Know-How

Equipment used:

  • Dell Studio 17 laptop with XP Pro 32bit
  • DAW desktop with Win pro 64 bit
  • 2 ND9 spl calibrators
  • Quest QC-10 SPL calibrator
  • ACO Pacific 511E SPL calibrator (#011010)with Q1,Q3 & Q7 adapters
  • matched pair of Earthworks M30 measurement microphones with windscreens (#1764 & 1766)
  • ACO Pacific:
    • PS9200 power supply with 50volt modification (#9818)
    • 4012 preamp (#98187-3)
    • 7012 mic capsule (#21720)
    • 7047 mic capsule, G.R.A.S. 40AM mic capsule
    • WS-1 windscreen
    • 4212 preamp
    • 4048 preamp
    • PSIEPE-4
  • Smith & Larson Woofer Tester 2
  • Presonus Firebox and Firestudio 2626 firewire microphone preamp interfaces
  • Behringer DCX2496
  • Symetrix 511
  • Crest Pro 8002 amplifiers
  • Powersoft Digam K10 (Klein & Hummel KPA-2400) amplifiers
  • Zod Audio 24HP-DSP amplifiers
  • Ideal Industries 61-702 DMM
  • LinearX VI Box.

Software:

  • ARTA
  • STEPS
  • LIMP
  • REW v5
  • Spectrum Lab
  • RplusD
  • NCH tone generator
  • CEA2010
  • S&L Woofertester2

The Firebox without compensation is +/-0.5db between 10hz and 20khz. The PS9200 powersupply and 4012 pre-amp are +/-0.5db between 10hz and 20khz. The M30 and ACO 7012 microphones are within +/-0.5db tolerance from 10hz to 20khz or better. The Behringer DCX2496 is measured flat within +/-0.5db from 7-19khz. The Crest Pro8002 amplifiers used are flat within +/-0.5db from 10hz-17.5khz. The Powersoft Digam K10/Klein & Hummel KPA-2400 amplifiers measure +/-0.8db 10-22khz, The Zod Audio 24HP-DSP amplifiers are +/-0.5db 10-20khz or better. The total signal chain should be within +/-0.5db from 10 to 17.5khz. This is good enough for reasonable accuracy in the primary range of interest for subwoofers, 10-200hz.

Measurements:

For drivers the tests start with a measurement of the drivers small signal parameters (TSP's) and generation of a free air impedance plot using LIMP software or occasionally a Smith and Larson Woofer tester 2. The drivers get a brief breaking in with pink noise and sine waves in free air, while doing a check for proper driver function prior to the measurements. Afterward another impedance measurement is taken with the driver/s loaded into the enclosure to be tested. The parameters collected will often vary somewhat from the manufacturers specs. The results depend heavily on the test methods and setup used. Additionally after discussion with several very well known and respected driver manufacturers a variation of up to 25% in some parameters is considered acceptable. Also many parameters are easily affected by the environmental conditions, most notably the suspension stiffness. The added mass method is used to lower the FS of the system the required amount to calculate the parameters. This is accomplished using 6.5 gram ceramic magnets clamped through the driver cone. This method is much more accurate in our experience than the known air volume "test box" method, but is still prone to some error. Ideally mms of the driver would be directly measured by weighing the assembly but this requires access to the parts themselves or cutting up a driver for this purpose. Typically this is not feasible so the added mass method is used. The impedance curve is often overlooked by many in favor of other more "exciting measurements, but the impedance measurement along with a few other facts about the driver or system can often give a lot of information and insight into how the system will behave. For example when looking at large subwoofer drivers the "Q" and magnitude of the impedance peak can offer insight into how strong the motor system is, the placement of the peak/s can indicate the FS of the driver or the tuning of the driver/enclosure system, the damping of the system can be hinted at and small anomolies in the upper part of the impedance curve can indicate resonances or vibration issues in the cone, frame, enclosure or port. With large subwoofer drivers if the impedance rapidly rises back up after the impedance peak this typically indicates a large amount of inductance in the motor system and a rolled of upper bass spectrum with diminished sensitivity.

Then on to the outdoor ground plane testing. The systems are placed on the ground out in a large field with the nearest large objects being 60ft or more away and the microphone is placed pointing at the subwoofer at a distance of 2 meters from the nearest enclosure edge or face. In this manner the effects of any boundarys or room acoustics can be eliminated leaving the response of the speaker minus the influence of any strong reflections. In a typical room the interaction between the boundarys and the radiation of the system being tested will serve to completely obscure the raw response of the unit and measurements taken in such an environment could only be compared to other measurements taken in this same space. Subs having multiple radiation sources will be placed so that the driver is closest to the microphone or equidistant in the case of multiple drivers and any ports, pr's or other auxillary radiators will be placed as close to the microphone as the cabinet geometry allows. All measurements are taken at a distance of 2 meters from the DUT unless otherwise specifically noted for the measurement. Systems having multiple radiation points or having drivers firing from 2 or 3 opposing enclosure faces will have measurements taken at a distance of 10m's and 2m's and used to develop a compensation file in order to properly compare the output power of these systems with conventional direct forward radiating systems. Outdoor groundplane measurements are what is considered to be half-space and impart 6dB of gain from the single boundary's reinforcement when compared to an anechoic free-space measurement taken from the same distance. Increasing the microphone distance from 1 meters back to 2 meters conveniently reduces the SPL by 6dB back to a level equivalent to a 1m anechoic or free-space measurement. It also has the added effect of allowing all of the radiation from a system that may have multiple radiation points to blend better at the microphone. With many larger sub systems the enclosure itself is large enough and the distance between radiation elements is large enough that the system itself is providing some loading to the output of the system delivered to a microphone at only 1m distance. Effectively it could still be in a quasi nearfield environment. Also the distance between radiation points on the system may be large enough to account for a significant percentage of the 1m distance which will misrepresent the total output makeup of the system by exaggeratting the output of the closest radiator and underrepresenting the output of those further away relative to each other. Some extremely large systems can still exhibit some effect at 2 meters. For these reasons the standard distance used is 2 meters and not 1m as it represents too many compromises for larger systems.

Test Procedure

1. A raw 1 watt (nominal fixed voltage)/1 meter response measurement with no signal shaping. This test is only performed on passive systems. (Note: Originally this test involved 1w maximum input power calculated using the impedance measurement taken of the driver/s in the enclosure. The minimum impedance measured in the subs intended pass band was used to calculate what voltage input will generate 1w of power into the load at the minimum impedance point. This limits the DUT to receiving no more than 1 watt of power at any point in the intended frequency range of use. Other methods of using a fixed voltage such as 2.83v without regard for the enclosures complex impedance profile can lead to apparent advantages for lower impedance cabinets as they will in reality be receiving higher power during the measurement but does not account for the fact that the maximum voltage available from an amplifier into the lower impedance is usually less. The method used here we believe represents a better indicator of true efficiency and some indication of how difficult of a load the speaker will present to an amplifier. However this resulted in odd voltages being used like 1.8v, 3v, etc...After being asked multiple times to standardize the voltages it was decided to settle on voltages of 1 volt for 1ohm systems, 1.41 volts for 2 ohm, 2 volts for 4 ohm, 2.83 volts for 8ohm and 4 volts for 16ohm systems. The impedance of the systems is still determined by looking at the minimum impedance measured between 10 and 200hz. Whichever range the minimum impedance of the system is closer to dictates what type of impedance it will be considered. For example a system with a Z min of 1.3 ohms would be considered a 1ohm system not a 2ohm system. A system with a zmin of 6 ohms would recieve 2.83 volts and be considered an 8 ohm system, but one with a Z min of 4.8 ohms would be considered a 4 ohm system. In this way it prevents inflated sensitivity ratings by testing 2 ohm systems with a full 2.83 volts for example. Obviously this is not perfect but by looking at the measured data and the voltage supplied one can calculate what the system would measure at a different voltage input anyway.) The results of this type of measurement give insight into how loud the system will be with a specified input voltage and in conjunction with the impedance measurement can give an idea of system efficiency and how heavy of a load the speaker is on the amplifier driving it. Bear in mind that this does not indicate necessarily which system will be louder as that also depends on compression effects, thermal heating and driver excursion limits. It does however allow you to calculate the maximum potential output with a specific amount of amplifier voltage.

2. A raw 100 watt nominal(10x the 1 meter voltage) / 10 meter response measurement with no signal shaping. The voltage is increased by 10X over what was used for the 1w/1m sensitivity test and the microphone is moved back to a 10 meter distance. This is to get a look at the response shape at a greater distance where the effects of a large baffle or enclosure, or widely spaced radiation points on the system response should be mitigated and the power is increased to a higher level, which could start indicating some compression or variation on the FR in some less robust systems. This also gives a glimpse at far field performance if the speaker is intended for use in large spaces at high volume and allows for far better integration of the output of systems having multiple radiators. This test is also only performed on passive systems. This measurement should be examined just as the 1 meter sensitivity measurement and in fact comparison of the two measurements from different distances is often helpful to gauge nearfield effects from the proximity of the DUT to the microphone at a 1m distance. In general the 10 meter higher power sensitivity measurement is found more useful and accurate to describe the system behavior because of this.

3. Basic frequency response shape. With powered or complete active subwoofer systems the response will be measured with a variety of different settings in order to gauge the effectiveness of the on board controls. Typically this would involve various settings of the low pass filter, a few different settings of any internal EQ bands available, all preset response curves available, damping settings, different modes of operation, plugged ports, sealed/vented, etc...A combination of settings will then be selected as the "basic response" to be used for the rest of the measurements. Typically the subwoofer will be configured for whatever response shape is widest bandwidth at the top of its frequency range and for whatever deep bass response appears to be most natural and unequalized without heavy low frequency boost. This is done because EQ eats up amplifier quickly and if done in the deep bass as is common it will also eat up driver excursion quickly. This will cause the maximum output measurements to be limited very early, by the deepest bass, wherever the EQ boost is heaviest and this will result in the testing being stopped before the unit is pushed very hard in the upper parts of it's bandwidth. Ideally we want to see what the DUT is capable of full bandwidth and not just limited to a small frequency band. Once these settings are settled on the basic response of the DUT as it will be set for all future measurements will be captured over a 10-200Hz bandwidth. In most cases an extended range frequency response was measured to see the response well past 200hz as well. Active systems typically are filtered not much past 300hz or so, but passive systems will be run without any top end limiting so they will be measured up higher in frequency to get an idea of the drivers potential capabilities and issues in the midbass. Passive units will be measured sans any processing at all and will also be measured with a suitable protective high pass filter applied if it is required. If so both responses will be shown for comparative purposes.

4. Impedance measurement. (For passive systems only.) LIMP software is used to measure the impedance of the system (A voltage of about 1 to 2 volts is used typically.). This can then be compared against the raw driver impedance measurements and the impedance measurement of other drivers tested in the same enclosure. The impedance measurement is very useful for identifying resonances or vibrations in the system, gauging how close the real system is to any simulated performance and calculating how much actual current or power is being applied at a certain voltage input. It can also help identify the tuning of helmholtz resonator systems and the Fb of horns.

5. Long term output compression sweeps. After setting the DUT's output level to correspond to an output of 90dB at 50Hz at a 2m distance using a 50hz sine wave signal, the response is measured using an ascending 0-120hz sine wave sweep of 24 second duration. This is the base measurement and output level which output compression is gauged against. Then measurement sweeps are taken back to back while the input signal is progressively increased by an amount that should cause the DUT to produce 5dB greater output each time up to the point where the DUT starts compressing the output in an obvious manner by 3dB or more or starts to emit distress noises. Typically the final loudest sweep measurement will only be increased by 2 or 3 dB as the DUT is already showing signs of reaching its limits and a further 5dB increase in output is simply too much to ask. After the loudest measurement is completed the input signal will be adjusted back down to the starting 90dB at 50Hz base level and recaptured and compared to the first "cold" 90dB measurement. This shows the effects of long term heating in the driver voice coil/s and motor/s. 2 types of graphs are generated from this test set. The basic overlay of each sweep which will show how the response changes with each increase in level and another graph which shows only the amount of compression that is occuring in the systems output referenced to the initial 90dB "cold" measurement. Ideally what you are looking for is the least amount of compression throughout the entire bandwidth at the highest sweep levels.

6. THD and distortion by component measurements are captured using long duration 256K, 8000z SR sine sweeps generated by ARTA software, starting at a level corresponding with the lowest output level used in the power compression tests (90dB @ 50Hz @ 2 meters) and is increased by increments of 5db for each measurement from there mirroring the levels used during the long term output compression sweeps. The relatively high level of background noise present at the test site has some effect on the lower volume measurements accuracy especially in the deep bass below 20Hz where most systems will have output well down into the background noise floor at the more modest drive levels. Distortion is usually low enough at lower volume outputs to not worry about to begin with, so the data for only the 3 or 4 highest output levels is presented for most systems. When considering distortion of the signal by the loudspeaker obviously the lower the amount the better but there are other considerations to look for such as the harmonic make-up. Generally a good result is a system that has low distortion even at very large outputs and has a distortion profile that does not have abrupt changes in distortion magnitude as these typically indicate other things about the system such as a severe resonance and that it may be more audible. Also when considering the harmonic makeup of the distortion the general guideline is that the lower order and even order harmonics are less offensive than the odd and higher order harmonics. Odd order harmonics do not blend in with the fundamental in a pleasing manner like even order harmonics do so they tend to stand out more. For example a sub with 12% 2nd harmonic distortion probably won't sound bad at all, but a sub with 12% 5th harmonic distortion would be much more obviously heavily distorted to your ear and "wrong" or "overdriven" sounding. 2nd harmonic distortion can be quite high indeed and the bass system can still sound perfectly acceptable or even good. Large amounts of bass frequency H2 are rather hard to hear with the rest of the frequency bandwidth included. The 3rd harmonic or H3 is higher in frequency and is an odd order harmonic so it is a little more offensive but it still takes quite a large amount of H3 distortion to make the bass sound obviously distorted or bad. The 4th harmonic is a little more easy to notice since it is now typically going to fall well outside of the bass range and sound like something your bass system should not be producing, but typically 4th harmonic distortion in bass systems is typically well down in level compared to the 2nd and 3rd harmonics which dominate usually. Once you get to the 5th harmonic and higher it takes much less of a percentage of the output to become audible. Thankfully most bass systems do not output large amounts of harmonics that high up in band unless they are severely overdriven or have other issues.

7. Spectral contamination was tested at various drive levels using a 10 tone equally spaced sine excitation centered between 20 and 77hz to test the amount of self noise each DUT generates. This type of test is very rough on the system as the driver and amplifier both run into overload much quicker than would be expected. The base level was 90db total and the level was increased by a nominal 5db until there were obvious overload or distress noises. Ideally what you would see with this test is nothing other than the 10 sine tones used indicating a very clean output. Distortion and spurious noises would start to show up in between the center frequencies of this test on the resulting graph. A cleaner more powerful result would have sharper defined center frequencies and deeper notches in between the center frequencies at higher output levels indicating a much cleaner more accurate reproduction of a very tough signal. (This test has been discontinued. It was difficult to setup, very rough on the systems at louder output levels and more importantly it did not seem to give much useful information that could not be gleaned from the other standard output and distortion tests already.)

8. Maximum long term output is derived from the highest sweep level attained during the long term output compression testing. This is the highest long term sustained level attainable by the DUT within it's entire intended bandwidth, at which point some limiting condition whether amplifier, or excessive: excursion, noise, compression or distortion is in evidence in some part or all of the frequency range. (Note this does not mean that the subwoofer will handle a 100% duty cycle sine wave at this power level indefinitely! It will not!) The long term output compression test is meant to simulate the effects of high duty cycle use on the system but in a much more rapid manner. The long duration sine wave sweeps at the highest power levels are brutal on the systems and are meant to simulate many hours of playback of typical material with a MUCH lower duty cycle.Sine waves are 100% duty cycle so the voice coil gets heated relentlessly and in fact this test type kills more drivers than any other. Heavily compressed electronic music is only 25% duty cycle at mostand typical music is far less than that so the average power is far less than that of the test signals used here.Therefore the voice coil and amplifier have some "downtime" to recharge and cool off a bit. Don't think of this measurement as output that can be sustained indefinitely...Think of it as a survival point...The sine wave test signals may generate and average power of 1500w from the amplifier into a system. Actual music or movie content that produces peak outputs of the same SPL levels might require the same amount of peak power but the average power seen over the same time period is more likely to be 200 watts than anywhere near the 1500 of the sine wave.

9. CEA2010 maximum short term distortion limited output. Maximum RMS short term output is measured at 2m ground plane. This test involved short 6.5cycle duration shaped sine bursts centered at 1/3rd octave intervals of 20, 25, 31.5, 40, 50 and 63hz. Additionally further test frequencies have been added at both ends of the spectrum (10, 12.5, 16, 80, 100 and 120hz) to better represent more of the full bass bandwidth as long as the systems were capable of reproducing meaningful output in those bands. The DUT's output level is then increased while the distortion is monitored until either the DUT stops gaining in output level, obvious signs of distress are encountered, or the prescribed stair step distortion threshold for any harmonic is exceeded. The distortion thresholds get lower at higher harmonics of the fundamental test frequency. For example the 2nd harmonic isallowed to reach -10dB from the fundamental before failing the test but the 3rd harmonic is only allowed to reach -15dB. By the final band if any of the harmonics or noise reach -40dB it causes a fail. The stair stepping down of the allowable percentage of distortion is supposed to correspond with our subjective acuteness to the presence of various harmonic distortions. In general we are much more sensitive to higher harmonics so they are limited much more tightly while the relatively benign 2nd and 3rd harmonics can be allowed to grow rather high before becoming offensive. In theory this works well but it is a complex subject. For instance some systems may be failing for a harmonic that is -35dB down in the final bandwidth while still sounding acceptble to the ear while another may sound obviously distorted with a lot of H2 H3 and H4 distortion yet it passes because of the stair step thresholds. Also two subwoofers may produce a very similar output of say 110dB at 31.5Hz and both may pass but one may sound utterly clean with mostly 2nd harmonic distortion while the other may have higher 3Rd and 5th harmonic distortion but sound a lot more audibly distorted despite having lower total THD. Also many systems may sound ok while failing the distortion thresholds while another sounds very bad and overdriven while passing. This is because when driven to their limits some units exhibit otehr overload noises or bad sounds that are not harmonically related to the signal input into the subwoofer. These types of sounds are far more audible and offensive than any harmonic related output as these rattles, buzzes, clacks, pops and other artifacts do not blend in with the sound at all and are obviously non-harmonically related to the output at all and typically are a wider bandwidth mechanical sounding noise that is higher in frequency. The CEA-2010 program will sometimes pick these up but not always. Again CEA-2010 provides a good guideline and judging output capabilities but despite being considered clean output according to the distortion thresholds there can be a very large variation in the quality of output from unit to unit despite both recieving passing results. This testing is used to measure distortion and also the maximum useful short term output of a device over various bandwidths. It can give a sense of how much dynamic reserve a system may have. (Note:CEA-2010 documentation actually calls for a measurement from a 1 meter distance and a peak SPL measurement to be reported, however at Data-Bass we follow a reporting method that is a much more typical RMS related SPL and it is referenced at a further distance of 2 meters. There are a number of logical reasons why this method makes much more sense than reporting the peak/1m output. First the peak versus RMS reporting method. These differ by almost exactly 3dB with the peak result being higher of course. However the actual SPL produced by the speaker system does not change. It is just a difference in the way it is reported and calculated from the captured waveform. So why choose one way over the other? An RMS SPL calculation is much more common to see in data and measurements than peak SPL #'s. Much more common. The majority of audio measurement programs default to reporting the rms SPL not peak for the majority of measurements which indicate an SPL and this holds true for loudspeaker development simulation software as well. Again the actual output of the bass system has not changed just the way it is being reported. For example if we put a sine wave sweep into a subwoofer that produces 110dB at 50Hz at 2 meters using a typical rms SPL calculation from the waveform and then without changing the output level at all run the sine wave sweep again with a peak calculation the SPL will appear 3dB louder however the actual bass systems output has not changed at all. The problem is mixing the 2 calculations in one set of measurements. If reporting a peak SPL # for only CEA-2010 versus the rest of the measurements it will have the effect of confusing many people looking at the data. You may look at the maximum long term output measurement and note that the system produced 110dB at 50Hz and then look at the CEA-2010 burst output measurement at 50Hz and see 116dB and think that the subwoofer is capable of bursting 6dB greater output at 50Hz short term but this is not true if the CEA-2010 results are peak results because it adds 3dB because of the different calculation from the peak of the waveform...Actually the subwoofer can produce 3dB greater burst output not 6dB. This is why the CEA-2010 SPL results here are reported in RMS not peak so that they are directly comparable with the rest of the measurements that indicate an SPL presented here without needing to do a calculation in your head first. A 2 meter distance is chosen for measuring/reporting purposes instead of 1 meter for a number of reasons as well. The CEA-2010 documentation appears to show that the reporting methods were primarily intended or marketed to smaller home audio powered subwoofers, but this type of testing actually was used by Mr. Keele long before the documentation made it a standard and it is not limited in usefullness to only subwoofers, bass systems and or active speakers. It can be used with full range and much larger more powerful devices as well. Here at Data-Bass the subwoofers tested may range from a tiny sealed 8" subwoofer to a behemoth horn loaded monster with multiple 18's intended for arenas...With these larger cabinets and subwoofers having multiple points of radiation a 1 meter distance may still be in the nearfield of large cabinets and the size of the enclosure itself can artificially boost the SPL recorded at such a close distance. Also if the cabinet has drivers or radiation points on differing enclosure faces it is impossible to aim all of the output at the microphone and you may end up with one radiator at a much greater distance from the microphone element than another which can skew the response shape and under represent the output power of the device. Moving the microphone back a further distance to 2 meters helps mitigate some of these effects of large cabinets and allows the the output of multiple radiating surfaces to blend together better by minimizing the difference in distance to the microphone element to a larger extent. Also a ground plane measurement is what is known as a half space measurement because there is just the one endless boundary under the microphone and DUT which provides a near perfect 6dB increase in output over anechoic or full space. Using the inverse square law we know that if the microphone is moved back to 2x the distance from the DUT the output should drop by 6dB. So by reporting results at 2 meters instead of at 1 meter the 6dB drop from the inverse square law cancels out the 6dB gain from measuring in halfspace versus full space and what you end up with is output that is approximate to a 1 meter anechoic or fullspace measurement. Yet another reason to measure at 2 meters is to allow headroom in the microphone signal chain. Some of the larger more powerful bass systems tested easily produce outputs in excess of 130dB at 2 meters which would put them in the neighborhood of 140dB at 1 meter which can start to cause issues with clipping in the microphone preamp and soundcard input. In fact a few of the most powerful systems have needed to be measured at 4 meters to ensure that there was sufficient headroom in the measurement chain! Lastly the other 2 large repositories of public CEA-2010 measurements for subwoofers also reported their results in 2 meter groundplane RMS format likely for the same logical reasons presented here. Despite being a "standard" CEA-2010 results can vary quite a bit and have been hit or miss between the various parties taking them, additionally there are variations in the way they are reported, with the most common being either 2 meter groundplane rms or the 1 meter groundplane peak that is specified in the literature. For these reasons it can be difficult to meaningfully compare data from different sources and is often a source of confusion or debate. It is best to just avoid making comparisons to data collected by other sources to those from Data-Bass as it may or may not be reliable and we cannot control the procedures or equipment involved.)

10. A group delay graph is derived from previous measurements. This measurement indicates how much delay or storage of energy versus frequency is taking place in a system. Larger values in MS indicate more latency or energy delay.Ideally there would be zero group delay indicating an incredibly well damped system. In reality there is always some. Generally it is held that as long as the delay is below 1.5 cycles it is inaudible and most who claim to hear differences in delay between speakers would say that anything less than 1 cycle of delay is nothing to worry about. Additionally the deeper in frequency, the longer the wavelength and less sensitive our ears are to it the less audible a long group delay will be. Generally anything below 25-30Hz is not regarded as of much concern unless really bad. For example 40ms of delay at 25Hz is probably not a big deal and not very audible but 40ms of delay at 80Hz is going to be much more audible and of concern. Typically sealed, IB and other lower order systems offer the least amount of delay...Bass reflex systems typically have increased delay near the resonator tuning. Even higher order systems like horns and Bandpass enclosures can have more group delay and more complex delay plots. Additionally equalization and other signal processing, such as high pass filters and boost or parametric signal shaping can increase the group delay. Usually this measurements indicates nothing of concern unless the group delay is exceeding 1.5 cycles above 25Hz.

11. Waterfall and Spectrogram plot graphs are derived from previous measurements. These plots are both different ways of representing the amount of energy storage or decay rate in the systems. Resonances, ringing or other issues will show up with a long and slow decay rate. Look for frequencies that seem to hang around longer than the rest and seem to not want to decay away. For Waterfall graphs look for a response that drops off rapidly and cleanly by 25 to 30dB within the first 100ms. Unless there is a very obvious issue with ringing in a particular frequency band there is not usually anything of particular sonic consequence in these graphs. As with the group delay chart most of the energy storage will occur below 30Hz where it is not of much concern. Pay much more attention to issues higher up in frequency as they become much more audible than they are in the depest bass.

12. High power impedance measurements. An impedance measurement will be taken at the power levels corresponding to the voltage inputs used during the long term output sweeps. The results will be used to gauge changes in the impedance magnitude and phase that occur once high power is input into a system and what effects it may have on the load presented to the amplifier. At higher power levels many changes will occur such as changes in impedance magnitude, the sharpness and magnitude of the peaks, the placement of the dips and peaks will shift, the average impedance may rise, etc.This can give clues as to how hot the voice coil and motor of the driver is getting and whether the resonance of the system is shifting, the ports are compressing etc. (Note: This test takes a completely different setup and calibration from the rest and is lengthy and time consuming. It is also every bit as demanding on the DUT as the long term output compression sweeps. That being the case this test is only done occassionally on a particular system in an effort to provide general, rather than system specific knowledge as to what happens to the impedance at high power levels.)