Single sheet TH challenge

[60ndown]

ok thanks again, im a big fan of the KISS principle

[David McBean]

Originally Posted by Wayne Parham

The reason for the difference is mostly due to one curve being generated by a constant voltage source and the other being driven with constant power. Remember that when we measure, we're dealing with amplifiers that are essentially constant voltage sources.

Hi Wayne,

Exactly - and because of this, the result is only going to be an approximation to the actual 1 watt / 1 metre performance - which is the point I was trying to make in my example :-).

Since the measured result is only going to be an approximation at best, why not just use the published rated impedance of the driver, rather than going to the trouble of finding a value of Zmin for the system? The difference between Re and the magnitude of Zmin will not be much. Using Re rather than Zmin would mean that it is no longer necessary to identify the passband in order to establish a value for Zmin.

Kind regards,

David

[Wayne Parham]

Originally Posted by David McBean

Exactly - and because of this, the result is only going to be an approximation to the actual 1 watt / 1 metre performance - which is the point I was trying to make in my example :-)

Yes, that's the same point I was making. So we're square on that.

Originally Posted by David McBean

Since the measured result is only going to be an approximation at best, why not just use the published rated impedance of the driver, rather than going to the trouble of finding a value of Zmin for the system? The difference between Re and the magnitude of Zmin will not be much. Using Re rather than Zmin would mean that it is no longer necessary to identify the passband in order to establish a value for Zmin.

Not sure I understand what you are saying. The measured result is an approximation? Actually, the measured result is the real deal. The simulation is what is the approximation. I mean, I think the model is very good, but it is a model, after all.

But now that I consider your meaning a little more, I don't think your statement was about comparing the measurement to the model, I think you were talking about what single-point impedance value to chose when calculating drive voltage for measurements. If so, then, yes, that's the problem. That is the dilemma.

If we wish to make SPL measurements of loudspeakers at consistent power levels, then we must understand their impedances to calculate their drive voltage requirements.

I agree with you that Re is perfectly acceptable. Zmin or Re. Both are pretty close to the same, and either one is what I would use. I have been inclined to use the value of Zmin in the passband, usually the first impedance minimum after the first peak. But it is, in fact, usually almost exactly the same value as DC resistance. So I'm perfectly satisfied with using either one.

To further the thought experiment, for others that might consider this dilemma:

We can't just say constant power must be used, because this side-steps reality. The amplifiers used to drive these speakers are essentially constant-voltage sources. You can argue whether or not they should be constant-current sources, but that's another matter. The fact is, they are constant voltage sources.

So that being said, if we want to compare measurements between speakers, if we want to use uniform power values as reference levels, e.g. 1W, 100W, etc, then we have to understand impedance. As has been rightly shown, impedance fluxuates across the passband, both in amplitude and in phase (resistance/reactance).

So what do we do? Do we measure and use Zavg? Do we use Re or Zmin? Or do we use the manufacturer's advertised impedance and call it good?

I don't think using advertised impedance is appropriate, since it can be pretty far off. Most times, manufacturers round to the nearest multiple of 4, 8 or 16 ohms. Seems pretty common to find speakers that are advertised as 8 ohm speakers but are actually 6 ohms, for example. One can usually find accurate values in the datasheets, and so I would be comfortable using the manufacturers specs in most cases, just not the advertised (4/8/16Ω) value. But really, even if I trust spec sheets, measurements are there to verify, so I think it only makes sense to run an impedance sweep and use the values found.

So throwing out the advertised impedance, that leaves me with two choices, Zavg or Zmin. Honestly, in my opinion, both Zavg and Zmin are just fine. When you run the numbers, you find a worse case variance of about 1dB between Zavg and Zmin. It's usually only about 0.25dB to 0.5dB. But Zavg is not obvious and requires some calculation, whereas Zmin is dead simple and completely obvious. That tends to lean my decision towards Zmin.

Zavg is a little bit higher than Zmin due to the influence of the impedance peaks. And to find it, you have to define an upper edge of the passband, since this will determine the area under the curve. But using Zmin is not subject to interpretation. It's repeatable and easy for anyone to find in an instant. So since using Zmin gives results that are less than a decibel off the Zavg values, and since it tends towards the conservative where SPL is concerned, I have always found it to be the most attractive method for setting drive voltage.

[CRESCENDO]

If people are still looking for 3015LFs, there are some brand new ones in the swap meet section.

Carry on

[David McBean]

Hi Wayne,

Originally Posted by Wayne Parham

Not sure I understand what you are saying. The measured result is an approximation? Actually, the measured result is the real deal.

Sorry for the ambiguity :-).

I was just making the observation that the measured result gives the true 1 watt / 1 metre response at a single frequency only, not across the entire bandwidth.

Originally Posted by Wayne Parham

I agree with you that Re is perfectly acceptable. Zmin or Re. Both are pretty close to the same, and either one is what I would use. I have been inclined to use the value of Zmin in the passband, usually the first impedance minimum after the first peak. But it is, in fact, usually almost exactly the same value as DC resistance. So I'm perfectly satisfied with using either one.

As far as Hornresp is concerned, the attraction of using Re rather than Zmin is that it no longer becomes necessary to identify an appropriate passband for a particular speaker system, or to find the value of Zmin within that passband.

This makes it much easier to provide the sensitivity tool requested by Djim.

Kind regards,

David

[djk]

I'm sorry, but this is not worth the time to read.

I can set the drive voltage in any of the three simulation programs that I use, and look at the impedance minimums too.

That's all that matters in my mind.

[Wayne Parham]

Originally Posted by David McBean

As far as Hornresp is concerned, the attraction of using Re rather than Zmin is that it no longer becomes necessary to identify an appropriate passband for a particular speaker system, or to find the value of Zmin within that passband.

This makes it much easier to provide the sensitivity tool requested by Djim.

I think Re is fine, since it is almost always very close to Zmin.

On the other hand, my definition of Zmin being the minimum impedance above cutoff is equally determinite, and has the advantage of being something that all agree is in the passband. You'll find people that complain about DC resistance not being "accurate", since the speaker never has DC applied.

You do not have to define the passband to find Zmin. You only have to know one side of the passband - lower cutoff - which can be simplified to be the 1/4λ frequency. This is a useful simplification since Zmin will always be above that point. The other side - the mass-rolloff point - is not important for this determination. Horns are often used above this but impedance is rising, at least when driven with a speaker having a voice coil. So you can use an arbitrarily high frequency for the upper bounds. Use 20kHz if you want. The point is to find the minimum impedance above lower cutoff.

I think we're definitely splitting hairs here though. I'm sure you could use either Re or Zmin and get the same results to several significant digits.

[AndrewT]

Originally Posted by AndrewT

How do we carry out a sensitivity measurement of a real speaker?

Do we supply a band limited pink noise, or white noise, or some other defined noise, to a speaker and measure the voltage and the current at the speaker terminals and the SPL at the test distance?

Originally Posted by Wayne Parham

..............if we want to compare measurements between speakers, if we want to use uniform power values as reference levels, e.g. 1W, 100W, etc, then we have to understand impedance. As has been rightly shown, impedance fluxuates across the passband, both in amplitude and in phase (resistance/reactance).

So what do we do? Do we measure and use Zavg?

I asked a while back and now I see Wayne asking the same question.
How do we measure the power sensitivity of an actual speaker?

[Wayne Parham]

Actually, that was a rhetorical question. This is how I do it:

• Test Plan

[Djim]

Hi Andrew,

In the industry they usually measure the current. This allows changes in impedance at different power stages.