Over-sampling verses Non over-sampling. I have wondered about this myself, much like tube verses solid state. I think a lot of this is implication of the technology, but I agree with You in that it would be nice to hear the difference for myself. I may pick one of the Muse up, as I could always use it on surround channels if I prefer the Buffalo.
I also looked into the Fremen Amp Kit. I am not sure if I can solder the SMD components. I actually am better at soldering and repaired stained glassed windows then I am at electronic soldering. Much of this is due to the fact that I have far sighted vision and may need customized glasses to see something that small. I know my 3X glasses do not cut it for stuff that small. I will see how the Buffalo Build goes, then if I feel brave, I can try the SMD soldering.
Koldby, did You ever piece together a separate thread for your array project?
It might be time for me to do the same
Thanks for everyone's input.
Now, back to the measurements...
Allen
Thanks for the measurements X, they will come in handy if I do decide to build a better cabinet for the arrays.
ArtsyAllen agree your experience for JR MC thanks feedback.
For me a JR MC standard installation deliver very high quality audio stream, where if i run Foobar or MS MP i need to tweak buffer sizes inside programs if available or sometimes tweak my memory timing settings in computer BIOS to have good quality stream, where JR MC just sound right out of box. Some thoughts about the reason could be that their WASAPI interface is better code in that it is a licensed package and they have the expertise.
I disable my onboard RealTek codec but have tried running it without the driver package using just the native MS Win7 driver. For me that sounds really better but you will not have access to RealTek DSP engine and if I remember right only have a L & R channel, but JR MC still can offer WASAPI upon the codec.
Another positive experinece regarding MS native OS driver i have is a USB HIFImeDIY 24bit up to 96Khz relative cheap sound card. This card for ordinary stereo sounds better than my expensive Xonar Essence ST and M-Audio Delta 192. It can be me that is little weird but that's my experience and to be free of sound quality change with a new driver release version from ASUS or M-Audio is a liberation when running MS native.
For me a JR MC standard installation deliver very high quality audio stream, where if i run Foobar or MS MP i need to tweak buffer sizes inside programs if available or sometimes tweak my memory timing settings in computer BIOS to have good quality stream, where JR MC just sound right out of box. Some thoughts about the reason could be that their WASAPI interface is better code in that it is a licensed package and they have the expertise.
I disable my onboard RealTek codec but have tried running it without the driver package using just the native MS Win7 driver. For me that sounds really better but you will not have access to RealTek DSP engine and if I remember right only have a L & R channel, but JR MC still can offer WASAPI upon the codec.
Another positive experinece regarding MS native OS driver i have is a USB HIFImeDIY 24bit up to 96Khz relative cheap sound card. This card for ordinary stereo sounds better than my expensive Xonar Essence ST and M-Audio Delta 192. It can be me that is little weird but that's my experience and to be free of sound quality change with a new driver release version from ASUS or M-Audio is a liberation when running MS native.
Glad to read you got JRiver to work somehow Allen. It really is a powerful package. I still have the old version of JRiver without the new WMD diver.
I can still route the sound trough JRiver though and take advantage of their audio tools with the audio loopback tool. Eventually I'll upgrade and make it easier on myself.
I've made similar posts as yours about JRiver almost feeling like a marketing employee for them. As said, I store my DVD's on hard disk before playback though that eats up a lot of space, considering I run every hard disk in mirror it is quite expensive space. I'm saving up for a storage NAS with lots of terabytes.
Yesterday I did another measurement but got cut short before I had all info. But I did see better distortion figures than before just by letting JRiver upsample to 96 KHz prior to output to the (upsampling) DAC. Unfortunately I couldn't record a convolved signal due to time restrains but the left and right measurements I made were lower in distortion than I've ever had them (between the ambient noise I have to deal with). Quite puzzled with that result. My M1 DAC didn't accept a higher upsampled signal trough spdif. I wanted to try 176.400. And I couldn't get REW to play anything higher than either 44.1 or 48 KHz.
To be continued...
Edit: just found out I had my SPDIF signal restricted at 96 KHz... will try again soon.
I can still route the sound trough JRiver though and take advantage of their audio tools with the audio loopback tool. Eventually I'll upgrade and make it easier on myself.
I've made similar posts as yours about JRiver almost feeling like a marketing employee for them
. As said, I store my DVD's on hard disk before playback though that eats up a lot of space, considering I run every hard disk in mirror it is quite expensive space. I'm saving up for a storage NAS with lots of terabytes.Yesterday I did another measurement but got cut short before I had all info. But I did see better distortion figures than before just by letting JRiver upsample to 96 KHz prior to output to the (upsampling) DAC. Unfortunately I couldn't record a convolved signal due to time restrains but the left and right measurements I made were lower in distortion than I've ever had them (between the ambient noise I have to deal with). Quite puzzled with that result. My M1 DAC didn't accept a higher upsampled signal trough spdif. I wanted to try 176.400. And I couldn't get REW to play anything higher than either 44.1 or 48 KHz.
To be continued...
Edit: just found out I had my SPDIF signal restricted at 96 KHz... will try again soon.
Last edited:
Fundamentals of SPL advantages of line arrays
First of all, I hope that I am not messing around the discussions on JRiver!
I have some doubts on the fundamentals on SPL advantages of line arrays against point sources. Please correct me if the following 3 points ( taking Wesayso's line array in discussion here as example) are wrong.
1. The lower frequencies in a line array are positively reinforced whenever the individual sources are within half wavelength. So, for 25 sources, we get ~ 14dB boost in the bass region.
Let this extra boost take 50Hz to 85dB/Watt SPL at 1 m & let the array be equalized flat at 85dB for 50Hz & above.
2. Let each array be driven with 5 Watt each. Since each array has a sensitivity of 85db/W/m, we get 92dB at 1m.
3. A conventional point source drops in SPL by 6dB whereas a line source drops by 3dB per doubling of distance. So, a point source at 85dB at 1 m is producing 125dB at 1 cm from its surface. Now since the line source drops by 3dB, at 1m from the array we get 105dB for each individuals drivers working as hard as required to give 85dB at 1m if working alone as a point source.
In points 1 & 2, the difference of sound pressure roll-off with distance between a line & point source is not considered. So, considering all the three points together,
we should get 105 dB/W at 1m from the array
& 112dB for 5W at 1m & 128 dB for 200W at 1m listening distance for frequencies > 50Hz.
This is equal to 122dB at 4m listening distance from a line source against 96dB at 4m from a point source( with 85dB/W/m SPL), both driven at 200W @ 50Hz.
~ a difference of 26dB!!
Am I right? or did I mess or miss anything?
First of all, I hope that I am not messing around the discussions on JRiver!
I have some doubts on the fundamentals on SPL advantages of line arrays against point sources. Please correct me if the following 3 points ( taking Wesayso's line array in discussion here as example) are wrong.
1. The lower frequencies in a line array are positively reinforced whenever the individual sources are within half wavelength. So, for 25 sources, we get ~ 14dB boost in the bass region.
Let this extra boost take 50Hz to 85dB/Watt SPL at 1 m & let the array be equalized flat at 85dB for 50Hz & above.
2. Let each array be driven with 5 Watt each. Since each array has a sensitivity of 85db/W/m, we get 92dB at 1m.
3. A conventional point source drops in SPL by 6dB whereas a line source drops by 3dB per doubling of distance. So, a point source at 85dB at 1 m is producing 125dB at 1 cm from its surface. Now since the line source drops by 3dB, at 1m from the array we get 105dB for each individuals drivers working as hard as required to give 85dB at 1m if working alone as a point source.
In points 1 & 2, the difference of sound pressure roll-off with distance between a line & point source is not considered. So, considering all the three points together,
we should get 105 dB/W at 1m from the array
& 112dB for 5W at 1m & 128 dB for 200W at 1m listening distance for frequencies > 50Hz.
This is equal to 122dB at 4m listening distance from a line source against 96dB at 4m from a point source( with 85dB/W/m SPL), both driven at 200W @ 50Hz.
~ a difference of 26dB!!
Am I right? or did I mess or miss anything?
rockk19,
I don't know if you found this already, but James Griffin White Papers on Line Arrays explains all this and more.
http://www.audioroundtable.com/misc/nflawp.pdf
I don't know if you found this already, but James Griffin White Papers on Line Arrays explains all this and more.
http://www.audioroundtable.com/misc/nflawp.pdf
You're missing the part that you're actually cutting the midrange to achieve the average SPL. If you only cut like in the example to achieve the 85 dB SPL you're cutting the entire range from 50 Hz to about 2 KHz. At ~ 200 Hz they would be most efficient and run on way less power than the mentioned 1 watt or at louder levels way less than the 200 watt.
It would be interesting to see some measurements on free standing arrays in anechoic conditions to see distortion figures for that. I cannot achieve that kind of SPL in my living room without rattles from all kinds of objects in that room.
200 watt is still only 8 watt for each speaker. But with a single TC9 the distortion down low rises fast:
25 drivers do way better as we have seen, lowering distortion but only up to a point.
So I'm sure that is the limiting factor at 200 watt.
That's why I have them close to a wall or corner to boost the low end. Free standing you'd have to have a higher crossover point.
Ideally I would try to place them in a long rectangle room, close to the (long) side walls with appropriate damping on that wall and away from the back wall like a Beveridge line array. I bet that would work awesome. No way for me to try that though, I'd never get that trough the beauty commission here.
It would be interesting to see some measurements on free standing arrays in anechoic conditions to see distortion figures for that. I cannot achieve that kind of SPL in my living room without rattles from all kinds of objects in that room.
200 watt is still only 8 watt for each speaker. But with a single TC9 the distortion down low rises fast:
25 drivers do way better as we have seen, lowering distortion but only up to a point.
So I'm sure that is the limiting factor at 200 watt.
That's why I have them close to a wall or corner to boost the low end. Free standing you'd have to have a higher crossover point.
Ideally I would try to place them in a long rectangle room, close to the (long) side walls with appropriate damping on that wall and away from the back wall like a Beveridge line array. I bet that would work awesome. No way for me to try that though, I'd never get that trough the beauty commission here.
Last edited:
yes, less power required due to cut between 50 Hz & 2KHz is an added bonus ;P
I just wanted to make a reference at 50Hz for both line & point sources, just for simplicity.
Even if the Vd of your array matches a 12" subwoofer with 10mm Xmax, I have no clue how their distortion would compare. The array should give more bass considering benefit of 3dB per doubling of listening distance & for the same listening levels it will help it against distortion.
If I ever build an array, I'll take it out on an open farm field to do some distortion & excursion limited power handling testing!
I just wanted to make a reference at 50Hz for both line & point sources, just for simplicity.
Even if the Vd of your array matches a 12" subwoofer with 10mm Xmax, I have no clue how their distortion would compare. The array should give more bass considering benefit of 3dB per doubling of listening distance & for the same listening levels it will help it against distortion.
If I ever build an array, I'll take it out on an open farm field to do some distortion & excursion limited power handling testing!
It's the same driver. The numbers are 85, 65 and 45 db at the side. Not that much difference to Zaph I guess at ~10 dB down @ 85 dB.
Compare that to one my measurements with 50 Hz at the actual 85 dB:
Distortion still down 40 dB at 50 Hz.
Found one even louder than that:
(2 graphs to show difference for 5 dB more volume, still not accurate measurements due to ambient noise)
Without a wall nearby I guess I wouldn't get these kind of numbers. But I cannot be sure until someone actually tries.
These measurements were 2 arrays playing at 3 meter at calibrated SPL. You do the math.
Compare that to one my measurements with 50 Hz at the actual 85 dB:
Distortion still down 40 dB at 50 Hz.
Found one even louder than that:
(2 graphs to show difference for 5 dB more volume, still not accurate measurements due to ambient noise)
Without a wall nearby I guess I wouldn't get these kind of numbers. But I cannot be sure until someone actually tries.
These measurements were 2 arrays playing at 3 meter at calibrated SPL. You do the math
.
Last edited:
While I have played with IR only, I wouldn't call it a fair comparison.
After I finished the towers I played them for 2 - 3 weeks with EQ applied in a rough guestimate. After that I let REW decide on the EQ needed with about 20 PEQ boosts/cuts on each side. That didn't sound that well. After removing the narrow corrections REW suggested in it's auto EQ with a Q of 10 or even higher it was way more natural sounding.
Shortly after that I went the FIR route, and didn't look back. It immediately gave me more pleasing results and the ease of use was a huge plus. I can change/try new curves and other factors in seconds instead of redoing a lot of boosts and cuts. The imaging like I have now with FIR I didn't obtain with IIR. But 2 extra damping panels have been added so I cannot claim it's all due to FIR processing. I still use my graphical EQ to zoom in on good tonality and add those adjustments in my FIR settings if I'm pleased with it.
Still there are so many factors you can change with FIR filters that determine how it all sounds, if I change one little thing I can get surprisingly different results. I did notice I need quite a large window at higher frequencies for optimum focus of the imaging. Not surprising as the arrays do have a bit of time smear due to the different path lengths of the 25 drivers to your ears.
I was trying out shorter time windows in my filters for a while figuring I could get away with that due to the extra damping panels. It wasn't until I tried a longer window that the center snapped back into much better focus and made me smile again. With IIR PEQ I used even longer windows and for a fair comparison I'd need to use gating to adjust the higher frequencies.
Hope this somewhat answers your question?
After I finished the towers I played them for 2 - 3 weeks with EQ applied in a rough guestimate. After that I let REW decide on the EQ needed with about 20 PEQ boosts/cuts on each side. That didn't sound that well. After removing the narrow corrections REW suggested in it's auto EQ with a Q of 10 or even higher it was way more natural sounding.
Shortly after that I went the FIR route, and didn't look back. It immediately gave me more pleasing results and the ease of use was a huge plus. I can change/try new curves and other factors in seconds instead of redoing a lot of boosts and cuts. The imaging like I have now with FIR I didn't obtain with IIR. But 2 extra damping panels have been added so I cannot claim it's all due to FIR processing. I still use my graphical EQ to zoom in on good tonality and add those adjustments in my FIR settings if I'm pleased with it.
Still there are so many factors you can change with FIR filters that determine how it all sounds, if I change one little thing I can get surprisingly different results. I did notice I need quite a large window at higher frequencies for optimum focus of the imaging. Not surprising as the arrays do have a bit of time smear due to the different path lengths of the 25 drivers to your ears.
I was trying out shorter time windows in my filters for a while figuring I could get away with that due to the extra damping panels. It wasn't until I tried a longer window that the center snapped back into much better focus and made me smile again. With IIR PEQ I used even longer windows and for a fair comparison I'd need to use gating to adjust the higher frequencies.
Hope this somewhat answers your question
?
Over-sampling verses Non over-sampling. I have wondered about this myself, much like tube verses solid state. I think a lot of this is implication of the technology, but I agree with You in that it would be nice to hear the difference for myself. I may pick one of the Muse up, as I could always use it on surround channels if I prefer the Buffalo.
I also looked into the Fremen Amp Kit. I am not sure if I can solder the SMD components. I actually am better at soldering and repaired stained glassed windows then I am at electronic soldering. Much of this is due to the fact that I have far sighted vision and may need customized glasses to see something that small. I know my 3X glasses do not cut it for stuff that small. I will see how the Buffalo Build goes, then if I feel brave, I can try the SMD soldering.
Koldby, did You ever piece together a separate thread for your array project?
It might be time for me to do the same
Thanks for everyone's input.
Now, back to the measurements...
Allen
No not a thread for myself, but I babled along about my project here:
http://www.diyaudio.com/forums/multi-way/193015-stupid-cheap-line-array-19.html
and the next couple of pages in that thread.
Koldby
+1 to Wesayso's observations.
I have tried both the regular parametric EQ (IIR) and created the same curve in FIR and compared the two. I have never been a big fan of EQ, as it always seemed to take away from the 3D rendering of soundstage. It would seem this is less of an issue with the FIR filters. In fact, I am amazed at the tonality I can create with 99 cent drivers in a NSB array! My center image in not quite grounded, so I am thinking I need to learn and play around with the "windowing". I have a lot to learn, so stay posted...
What I can say so far, is FIR filters are worth the effort.
Allen
Since I moved my media center to a PC and using JRiver (sold the Mac as that was an extra and someone needed it more), I decided to give that DRC thing a try.
I went the easy route and used DRC Designer to make a sweep and spit out a filter. Keeping it simple, I just used the first one on the list, the ERB Psycho-Acoustic, and plugged the file into Jriver convolution plugin.
The result was... less than stellar. Actually, pretty bad, everything got muddied and boomy.
Not a good start but willing to give it another go, as you guys seem to vouch for this 100%.
Any pointers for my next try?
I went the easy route and used DRC Designer to make a sweep and spit out a filter. Keeping it simple, I just used the first one on the list, the ERB Psycho-Acoustic, and plugged the file into Jriver convolution plugin.
The result was... less than stellar. Actually, pretty bad, everything got muddied and boomy.
Not a good start but willing to give it another go, as you guys seem to vouch for this 100%.
Any pointers for my next try?
Perceval,
Have not tried DRC yet myself but think magicians wesayso and gmad talked much about a windowing size strength need to be set for the corrections, some clue data windowing size gmad use can be seen here #446 http://www.diyaudio.com/forums/full-range/265053-hypercube-loudspeakers-45.html#post4275564, hope this is not wrong and help a bit.
Have not tried DRC yet myself but think magicians wesayso and gmad talked much about a windowing size strength need to be set for the corrections, some clue data windowing size gmad use can be seen here #446 http://www.diyaudio.com/forums/full-range/265053-hypercube-loudspeakers-45.html#post4275564, hope this is not wrong and help a bit.
I have a few...
I use REW for the measurements as that gives me a visual control factor not present in the DRC Designer package. I use the longest sweep possible in REW.
First I run a sweep trough both left and right speaker. Looking at the impulse tab I can see if I have the microphone in the correct listening position. It should show a single impulse peak. Usually I have to move the microphone just a little to line up the peaks shown on the impulse tab (verify with a new sweep).
After that I'm pretty sure the microphone is in the sweet spot. I then run about 3 sweeps for the right channel and 3 for the left, observing the differences between the 3 sweeps. Usually they are almost exactly the same. (Except for distortion where differences can be seen due to outside traffic noise).
Look at the impulse graphs and pick one left and one right measurement that has the peak exactly lined up on the graph's "0". Export the impulse from REW to a 32 bit WAV mono file. These two mono files I convert each from 32 bit wav to 32-bit IEEE Float (16.8) PCM in Cool Edit. Then rename to the DRC Designer defaults: LeftSpeakerImpulseResponse44100.pcm and RightSpeakerImpulseResponse44100.pcm
Other packages might do too like Audacity, not tested that though.
Next I run DRC Designer to create the convolver filters.
Hope that helps....
Actually the standard "ERB" template is the one that has one of the shortest windows in the Low Frequencies (65 ms at 20 Hz) and quite a long High Frequency window (0.46 ms at 20 KHz).
So it doesn't control much of the room modes due to that short window at Low Frequencies. Run a "Normal" template to see what that does.
(500 ms at 20 Hz, 100 ms at 100 Hz, 10 ms at 1 KHz, 0.50 ms at 20 KHz)
If that's more to your liking you could adjust with the custom settings. If you're adjusting single full range speakers and not line array's you might need way shorter upper windows.
In Audiolense this info is given as cycles as well as milliseconds. That way you can see how many cycles (front/back movement of speaker cone) you are controlling.
You could use Audiolense to record the sweep. That one will give you even more control over the measurement than REW. I use REW because I can get the impulses. Do my thing with them and play trough JRiver's loopback to see what the convolved signal looks like. You actually need 2 soundcards for that. I use my Asus Xonar for record as well as play back in REW and loop the play back trough JRiver (File -> Open Live) to my Musical Fidelity M1 Dac. The draw back is I have to use Wasapi and not ASIO.
I can't look up every detail right now as I am in the middle of swapping computers. The hard way I might add, didn't want to setup all programs again and drivers etc. so I swapped the Hard Disk from old PC to new. It works but still have to add my second OS disk to get all my raid drives back. (it only works because I swapped all internals from the old PC to the new. Graphics card and disks were all only one or two years old)
But going to a much faster processor it is worth the pain.
I use REW for the measurements as that gives me a visual control factor not present in the DRC Designer package. I use the longest sweep possible in REW.
First I run a sweep trough both left and right speaker. Looking at the impulse tab I can see if I have the microphone in the correct listening position. It should show a single impulse peak. Usually I have to move the microphone just a little to line up the peaks shown on the impulse tab (verify with a new sweep).
After that I'm pretty sure the microphone is in the sweet spot. I then run about 3 sweeps for the right channel and 3 for the left, observing the differences between the 3 sweeps. Usually they are almost exactly the same. (Except for distortion where differences can be seen due to outside traffic noise).
Look at the impulse graphs and pick one left and one right measurement that has the peak exactly lined up on the graph's "0". Export the impulse from REW to a 32 bit WAV mono file. These two mono files I convert each from 32 bit wav to 32-bit IEEE Float (16.8) PCM in Cool Edit. Then rename to the DRC Designer defaults: LeftSpeakerImpulseResponse44100.pcm and RightSpeakerImpulseResponse44100.pcm
Other packages might do too like Audacity, not tested that though.
Next I run DRC Designer to create the convolver filters.
Hope that helps....
Actually the standard "ERB" template is the one that has one of the shortest windows in the Low Frequencies (65 ms at 20 Hz) and quite a long High Frequency window (0.46 ms at 20 KHz).
So it doesn't control much of the room modes due to that short window at Low Frequencies. Run a "Normal" template to see what that does.
(500 ms at 20 Hz, 100 ms at 100 Hz, 10 ms at 1 KHz, 0.50 ms at 20 KHz)
If that's more to your liking you could adjust with the custom settings. If you're adjusting single full range speakers and not line array's you might need way shorter upper windows.
In Audiolense this info is given as cycles as well as milliseconds. That way you can see how many cycles (front/back movement of speaker cone) you are controlling.
You could use Audiolense to record the sweep. That one will give you even more control over the measurement than REW. I use REW because I can get the impulses. Do my thing with them and play trough JRiver's loopback to see what the convolved signal looks like. You actually need 2 soundcards for that. I use my Asus Xonar for record as well as play back in REW and loop the play back trough JRiver (File -> Open Live) to my Musical Fidelity M1 Dac. The draw back is I have to use Wasapi and not ASIO.
I can't look up every detail right now as I am in the middle of swapping computers. The hard way I might add, didn't want to setup all programs again and drivers etc. so I swapped the Hard Disk from old PC to new. It works but still have to add my second OS disk to get all my raid drives back. (it only works because I swapped all internals from the old PC to the new. Graphics card and disks were all only one or two years old)
But going to a much faster processor it is worth the pain.
Last edited:
for a snappy responsive computer hope it sounds as good or better as the old one without too much trouble and tweaking and thanks DRC tips.
wesayso, thanks for the tips.
I tried the "Normal" setting for the creation of filters, and it sounded boxy, far from anything listenable.
Also, if I leave "Normalise the sound" in JRiver's plugin, the output is really low. Unchecked and the output is back to reasonable levels.
I will try using REW to generate the sweep and record that. I'll see if it improves things a bit. I'm still just getting my feet wet using DRC.
I tried the "Normal" setting for the creation of filters, and it sounded boxy, far from anything listenable.
Also, if I leave "Normalise the sound" in JRiver's plugin, the output is really low. Unchecked and the output is back to reasonable levels.
I will try using REW to generate the sweep and record that. I'll see if it improves things a bit. I'm still just getting my feet wet using DRC.