way passive professional speakers.  I'm getting a subtle variation in frequency response and I'm trying to determine whether it is in the crossover or drivers. 

I've ruled out the source/amps.  I've swapped tweeters and this was inconclusive, swapping mid range units is difficult but possible and I can access spares.

For my own peace of mind I'd like to measure the crossovers and I know this will be difficult because of the typical low impedance of drivers and the 10k input impedance of my measuring gear (REW/Fuzzmeasure and pro DAC/ADC). 

I don't need to measure the actual response of the circuit, just to isolate any differences between the sides.

Before anyone suggests measuring the whole rig acoustically I've never found this to be satisfactory when dealing with small and repeatable differences, I have recent experience on an install at a world famous studio where we could blatantly hear a difference that we could not measure.

Any suggestions?

best would be to measure across a suitable resistor. now this will change the exact frequency response you will be able to see differences between channels if they exist. just hook up a resistor slightly smaller then your nominal chassis impedance (say 6.8 ohms for a 8 ohms chassis) instead of your hf / mf / lf driver and measure across. usually drivers have Rdc some 15% to 20% smaller then Rnominal. Keep low to reasonable volume, otherwise you will need a fairly large resistor.

cheers,

Michael

Thanks Michael.  I'll need all the resistors in place to simulate having drivers in place?

it seems swapping the drivers should isolate the problem to driver or network.

Are the speaks symmetrical. If yes, perhaps swap L with R to establish that room (or your hearing) is not a factor.

JR

JohnRoberts said:

it seems swapping the drivers should isolate the problem to driver or network.

Are the speaks symmetrical. If yes, perhaps swap L with R to establish that room (or your hearing) is not a factor.

JR

Click to expand...

Swapping tweeters (where the issue is most present) wasn't conclusive.  This is a small but annoying variation. 

The room is perfectly symmetrical (hi end build out), as is the studio furniture.  Hearing also ruled out.

I've had this niggle for a while despite some driver changes a while back so I'm keen to rule out the crossover, without having to remove them if possible (it's not easy).

One of my hats is to do hi end studio speaker installs so I've got a lot of experience ruling out typical variables.

Cheers,
Ruairi

I would do crossover measurements with real drivers connected, using small power (1W or so) with pink noise and some
impulse stimulus as a source. The drivers have complex impedance, so replacing them with resistors may not show the difference between channels. 

the drivers could have different impedances and be the problem, so replacing them with resistors (non complex imp, I know) does take them out of the equation. that's what the OP wanted in the first place. things like gunked ferro fluid would change the HF drivers impedance for instance. The question asked is whether the dividing network has changed, due to component failure or aging. In one case I had a speaker with inverted mid driver, shoving slight measurement disagreement between L and R. only one mid driver has inverted, and that was a mistake on the driver itself. the apparent polarity was correct. a mistake in the chassis production. took me ages to find out, because the error was subtle and the visual check non conclusive. at least that way you can reduce the possibilities.

I would indeed have all three ways replaced by fixed & known resistors. I would not expect interaction between the bands, but hey you never know.

- michael

While a little work, you could Solder wires to both tweeters and feed them into two channels of a console. Flip the polarity on one and assign to the same bus. See how well they null. If the two drivers are getting a different frequency response they will not null ever. If they are getting different levels you may see this in the null, but small differences are not as visible. 

A very subtle difference may be a driver distorting only occasionally so hard to troubleshoot.

JR

ruairioflaherty said:

I'd like to measure the crossovers and I know this will be difficult because of the typical low impedance of drivers and the 10k input impedance of my measuring gear (REW/Fuzzmeasure and pro DAC/ADC). 

Click to expand...

There is absolutely no problem measuring the voltage at the tweeter's voice-coil with a medium-high impedance set-up.
I would recommend measuring the cross-overs with a dummy load resistor (it does not need to be the exact nominal value) AND with the tweeters connected.
You have to be careful with the power you apply to the system that may fry the input of your ADC.
The interaction between cross-over and driver is a complex matter. A trivial deviation of the tweeter response may generate noticeable transfer-function differences.

I have recent experience on an install at a world famous studio where we could blatantly hear a difference that we could not measure.

Click to expand...

I would put my money on acoustic differences. Conventional acoustic measurements measure sound pressure, which is only one of four components of the acoustic energy. The audition process may detect differences in velocity that a pressure microphone will ignore.
I will spare you the differences created by different decay characteristics.

abbey road d enfer said:

ruairioflaherty said:

I have recent experience on an install at a world famous studio where we could blatantly hear a difference that we could not measure.

Click to expand...

I would put my money on acoustic differences. Conventional acoustic measurements measure sound pressure, which is only one of four components of the acoustic energy. The audition process may detect differences in velocity that a pressure microphone will ignore.
I will spare you the differences created by different decay characteristics.

Click to expand...

Yes in this case it was definitely acoustic and differences in some relevant physical structures.  I always trust ears more than measurement, but both are very useful.

Thanks for the info re crossover measurement.

Cheers,
Ruairi

As you have REW and a professional soundcard with balanced inputs, just measure the response from the xover input to each of the drive unit terminals.  The units should be connected and preferably in the enclosures.  Put the balanced line inputs of your professional across each speaker unit in turn.

Use ONLY about 1V at the xover input / power amp output.

If you see a big difference remove the units, substitute 8R resistors and re-measure.  If the difference goes away, at least one unit is damaged.  Otherwise its the xover.