ZilchLab said:
I guess I don't understand. I thought Marcus number were the noise levels not the signal levels, but I am now confused what is what. My signals are about 94 dBSPL at the mic. The noise at that same position with the source off is probably around 70 dB. I don't do things in absolute numbers like this so I don't really know.
noah katz said:
True, I have had a lot of trouble with the crossovers BECAUSE good waveguides are big. I find, as I have said before, that getting the waveguide right trumps getting the vertical nulls right. Thats what I trade-off and I'll stand by it.
I measure 94 dBSPL in the nearfield of the speaker and 68 dBSPL at the listening position.
May be the 68 sb 86??
"Actually it seems this would highlight the one weakness of WG vs conventional systems - the requisite wider vertical driver spacing and associated nulls."
It's wavelength dependent as well. It's combination of the two.
Rob🙂
May be the 68 sb 86??
"Actually it seems this would highlight the one weakness of WG vs conventional systems - the requisite wider vertical driver spacing and associated nulls."
It's wavelength dependent as well. It's combination of the two.
Rob🙂
Got those numbers with white noise from my cheap SPL meter C-weighted. Not sure if I messed something up because I was busy doing other things in between...
Got the 200 Hz white noise sample from Earl which is more apartment-friendly. This is what I've achieved so far:
Sounds good! But needs some more work as the bass didn't pass my Seal "Human Beings" test. If you get that bass right than you made it.
I guess there are too many parameters I have access to (LF cutoff, filter characteristic, level, phase, delay). Earl, where is the priority? When setting up the first sub then I guess you have to live with big peaks and dips showing up? What is the main characteristics of the frequency response to look at in this step?
Best, Markus
Got the 200 Hz white noise sample from Earl which is more apartment-friendly. This is what I've achieved so far:
An externally hosted image should be here but it was not working when we last tested it.
Sounds good! But needs some more work as the bass didn't pass my Seal "Human Beings" test. If you get that bass right than you made it.
I guess there are too many parameters I have access to (LF cutoff, filter characteristic, level, phase, delay). Earl, where is the priority? When setting up the first sub then I guess you have to live with big peaks and dips showing up? What is the main characteristics of the frequency response to look at in this step?
Best, Markus
That's certainly more consistent with my measurement levels at 1M; RTA pink noise is usually lower than that by ~10 dB, tho. MLS and Sinusoidal, I let 'er rip. CLIO protests if it's too high, but I kinda already know that when it occurs. My "YEOW!" has already messed up the measurement anyway.gedlee said:
Listening to progam is not apples and apples, of course, but 68 dB would be appreciably lower than normal listening levels, 80 - 85 dB here....
Hey Pos,
Thats exactly what I'm trying to do! A bang4buck version using 41hz.com amp-9 and a Elliot Sound Project p09b crossover. Iv'e been reporting in this thread here:
http://www.diyaudio.com/forums/showthread.php?postid=1589987#post1589987
col.
Thats exactly what I'm trying to do! A bang4buck version using 41hz.com amp-9 and a Elliot Sound Project p09b crossover. Iv'e been reporting in this thread here:
http://www.diyaudio.com/forums/showthread.php?postid=1589987#post1589987
col.
In-room response
I've started mapping my setup (since this thread seems to be immersed in FFT impulse measurements) so that I can possibly improve the sound. Still figuring out the software (Sample Champion) to see if I want to purchase it.
Here is what I've measured at the listening position (~15 feet back from speakers, all units running with audyssey room correction) Raw data, no smoothing. I'm still debugging the setup, so there are artifacts in the data for sure, but the response looks fairly flat except for the peaks at ~5800 hz (probably membrane resonance) and at ~14khz (I can't hear much in this region anyway).
Scale is 7.5 dB / vertical division.
John L.
I've started mapping my setup (since this thread seems to be immersed in FFT impulse measurements) so that I can possibly improve the sound. Still figuring out the software (Sample Champion) to see if I want to purchase it.
Here is what I've measured at the listening position (~15 feet back from speakers, all units running with audyssey room correction) Raw data, no smoothing. I'm still debugging the setup, so there are artifacts in the data for sure, but the response looks fairly flat except for the peaks at ~5800 hz (probably membrane resonance) and at ~14khz (I can't hear much in this region anyway).
Scale is 7.5 dB / vertical division.
An externally hosted image should be here but it was not working when we last tested it.
John L.
John, what do we see? Real time FFT or windowed? If you want to evaluate the speakers, only free field measurements (i.e. windowed) are helpful. If you want to calibrate your speakers with subs then you need to have a much finer resolution (1 Hz would be perfect). And you only need to look at 20 Hz to 200/300 Hz.
If you're new to measuring loudspeakers then I suggest to start with ARTA:
http://www.fesb.hr/~mateljan/arta/index.htm
It has all you need and a big online user base.
Best, Markus
If you're new to measuring loudspeakers then I suggest to start with ARTA:
http://www.fesb.hr/~mateljan/arta/index.htm
It has all you need and a big online user base.
Best, Markus
markus76 said:
That noise file only has sound up to 200 Hz. so above that the fall is from the source.
With only one sub there will be a lot of peaks and dips. You need to find out whats real and spatial averaging helps as does frequency averaging. The first sub should get you a "mean" level just below where you want it to end up. So the priority here is the level and the LP frequency. Don't even worry about phase yet. You might want to do very wide frequency averaging at this stage, even 1 octave or at least 1/3 octave. Smoothness is not an citeria at this point.
Now turn on the second sub and look at finer detail - 1/3 octave. Adjust gain first until you see an effect, then adjust the phase and LP point for the lowest variations in the curve. You should be able to get much smoother at this point. Now add the third sub and you can go to full resolution. This sub should have only a marginal effect overall. Look for it to mostly fill in some remaining holes. You can try phase to knock down some peaks. What will happen is that everything you do will have its pluses and minuses. Just use your own judgement.
Finally, use some EQ on any remaiing peaks and dips. This has always gotten me good results.
Also, try doing the first steps simply by ear. You'd be surpsised how well this works. You can always look at the details after and try and improve it. The final EQ has to be done with measurements or you can seriuosly screw things up.
Or just do the first sub by ear and add the other two, or visa versa.
So for step 1 isn't the best start to do a nearfield measurement of the woofer and sub and then cross them over "traditionally" as I did?
Best, Markus
Best, Markus
If you are going to start using EQ then what is the point of having any passive components? You have to buy some form of signal processor. At the point where you buy the signal processor the argument that using passive components in a crossover and only one amp is cheaper, goes out of the window. If you are going to use a signal processor you may as well use it to do all the stuff that the passive components are doing and use it for your room correction too.
I use cheaper speaker parts, I have NO passive components between my amps and the speaker drivers (apart from the speaker cable). All my signal processing is done in the digital domain before it gets to my DAC which then passes the signal on to a very minimal linkwitz riley crossover (please look at the ESP P09B) where I balance the levels between the drivers.
The only expensive bit of kit in the equation is the Behringer DEQ2496 which gives me the flexibility of a 10 band parametric EQ and a 31 band graphic, per channel, plus other features ($350). I use that to take care of all signal manipulation. Iv'e avoided the cheap analogue inputs and outputs by only using the DEQ2496's digital In/out.
It's a shame I don't live around the corner from Markus as It would be great to do a comparison.
col.
I use cheaper speaker parts, I have NO passive components between my amps and the speaker drivers (apart from the speaker cable). All my signal processing is done in the digital domain before it gets to my DAC which then passes the signal on to a very minimal linkwitz riley crossover (please look at the ESP P09B) where I balance the levels between the drivers.
The only expensive bit of kit in the equation is the Behringer DEQ2496 which gives me the flexibility of a 10 band parametric EQ and a 31 band graphic, per channel, plus other features ($350). I use that to take care of all signal manipulation. Iv'e avoided the cheap analogue inputs and outputs by only using the DEQ2496's digital In/out.
It's a shame I don't live around the corner from Markus as It would be great to do a comparison.
col.
Maybe you should at least include a protection cap for the compression driver.
I do them same in my current system and it is quite frightening to connect a compression driver directly to an amp.
You have to be sure you amp does not generate noise at start, and has no DC offset. And you also have to make sure you turn on you amp after your processor, and turn it off before.
You can also build a protection system with relays.
A protection cap is really the minimum protection. If you are sure your amp does not generate DC offset or start noise then you could put the protection cap at line level, between the processor and the amp.
Another problem is that the amp needs to be really quite because of the sensitivity of the compression driver. In a passive system it is padded way down to match the woofer.
The 41hz amp9 has a mute circuit that I switch last that deals with any pops from other bits upon startup but yeah, your right I shouldn't be too purist and a polypropylene protection cap is good insurance.
col.
col.
markus76 said:
Naw, I'm not trying to calibrate with subs, I have 4 running, they are fully integrated with the mains, such that localization in the room is impossible, and bass is smooth and profound down to ~ 18 Hz. The mains are 6' BG75 planars as quasi linesource dipoles, the only lobing I get is 1 horizontal between it and the 25" neoplanar at ~45 degrees off axis, well off the listening angle. No vertical lobing at all, and the image is rock solid throughout the room. Room size is ~15'000 ft^3 and setup produces >105dB C avg. at lp 14 feet from screen/mains.
Yeah, I've played with ARTA... most of this is to try and confirm other measurements and or subjectively excellent observations many have made about this setup.
thnx for the input.... the plot is windowed non-smoothed FFT of MLS impulse
http://www.purebits.com/scoverview.html
John L.
EQ
Markus,
After you are setup as best you can with placement/treatments, whatever, if you like I can bring a DEQ2496. I don't know your setup, you may have better EQ.
Markus,
After you are setup as best you can with placement/treatments, whatever, if you like I can bring a DEQ2496. I don't know your setup, you may have better EQ.
markus76 said:
Markus - I would never bother with a nearfield of the subs when in a real room. It doesn't tell you much of anything because the room will dominate the situation. And I really hope that you didn't mean "and then cross them over traditionally". That would be a big mistake. I thought that I was clear on this point before.
col said:
If EQ is already there (as in Markus case) then use it - but ONLY for frequencies below 200 Hz. Otherwise passive EQ in the system is fine.
gedlee said:
I'm still talking about step one: The first sub has to extend the LF of the Nathan, so setting the lowpass cutoff higher than where the mains have dropped -6 db would automatically lead to a bump in the frequency response?
What subs to start the process? As I'm sitting near a wall I guess the sub #3 (see post #1) is the most important one and the one to start with?
Markus
ALL the subs extend the LF of the Nathan.
I would start with any sub BUT #3. I think that I said this before, but maybe I wasn't clear. Sub #2 would probably be a good choice since its in a corner. The LP filter needs to be set based on what the response is spatially averaged at the listener positions. What the near fields of the mains and the sub do is not important. Then add sub #1 and finally #3. Since #3 is so close it should be barely audible. This is exactly the characteristic of the last sub added. It should add almost nothing to the overall SPL and simply tends to fill in some holes. #3 should be so low in level that its barely possible to even tell if its on without moving your ear right up to it. If this isn't the case then something is wrong.
If you start with #3 it is almost certainly going to be too loud.
ALL the subs extend the LF of the Nathan.
I would start with any sub BUT #3. I think that I said this before, but maybe I wasn't clear. Sub #2 would probably be a good choice since its in a corner. The LP filter needs to be set based on what the response is spatially averaged at the listener positions. What the near fields of the mains and the sub do is not important. Then add sub #1 and finally #3. Since #3 is so close it should be barely audible. This is exactly the characteristic of the last sub added. It should add almost nothing to the overall SPL and simply tends to fill in some holes. #3 should be so low in level that its barely possible to even tell if its on without moving your ear right up to it. If this isn't the case then something is wrong.
If you start with #3 it is almost certainly going to be too loud.
- Status
- Not open for further replies.