inductive versus non-inductive resistors as loads for audio amplifiers

[calzap]

I was looking for some 8 ohm resistors to use for loads for testing some low power (20 W per channel) audio amp modules and assumed I would need non-inductive resistors of appropriate wattage.  When I started looking on eBay and other sources, I found non-inductive resistors, but  found even more resistors being sold as audio loads for which inductance was not mentioned.  I assumed these were regular wire-wound resistors and was curious as to their inductance and ultimate effect on impedance at audio frequencies.

There's a nice video on the topic on youtube:

I happened to have a couple of 8.2 ohm, 10 watt wire-wound resistors (no brand ID), and connected them to a DE-5000 LCR meter.  The meter said resistance was 8.3 ohms (within spec).  At 1k, 10k and 100k Hz, inductance was 2.1 uH.  At 100 and 120 Hz, readings were unstable, but around 1-2 uH.  So, if my arithmetic is right, impedances are 8.3, 8.3 and 8.4 ohms at 1k, 10k, and 100k Hz respectively, which agrees with conclusion of the youtube video.

Anything over 10 kHz is a waste for me, and most folks aren't going to hear much over 15 kHz unless they're pretty young.  If ordinary 8 ohm wire-wound resistors have such a low reactance in the audio frequency range, I don't see much point in paying extra for non-inductive resistors.  Any that I bought would need to be checked though.  Anybody encounter 8 ohm wire-wounds with high inductance?  It should be possible to make them, and it might be an interesting project to design and make one of high wattage.  I would guess most resistor manufacturers try to minimize inductance within pricing constraints.

Also tried the highest resistance wire-wound that I have, which is 25k ohms, 10 W, WEP brand (Workman Electronic Products is still around!).  DE-5000 said resistance is 24k (within spec).  I expected the inductance to be higher than for the 8.2 ohm resistors, and it was.   At 1k, 10k and 100k Hz, inductance was 4.1 mH.  At 100 and 120 Hz, readings were unstable, but around 2-4 mH.  Inductance didn't scale that far off being proportional to resistance.  Resistance was about 3000 times higher, and inductance was about 2000 times higher.  25k resistor probably has higher resistivity wire as well as more coils.  So, impedances are 24.0k, 24.0k and 24.1k ohms at 1k, 10k, and 100k Hz respectively.  Again, not so different from DC resistance in the audio range although not a useful resistance for testing audio amps.

  Mike in California

[edpalmer42]

Have you ever looked at the impedance curve of a speaker?  It bears a strong resemblence to a mountain range!  You have to use a really broad brush to assign a single value for the speaker impedance.  The value depends on the speaker, the crossover, and the enclosure.  As a result, the inductive effects of a load resistor are not significant.  Even if you can measure the effects, they will be negligible compared to the effects of the speaker itself.

Ed

[calexanian]

Generally you are fine. As you pointed out sometimes the actually approximate the impedance of a speaker voice coil less the dynamic movement and resonant frequency. Generally your larger wire wound resistors have allow enough impedance that it does not really matter for audio, but if you are doing some real measurements you defiantly want to invest in non inductive types for a reference.

[dom0]

An audio amplifier that is unstable with the inductance of any small power resistor (50, maybe 100 W?) will certainly not be stable with any speaker in the real world.

[T3sl4co1l]

In my experience, nichrome resistors (that aren't made intentionally badly, i.e. a maximal inductance construction) have a crossover frequency around a MHz, give or take.  This isn't quite a material property (it depends on geometry), but it's proportional to it: conductivity is in units of S/m, while permeability is in units of H/m.  The product of these units is s/m^2, an odd thing, but I think that might be saying something like, the time constant per total winding area (i.e., solenoid cross sectional area * turns, give or take a geometry factor because the turns aren't perfectly coupled).

Which is why copper has a low crossover frequency (typically ~10kHz), making even short lengths of copper a poor choice for low value shunt resistors, when any kind of AC is involved.  Whereas copper inductors are quite good at modest frequencies (if Q ~ 1 at ~10kHz, then Q ~ 100 at ~1MHz?).

I suppose one should expect noninductive types to be maybe 1-10 times higher in crossover frequency.  The 1 includes low-ohm types that are straight (axial) wires embedded in cement; there's no such thing as a noninductive construction in that case, so don't waste your money.    Better construction (Ayerton-Perry) non-inductives probably have an even better ratio than 10, but the winding is denser (high resistance --> high inductance) to begin with, so the construction has "more to fix", so to speak, than usual.

Bifilar non-inductive construction isn't very good, because at best, it makes a parallel pair transmission line, which still has inductance between the leads.  That said, this can be made quite good if the characteristic impedance is matched to the DC value, plus something or other about termination at the shorted end.  It's noteworthy that the thermal distribution will be poor at higher frequencies, because the resistor acts like a length of really lossy transmission line: the AC goes in, decaying exponentially along the way, roughly in terms of wavelengths.

Thick film (ceramic substrate, heatsink required) types are generally quite good, and can be made out into the GHz (of course, those fancy ones you have to pay for).  The main downside for common types (e.g., TO-220 style) are lead inductance and unspecified package capacitance, which may lead to worse behavior in the 100MHz+ range.

So... ahem... for audio applications, yeah, don't worry about it.

Tim

[suicidaleggroll]

A speaker is, essentially, just a big inductor (that's moving through a magnetic field).  The tiny bit of inductance in a wire-wound dummy load will be insignificant compared to the inductance in any speaker you would actually be powering with the amp, so it's basically a non-issue.

[Circlotron]

Probably the one good thing about a purely resistive audio dummy load is that yours can be identical to Fred-next-door's dummy load so should give equal test results. Two different loudspeakers as loads are about as similar as two different people.

[calzap]

Thanks for the replies; it's been educational.  I have a very nice 250 W Celwave dummy load I use for RF that's good from DC to 2 GHz.  Problem is it's 50 ohms.  Resistor is made by EMC Tech and is BeO or AlN, not sure which.  I suppose I could buy 5 more and put them all in parallel to come close to 8 ohms!  But, I ordered a couple of plain 100 W, 8 ohm wire-wounds.    Should be good enough to test my 15 W/channel TDA7297 modules.

Mike in California

[DrGeoff]

The DC resistance of the voice coil for an 8 ohm speaker is usually around the 6-7 ohm value (a 6.8 ohm resistor is a good start). However as pointed out, unless you have applied a compensation network to the speaker voice coil, the impedance changes with frequency.

[GK]

Regardless of the need or not for a purely resistive load, it's pretty trivial to compensate an inductive R with a series R-C in parallel.

For example:
8 ohm resistive load with 10uH of inductance.

fc = R / (2.pi.L) = 8 / (2.pi.10uH) = 127.324 kHz.

So therefore make Ccomp = 1 / (2.pi.fc.Rcomp) where Rcomp = Rload.

1 / ( 2.pi.127324.8 ) = 156.25nH.

Rcomp of course needs to be non-inductive and beware that it may need to dissipate a few watts at 20 kHz if driven from a high power amplifier with a high voltage output swing.

I have a 1kW 50 ohm RF dummy load compensated this way that is made of jug elements and is resistive to the extents of the HF band.

[dom0]

This technique even has a name. Well, two, actually ; Zobel networks (the general case) and Boucherot cell (the inductive compensation case).

[LukeW]

Why would you want a realistic test dummy load for an audio amplifier to be non-inductive?

Are your loudspeaker coils non-inductive?

[Circlotron]

^^ Does your loudspeaker have the same inductance as mine. No.
Does your 8 ohm non inductive load have the same resistance as mine. Yes.

It may not be a real world test but it is useful for comparisons. Sort of like rolling road fuel consumption tests for new cars. Only partially connected to reality but very easy for many different test labs to reproduce and therefore compare accurately.