Ilkka R
Second Unit
- Joined
- May 19, 2004
- Messages
- 270
- Real Name
- Ilkka
There has been a lot of discussion about intermodulation distortion (IMD) over various Internet discussion boards. Very few IMD measurements have been done on any subwoofer, most of the discussion has been basically guessing and making overly generous assumptions related to linear excursion of the woofer etc. I'm not a fan of assumptions, I'm more facts and measurements type of guy. I and Ed Mullen have been working around this IMD subject for a quite some time now and have been trying to find some common ground concerning many different questions relating its measurement and calculation. We are not sure yet whether to include this new test into the already quite wide battery of tests we take.
Theorizing can only go so far, but to get further we need some measurements and actual data so that we can see if our theories and hypothesis were correct. Since we still have a quite thick layer of snow here in Finland, I had to do my testings indoors, but that don't necessarily mean unrealiable results, especially when one is only comparing two different systems and knows the limitations of such way.
Hypothesis: Subwoofers with high excursion woofers suffer from higher levels of IMD than subwoofers with lower excursion woofers.
There has been also discussion about the relevance of the weight of the cone, weight of the amplifier and some other variables. In order to prove or disprove our hypothesis, we need two subwoofers, other equipped with a high excursion woofer, heavy cone and light amp and another equipped with not so high excursion woofer, lighter cone and heavier amp. I haven't actually measured a single one of these parameters, they are just assumptions based on some highly educated guesses made by several self-educated members of the AV community. And the test products are...Axiom EP-600 and SVS PB12-Ultra. If someone doesn't know if by now, Axiom is the one with a heavier amp.
In order to exclude the effects of room, one has to go as close to a subwoofer as possible. But not too close, because as we all know, SPL rises 6 dB everytime we half our measuring distance (if there is no reflective field, so it doesn't apply 100% when indoors), and because my microphone is only good up to 120 dB. My own hypothesis was that there is not much IMD at low sound pressure levels, so I needed to use as high levels as possible. After making some frequency response measurements with both subs, I decided to use 4" measurement distance. Using that distance FR curves were as close to each other as possible and there was still a good amount of distance for that high excursion driver to move before hitting the tip of mic. More info about my gear and how I calibrate them can be found here.
Intermodulation distortion: Nonlinear distortion characterized by the appearance, in the output of a device, of frequencies that are linear combinations of the fundamental frequencies and all harmonics present in the input signals. Note: Harmonic components themselves are not usually considered to characterize intermodulation distortion. When the harmonics are included as part of the distortion, a statement to that effect should be made.
When we measure total harmonic distortion, we use only one fundamental frequency at a time. Already then there are some amounts of IMD present, because the fundamental and the numerous harmonics interfere with each other, but that is not the correct way to produce IMD. In order to produce measurable levels of IMD, we need to use two fundamentals at different frequencies. There are some stardard frequencies for measuring for example IMD of an amplifier (for example 19 kHz and 20 kHz), but there are no standards for subwoofers. In the discussions with Ed Mullen, we decided to use fundamentals at 30 Hz and 70 Hz frequencies. At the moment we use only one pair of frequencies, but it is possible that we use more in the future. There are numerous reasons why spesifically these two were chosen, but that discussion will happen another time. The frequencies of the two-tone intermodulation products can be computed by the equation:
M f1 ± N f2, where M, N = 0, 1, 2, 3, ...
The order of the distortion product is given by the sum of M + N. The second order intermodulation products of two signals at f1 and f2 would occur at f1 + f2, f2 – f1, 2f1 and 2f2. We decided to calculate IM components up to 6th order, because that means the highest frequency of measured IM component is already as high as 380 Hz and the amplitude of components above that frequency is so small, it won't affect on total IMD value. As it was said before, harmonic components aren't usually included as a part of IMD, so we had to exclude every matching IM and harmonic component. Here is a screenshot of the IMD spreadsheet I used to calculate the results to be shown later in this thread. The needed equations can be found from the Internet, so you can make your own if you like. Although it needs some research and hard work, so no, unfortunately I won't share that spreadsheet. That is a proprietary Rissanen/Mullen spreadsheet.
In order to measure the IMD for these two subwoofers, I removed the base plate from the Ultra and placed both subs on the floor of my room (one at a time of course), so that the cones were exactly at the same place and the distance to the microphone was 4" from the level of upper surrounds (During this operation, I noticed that taking off the base plate actually changes the tuning frequency of a subwoofer just a little bit. With the base plate on, Fb of the Ultra was around 21 Hz and around 21.8 Hz sans base plate (20 Hz native tuning frequency i.e. without any port plugs). This could be one explanation for some people reporting differenced after tilting the subwoofer on its side and removing the base plate. Higher tuning frequency changes the FR profile just a bit, which can be heard as an added emphasis in the mid bass range.) I measured also the tuning frequency for the EP-600, and it was exactly the same ~21.8 Hz as for Ultra sans base plate. These measurements were done using a 1" distance. The frequency responses at 4" won't be similar, even with identical tuning frequencies, because Axiom has its vent right next to its woofer, when SVS' vents are located on another side of the subwoofer. Used 30 Hz and 70 Hz fundamentals are therefore good, because neither one of them matches with the tuning frequency of these subwoofers. That means the place of the vent doesn't cause a large difference/error here.
Here are the measured frequency responses using 4" distance. Naturally SVS has still a dip at Fb and Axiom will roll off sharply below 20 Hz and above 100 Hz due its almost brickwall filters, which can not be bypassed. But still the FRs are within +/-1 dB in 30-100 Hz range, which is really good.
I tested both subwoofers for IMD and THD (calculated for both 30 Hz and 70 Hz fundamentals) using three sound pressure levels: 110 dB, 115 dB and 120 dB. Both subwoofers could have gone even higher, but unfortunately my microphone started to clip at higher levels, so I couldn't measure them. There won't be this problem outdoors, when one can use 2 m distance. The output spectrum for each subwoofer and level was recorded and evaluated with S/T SpectraPro, and later manually calculated using the showed spreadsheet. Also the THD was calculated manually using a spreadsheet of its own. Output for each frequency pair were first calibrated at 100 dB level and then rised respectively. Axiom suffered from around 0.5-1 dB of compression at 120 dB test level, so adding input level by same amount was needed. SVS didn't exhibit from compression at used test levels. Since the maximum IMD and THD levels were around 10%, the total SPL (total power) showed by Spectra is basically only the output of two fundamentals at 30 Hz and 70 Hz. The relative levels of two fundamentals weren't adjusted, although they were naturally affected by the FR profile of each subwoofer (also by compression if there was some).
Next series of graphs and tables show the IMD and THD levels for both subwoofers at three different levels. I also calculated the sum of IMD and THD, which naturally can not be done by a simple plus operation, but using the (a^2+b^2)^0.5 formula instead.
Here are the IMD spectrums at 115 dB SPL for both subs. You can self calculate the IMD and THD levels, if you have the needed formulas/spreadsheets.
Previous spectrums overlaid (only a link due large size of the graph).
http://personal.inet.fi/private/ilkk...arison_115.png
Conclusion: High excursion woofers don't seem to suffer from excessive IMD. It can be also seen that IMD isn't a large problem with either subwoofer, combined THD levels are higher at every tested SPL. Also it seems that THD and IMD are quite strongly connected to each other. Subwoofer with a high THD levels will MOST LIKELY suffer from high IMD levels as well. Naturally much more subwoofers must be still tested, but I won't believe things would turn upside down.
The next question will naturally be, do we even have to measure IMD levels or does THD tell us the same information? Does measuring IMD add something we already don't know by measuring THD?
-Ilkka Rissanen
PS. I posted this first on AVS, but David Bott deleted it right away and then banned me. The reason was: Marketing on the site is not welcome and we believe that is why you are now here. Your account is now closed.
Yeah, thanks for nothing.
edit: More results can be found at here.
Theorizing can only go so far, but to get further we need some measurements and actual data so that we can see if our theories and hypothesis were correct. Since we still have a quite thick layer of snow here in Finland, I had to do my testings indoors, but that don't necessarily mean unrealiable results, especially when one is only comparing two different systems and knows the limitations of such way.
Hypothesis: Subwoofers with high excursion woofers suffer from higher levels of IMD than subwoofers with lower excursion woofers.
There has been also discussion about the relevance of the weight of the cone, weight of the amplifier and some other variables. In order to prove or disprove our hypothesis, we need two subwoofers, other equipped with a high excursion woofer, heavy cone and light amp and another equipped with not so high excursion woofer, lighter cone and heavier amp. I haven't actually measured a single one of these parameters, they are just assumptions based on some highly educated guesses made by several self-educated members of the AV community. And the test products are...Axiom EP-600 and SVS PB12-Ultra. If someone doesn't know if by now, Axiom is the one with a heavier amp.
In order to exclude the effects of room, one has to go as close to a subwoofer as possible. But not too close, because as we all know, SPL rises 6 dB everytime we half our measuring distance (if there is no reflective field, so it doesn't apply 100% when indoors), and because my microphone is only good up to 120 dB. My own hypothesis was that there is not much IMD at low sound pressure levels, so I needed to use as high levels as possible. After making some frequency response measurements with both subs, I decided to use 4" measurement distance. Using that distance FR curves were as close to each other as possible and there was still a good amount of distance for that high excursion driver to move before hitting the tip of mic. More info about my gear and how I calibrate them can be found here.
Intermodulation distortion: Nonlinear distortion characterized by the appearance, in the output of a device, of frequencies that are linear combinations of the fundamental frequencies and all harmonics present in the input signals. Note: Harmonic components themselves are not usually considered to characterize intermodulation distortion. When the harmonics are included as part of the distortion, a statement to that effect should be made.
When we measure total harmonic distortion, we use only one fundamental frequency at a time. Already then there are some amounts of IMD present, because the fundamental and the numerous harmonics interfere with each other, but that is not the correct way to produce IMD. In order to produce measurable levels of IMD, we need to use two fundamentals at different frequencies. There are some stardard frequencies for measuring for example IMD of an amplifier (for example 19 kHz and 20 kHz), but there are no standards for subwoofers. In the discussions with Ed Mullen, we decided to use fundamentals at 30 Hz and 70 Hz frequencies. At the moment we use only one pair of frequencies, but it is possible that we use more in the future. There are numerous reasons why spesifically these two were chosen, but that discussion will happen another time. The frequencies of the two-tone intermodulation products can be computed by the equation:
M f1 ± N f2, where M, N = 0, 1, 2, 3, ...
The order of the distortion product is given by the sum of M + N. The second order intermodulation products of two signals at f1 and f2 would occur at f1 + f2, f2 – f1, 2f1 and 2f2. We decided to calculate IM components up to 6th order, because that means the highest frequency of measured IM component is already as high as 380 Hz and the amplitude of components above that frequency is so small, it won't affect on total IMD value. As it was said before, harmonic components aren't usually included as a part of IMD, so we had to exclude every matching IM and harmonic component. Here is a screenshot of the IMD spreadsheet I used to calculate the results to be shown later in this thread. The needed equations can be found from the Internet, so you can make your own if you like. Although it needs some research and hard work, so no, unfortunately I won't share that spreadsheet. That is a proprietary Rissanen/Mullen spreadsheet.
In order to measure the IMD for these two subwoofers, I removed the base plate from the Ultra and placed both subs on the floor of my room (one at a time of course), so that the cones were exactly at the same place and the distance to the microphone was 4" from the level of upper surrounds (During this operation, I noticed that taking off the base plate actually changes the tuning frequency of a subwoofer just a little bit. With the base plate on, Fb of the Ultra was around 21 Hz and around 21.8 Hz sans base plate (20 Hz native tuning frequency i.e. without any port plugs). This could be one explanation for some people reporting differenced after tilting the subwoofer on its side and removing the base plate. Higher tuning frequency changes the FR profile just a bit, which can be heard as an added emphasis in the mid bass range.) I measured also the tuning frequency for the EP-600, and it was exactly the same ~21.8 Hz as for Ultra sans base plate. These measurements were done using a 1" distance. The frequency responses at 4" won't be similar, even with identical tuning frequencies, because Axiom has its vent right next to its woofer, when SVS' vents are located on another side of the subwoofer. Used 30 Hz and 70 Hz fundamentals are therefore good, because neither one of them matches with the tuning frequency of these subwoofers. That means the place of the vent doesn't cause a large difference/error here.
Here are the measured frequency responses using 4" distance. Naturally SVS has still a dip at Fb and Axiom will roll off sharply below 20 Hz and above 100 Hz due its almost brickwall filters, which can not be bypassed. But still the FRs are within +/-1 dB in 30-100 Hz range, which is really good.
I tested both subwoofers for IMD and THD (calculated for both 30 Hz and 70 Hz fundamentals) using three sound pressure levels: 110 dB, 115 dB and 120 dB. Both subwoofers could have gone even higher, but unfortunately my microphone started to clip at higher levels, so I couldn't measure them. There won't be this problem outdoors, when one can use 2 m distance. The output spectrum for each subwoofer and level was recorded and evaluated with S/T SpectraPro, and later manually calculated using the showed spreadsheet. Also the THD was calculated manually using a spreadsheet of its own. Output for each frequency pair were first calibrated at 100 dB level and then rised respectively. Axiom suffered from around 0.5-1 dB of compression at 120 dB test level, so adding input level by same amount was needed. SVS didn't exhibit from compression at used test levels. Since the maximum IMD and THD levels were around 10%, the total SPL (total power) showed by Spectra is basically only the output of two fundamentals at 30 Hz and 70 Hz. The relative levels of two fundamentals weren't adjusted, although they were naturally affected by the FR profile of each subwoofer (also by compression if there was some).
Next series of graphs and tables show the IMD and THD levels for both subwoofers at three different levels. I also calculated the sum of IMD and THD, which naturally can not be done by a simple plus operation, but using the (a^2+b^2)^0.5 formula instead.
Here are the IMD spectrums at 115 dB SPL for both subs. You can self calculate the IMD and THD levels, if you have the needed formulas/spreadsheets.
Previous spectrums overlaid (only a link due large size of the graph).
http://personal.inet.fi/private/ilkk...arison_115.png
Conclusion: High excursion woofers don't seem to suffer from excessive IMD. It can be also seen that IMD isn't a large problem with either subwoofer, combined THD levels are higher at every tested SPL. Also it seems that THD and IMD are quite strongly connected to each other. Subwoofer with a high THD levels will MOST LIKELY suffer from high IMD levels as well. Naturally much more subwoofers must be still tested, but I won't believe things would turn upside down.
The next question will naturally be, do we even have to measure IMD levels or does THD tell us the same information? Does measuring IMD add something we already don't know by measuring THD?
-Ilkka Rissanen
PS. I posted this first on AVS, but David Bott deleted it right away and then banned me. The reason was: Marketing on the site is not welcome and we believe that is why you are now here. Your account is now closed.
Yeah, thanks for nothing.
edit: More results can be found at here.