This sounds like a recently purchased house and the previous owner took all of the equipment out.

Are the volume controls autotransformer type? If so they may be impedance matching without a switch. This type of control was common. The color of the transformer indicates the ratio.

The original installation may have included a large matching transformer that has been taken out. In this case the controls are probably 1x.

You could get a very crude sense of the situation by measuring resistance at the head end and the primary resistance of one autotransformer. Best to have all of the controls set the same. A better plan would be to measure the impedance, but this will require a series resistor, audio generator, a good AC millivoltmeter.

The way we created impedance curves with our analog Brule & Kjaer equipment was to put 10 volts of audio from a power amp through a 1000 ohm resistor, then to the speaker. 1 ohm of speaker impedance would have roughly 10 mv of signal across it. 8 ohms, 80 millivolts. (A power amp and an ac meter capable of resolving millivolts were required for this.)

If the installation was done without cable drops to the head end for each speaker or a known common location, I seriously doubt they used something so expensive and elaborate for maintaining speaker load. If they used anything pre-made, it would probably be this (or the equivalent from another company)-

[Link: jamo.com]

Hopefully, they didn't do what the installers did at one job I worked on- solder wire-wound resistors to zip cord and staple the wire to the floor joists. They didn't use a heat shield, either- the heat from the resistors charred the wood in a 6" circle.

Why not just use a 1 Ohm resistor and 1V of signal?

No, this is not as convenient. Do a few calculations and you'll see why. The idea is to use a series resistor that is orders of magnitude greater than the magnitude of the impedance that you are measuring. This transforms the constant voltage oscillator output into a constant current generator driving the load. (the impedance that we are measuring is essentially a short circuit compared to the series probe resistor) Since the current in the circuit is now constant, and we picked a clever series resistor value, one can just shift a decimal point in your head to derive the number.

And, if you use a 10K resistor, most audio oscillators can directly drive the load, eliminating one piece of equipment to carry around, but you need a better voltmeter. Take care when picking the voltmeter. Most garden variety digital volt meters have a terrible frequency response and are useless for audio work.

---

Obviously, the current is not exactly constant in our simple circuit because the magnitude of the impedance varies, but we don't really need to discriminate between 8.000 or 8.005 Ohms (or similar). Since the magnitude of the impedance varies with frequency, accuracy decreases somewhat as the measured impedance approaches our probe resistor value, but we really don't care about the accuracy near 100.000 or 1000.000 Ohms, we only care about the low point and this will likely occur under 100 Hz and 10 Ohms.

For finding WHERE Fs is, this is fine, but it doesn't give an accurate value if the resistor's tolerance isn't tight. I read that as using current to derive impedance and since none of the range between the peaks is important, some nominal number of samples near each can be used in an Excel sheet to find the results.

Why use a direct connection from an oscillator when the speaker(s) will ultimately be driven by an amplifier? Why not just use the amp?

This has raised some interest.

The 1000 ohm resistor, driven by a power amp at 10 volts, is a 1000 ohm source.

If you put 1 ohm in series with it, you'll get 10 mV across the speaker. But if your speaker had 100 ohms impedance, you'll get something like 91 millivolts, because 100 ohms is not appreciably smaller than the source impedance. This approach only is successful, as buzz says, if the resistor is substantially larger than the speaker impedance.


You don't need an accurate value. You need an approximate value. If you know, because you've thought about it, that the number won't be exact, you won't think of it as exact. The important thing is that you'll still have very useful information.


I'm not sure what you're talking about. When we used the 1000 ohm/10 volt method, we were printing curves of level (ohms) versus frequency, so we saw the entire thing, not just a few points in a worksheet.
Of course the range between the peaks is important! An 8 ohm speaker might have a 40 ohm resonance at 80 Hz and, if the crossover is good, no resonances at all above that. And what do you find in the average 8 ohm speaker once you're about an octave above the bass resonance? You find quite a portion of frequencies with an impedance of about 8 ohms. In other words, the range between the peaks is exactly what you want to know about.


Indeed. You can use a direct connection from an oscillator, but you have to know its source impedance. The 1000 ohm method sets up a way to read impedance almost directly from a meter. If the oscillator had a source impedance of 500 ohms, you'd have to do some math to get the impedance value.

You can't "just use the amp" because you need the oscillator to provide an audio frequency signal.

This is, by the way, a method of getting the actual impedance of the speaker. You know how everybody freaks out when someone mentions measuring speaker impedance with a DC voltmeter, and then I respond (again) that you can learn things by measuring this way, but you can't tell what the actual impedance is? Well, this gets you the actual value (as long as the resistor is substantially greater in resistance than the impedance).

There is nothing wrong with using an amplifier between the oscillator and the resistor, but it is something extra to carry around and connect to a power outlet. You can argue that using an amplifier allows running at a higher voltage and this will increase accuracy somewhat, but we don't need 0.2% accuracy. 10% accuracy would be fine. The goal is to determine if whatever is connected through the speaker wire network will stress a power amplifier because of a low point in the magnitude of the impedance. (It also pays to check for ground faults.)

In any case this is a quickie back of the napkin approach because the test does not directly measure any reactive components. A very reactive network could push a low quality amplifier into an unstable operating region or, if there is a hidden matching transformer, the transformer or autotransformer controls could saturate if hit with too much current.

When facing this sort of system takeover, before connecting any equipment, I'll usually pull the volume controls and at least check to see if they are burned and wired correctly. If there is any question about the network topology, I'll disconnect the controls from their feed and measure the network, expecting to find open circuits all around. I'll also short at least a couple of wires at the control end in order to make sure that I have a direct wire connection between the two points. I'll use my circuit toner to make sure that all of the speakers make noise.

If you were using B&K test equipment, why didn't you just get an impedance meter?

The standard way to calculate impedance is by using the voltage drop across the voice coil and doing the calculations. It requires two meters, or, connecting the meter across the coil but it can't be a VTVM. Why not do it that way? You don't have a phase plot, but at the time, I'm not sure that was used for speaker design.

"You can't "just use the amp" because you need the oscillator to provide an audio frequency signal."

Duh. Really? The function generator wouldn't be connected to the amp? GMAB. The assumption that I didn't know this is almost offensive. How the hell would connecting ONLY an amp, with no input signal, be useful for this?

This is a technical discussion. You asked why not just use the amp.

We were not calculating impedance. We were looking at impedance curves and response curves so we could tweak the crossovers for the best audio response. You can't make an impedance curve with an impedance meter.

I should back up and say that the 1000 ohm setup was designed specifically to give an output for drawing an impedance curve. We wouldn't have taken the time to make such a thing just for a single frequency measurement.

The first time I saw this I was astounded! The great, low-distortion midrange we were considering had an overall rising response. My boss, Richard May, managed to create a midrange crossover with an overall declining output. Result? Flatter curve than possible if we had not been able to tweak the impedance. One impedance calculation, even several, would not have made that possible.