Ivan Beaver wrote on Thu, 02 August 2007 07:20 |
The impedance issue is where it really gets to be picky/touchy. If you do an impedance sweep, then what is the actual impedance? WHo is going to determine that? It is generally considered NOT to be the lowest point on a curve, but rather a "average" that relates to a standard "number". In just about every loudspeaker, there are more points that are much higher in impedance than those that are lower.
|
The impedance sweep is quick and easy to do. I see no reason to omit it. At least then, people can see the exact impedance curve rather than depending on an advertised impedance value.
Many manufacturers have already expressed concern that without impedance being taken into consideration, their loudspeakers SPL appears to be lower than others.
We chose minimum impedance because it is a fixed tangible value. Average impedance is subject to interpretation. Minimum impedance works very well for setting power reference levels to compare SPL between speakers.
Ivan Beaver wrote on Thu, 02 August 2007 07:20 |
I used to be in the other camp and was wanting to simply put 1 watt in(changing voltage with different impedance), but started to realize all of the issues this brings up and am now in the camp of 2.83V and figure out the wattage that was applied from there. After all the wattage applied is different for all points on the curve and is only valid for one (and equal points of impedance).
|
I agree with you about measuring using a fixed voltage. In the technical sense, all measurements are made with a fixed voltage anyway. Amps aren't contant-power sources. But the point is, it's nice to gather data that reflects SPL at different reference levels, and many people are used to seeing power level references.
We measured using 28.3v, which from 10 meters away gives the same SPL as 2.83v at 1 meter. We also measured using 100 watts at 10 meters. This gave the same SPL as 1W/1M. After that, we doubled power and ran another sweep, continuing to do this until we reached maximum power. I kind of like that approach. You get both voltage reference charts and power reference charts.
Ivan Beaver wrote on Thu, 02 August 2007 07:20 |
I do not know of any manufacturer (there might be some) that uses the lowest point on the curve as the impedance number. So if you do that, then somehow you are saying that you are "better" than the rest of the industry and want to do things differently than currently accepted. That is very open to discussion.
|
Most manufacturers publish an advertised impedance, which is the closest multiple of 4 or 8 ohms. It's almost an arbitrary number.
Ivan Beaver wrote on Thu, 02 August 2007 07:20 |
SO the nubmer you choose is now a determination by who?
|
Exactly. That's why we chose minimum impedance. Minimum impedance isn't a value that requires interpretation or calculation.
Average impedance is always higher than minimum impedance. Using minimum impedance to calculate power, one sends a smaller voltage to the speaker than they would if they calculated using average impedance. The signal presented to the speaker then causes it to dissipate the expected power level at frequencies where impedance is minimum. At other frequencies, where impedance is higher, power dissipation is lower. So the measurement reflects a conservative SPL. It is, however, uniformly done across all speakers without interpretation.
The only way to accurately find average impedance it to take the measured impedance chart and find the area under the curve. But even this is subject to interpretation because you have to choose the frequency range to use. If you use DC to 100Hz, average impedance will be lower than if you choose 20Hz to 20kHz. At lower frequencies, reactive impedance from voice coil inductance isn't significant but at high frequencies, it dominates. So no matter what you do, average impedance is an interpretation.
Sure, you can say the range chosen should be the intended bandwidth of the device. But even that gives a little "wiggle room" because the band chosen for determining impedance can be shifted down, say from 30Hz-150Hz to DC-150Hz. This would give a lower average impedance, even if precisely calculated using the area under the curve. A small horn would be used over a different frequency range than a larger horn, so do you set a fixed band to calculate average impedance? Or do you let someone pick it? This leaves room for interpretation, which gives some ambiguity. It lets the person driving make choices that can offset the results, even if average impedance is precisely calculated.
That's why I suggest minimum impedance be used. If someone else has a suggestion for a method that can be accurately and uniformly applied, I'd love to hear it.
Ivan Beaver wrote on Thu, 02 August 2007 07:20 |
And to further complicate it, you now have to reajdust the sending level to change the output voltage. So now you have adjusted something. It is all of these little adjustments that start to be the "problem", and start to bring up question regarding how the data was collected.
|
To measure SPL at a fixed power level across various loads, you have to adjust voltage to account for impedance, that's right. But if you don't, then you have to massage the data to come up with the same thing. I think it is much better to have the exhibitor right there, watching the dial and seeing the voltage level set for his speaker before the test is run. He is then comfortable that his speaker has received the signal appropriate for the power tested for. Then when he sees the charts published, he can be comfortable that his speaker got a fair test.
Ivan Beaver wrote on Thu, 02 August 2007 07:20 |
WIth the TEF you can change the scales with post processing to account for different wattages/voltages so you can do a comparisom anyway you want. That is what Tom did after the NY shootout, but people were generally not interested ins seeing that, partly because they did not trust him, but I will tell you, Tom is one of the most anal people in this industry regarding real specs, and what he does can be trusted totally. He was not trying to play any games, but just rather present data in a way that would be easier for some to understand.
|
I don't think we should post-process because of the trust factor.
To adjust a response chart to account for impedance differences, you still have to calculate using an impedance figure. This figure must be interpreted at some point, be it minimum impedance or average impedance. Then the data has to be massaged to account for the difference. I personally would prefer that the impedance be measured rather than using advertised impedance or some other arbitrary figure to calculate an offset and then later massage the data.
I would welcome looking at other approaches to calculate power, but I don't think we should avoid power-reference measurements, choosing to make
only voltage-reference measurements instead. I think we should do both.
Ivan Beaver wrote on Thu, 02 August 2007 07:20 |
Is it better to have someone who has no relation to a manufacturer and has little experience in doing all of these types of measurements or someone who does this all the time and is aware of all the issues/problems that arrise and can head them off ahead of time and ensure accurate data?
|
Naturally, it would be best if we had a qualified, independent and unaffiliated third-party. If that's not possible, we should self-police. But if we are going to self-police, we must be extra careful to make everyone comfortable.
Ivan Beaver wrote on Thu, 02 August 2007 07:20 |
If some people don't trust others to do measurements with someone looking over their shoulder. then I think they personally have a problem.
|
Absolutely. The measurement process and data should be open for inspection.
That's how we did it the past two years. I did the test plan and measurements in 2005, and everyone watched. When I set the voltage levels for the fixed-power measurements, I showed the exhibitor being measured the impedance chart, picked the minimum impedance, calculated voltage required to set power and showed the exhibitor that. Then turned the LMS oscillator on and set voltage, showing the exhibitor the meter. Everyone watched, the whole thing was open and I think everyone felt comfortable with the process.
David Lee did the measurements in 2006 and he basically did the same thing, but with a different system (Praxis) which had a different setup.
I am definitely open to suggestions and don't think I should be the final decision-maker, just like I don't think you or Mark or Tom should be. I think everyone would be better served with a concensus. So I definitely appreciate your input.
I think we agree in most things, but there are a few I think we might still need to hash through. Fixed-power measurements and impedance determination are two of them. Please look through my comments above and tell me what you think.