|
|
Post by nickwin on Apr 23, 2015 9:38:31 GMT -5
Ok, so this is a problem I have had since I started using Dirac for Mac on the XMC1. Basically my subs and mains will measure very well after running Dirac, but once I send a signal from my mains to my subs by using the crossover in the XMC1, my bass response gets all jumbled up. Beyond that, the bass response I get is dependent on which speaker it is being routed from. Even though Im using a 90z crossover, the bass routed from my center would measure totally different than the bass routed from my front left, or front right... Even weirder, when routing bass from my mains, peaks that show up when measuring my mains full range are showing up in the subwoofer response, (look at the 20hz peak in the graph I attached, that peak is present on my front left NOT in my subwoofers natural response.) This is happening with a 90z crossover active. Its almost like the crossover is being bypassed in the XMC1. Up until recently I thought this was just the nature of measuring and EQing more than one speaker in isolation and summing them, but yesterday I took some new measurements in Dirac, created a new filter set, and found when I measured it in REW that the bass was perfectly smooth with that filter set. The bass passed from my mains to my subs with this filter set looked exactly like the LFE channel on its own, no weird 5-10db peaks introduced when using the crossover like I usually get. Also with this filter set the bass was exactly the same regardless of which speaker it was being routed from. If I overlay my 5 main speakers on top of each other with that filter set, they lined up perfectly below 80hz. This is where it gets really weird, I took another set of measurements yesterday and created new filters from them. These produced the same results I usually get with the weird summing behavior. I decided to go back to the previous set of filters (the good ones), but when I created another set of filters using the set of measurements that produced the good results, It gave me the same lumpy response I usually get. So same measurements, same project file, same target curves, same crossover settings totally different bass response results. Im trying to narrow down the cause of these differences. At this point I don't know if its Dirac or the XMC1 thats causing these inconsistencies. Attached is the bass response of my front left routed to my sub with a 90z crossover. Neither line have any smoothing, the green line is the one that is being routed to the sub properly. Note the 5db peak at 20hz on the purple line. That is not present in my subwoofer response (but it is in my front left speaker). Attachments:
|
|
|
|
Post by nickwin on Apr 23, 2015 9:49:52 GMT -5
Here are the bass responses of my left center and right fronts with a 90z crossover (so what you see is output from my subs). Non of these have any smoothing applied. I ran the center channel sweep in the "smooth" graph a few db higher, but you can see that the bass response routed from all 3 speakers is identical as it should be. These were created from the the same set of measurements. As far as Im concerned the sub should produce the same response regardless of which channel the signal originates from. Obviously there will be some differences around the crossover point where the speaker and sub interact, but this should only be around the crossover, not at 60-20hz. Unfortunately I haven't been able to duplicate the "good" results. Attachments:
|
|
|
|
Post by socketman on Apr 23, 2015 14:36:02 GMT -5
It seems like if you try and modify an existing filter set things get messed up but if you run the measurements and create a filter set everything is fine. Can you actually hear a difference or is it strictly measurements .
|
|
|
|
Post by nickwin on Apr 23, 2015 19:57:54 GMT -5
I thought the bass resulting from the filters that look extra smooth in the graph sounded extremely tight, more so than normal, but I saw the measurements first so that may have changed my perception. It just seems like anytime Im using a crossover nothing behaves like it should. This is what happened to bass routed from my front left when added 1db to the LFE channel. The plus 1db plot is 3db up at 28hz and actually lower at 20hz. I can add 10db gain to my LFE channel the response will keep the same shape when the signal is going to the LFE channel directly. The weird thing is I didn't get this behavior with the filters that produced the smooth response. When I added up to 6 db it raised the entire response like you would expect it to while keeping the same overall shape, even when routed through the crossover. Attachments:
|
|
edrummereasye
Sensei
"This aggression will not stand, man!"
Posts: 438
|
Post by edrummereasye on Apr 25, 2015 2:03:19 GMT -5
Something crazy is going on, for sure...I wish I had time to test on my unit, but...I haven't even gotten around to running Dirac yet...broke in new speakers, need to play with positioning...plus my main sub is dead, gotta pull the board and send it to Outlaw, I suspect... I know ansat mentioned a couple things he couldn't explain, including maybe something with the xover...but don't know if it was as bad as what you're seeing.
You're measuring at the output, or at the speaker? Won't get far with Emo unless you're measuring at the output...
Would be great if some other REW-capable folks could see if they can duplicate this...need to determine if this is a one-off with your unit, a defect that affects some units, or a "state" that a unit can somehow get into...something goes wack internally, or with the filters upload...
|
|
edrummereasye
Sensei
"This aggression will not stand, man!"
Posts: 438
|
Post by edrummereasye on Apr 25, 2015 2:19:20 GMT -5
Out of curiosity...have you taken any measurements with Presets 1 and 2 , i.e. , crossovers active, no Dirac? How about with PEQ filters and crossovers engaged?
|
|
tubby
Emo VIPs 
Route 2 in Weekapaug!!!
Posts: 408
|
Post by tubby on Apr 25, 2015 9:30:48 GMT -5
I 2nd the recommendation to measure straight from the output to eliminate any room/speaker influence. Also if you can't let's see the sub response by itself without the mains playing. To be clear, let's see the redirected bass from FL, from, and C on one chart but only the output from the sub, turn off the amp for the main speakers. Actually add an LFE signal as well if you can. These should all be the same if everything is woking properly.
How do have REW setup? Are you connected with HDMI? What mode on the XMC?
|
|
|
|
Post by nickwin on Apr 25, 2015 11:33:15 GMT -5
So far Ive only taken speaker measurements from the MLP. My set up is Macbook Pro > HDMI out to XMC1. XMC set to stereo or surround (I get the same results either way). Im using an EMM1 that has been calibrated by Cross Spectrum. Im hoping someone else will test there XMC so I know if its my unit. I'm pretty burned out on taking measurements at the moment. Ive been playing around with Dirac and REW for a few weeks now. Ive got a filter set Im happy with and it sounds really good even though Im still getting the weird sub behavior. Im sure my curiosity will get the best of me at some point and ill try and get some measurements directly off of the XMC. That said Im confident that what Im measuring is real because what Im looking at is far enough below the crossover that speaker interaction shouldn't be an issue. Im wondering if something weird is going on with the with how the XMC applies filters to bass that has been routed from "small" speakers. Maybe the the filters for the main speaker AND the subwoofer are being applied? I got in touch with a developer at Dirac that has knowledge of the XMC, he's taking a look at some of my measurements and is going to let me know if this looks like something Dirac itself could cause or not. Its really looking like this is originating in the XMC though. If anyone else could test there XMC that would be awesome. Id really like to know if this is just my unit or if its more widespread. Here is a graph of my sub response and my front left routed to my sub so you can see what the sub looks like on its own. It's the behavior of the routed bass below 30hz that really has me scratching my head. Iv also noticed that the response of routed bass changes even from sweep to sweep. This doesn't happen with the LFE channel and it definitely doesn't seem right. Attachments:
|
|
|
|
Post by nickwin on Apr 25, 2015 11:44:39 GMT -5
On a side not I was able to fix the huge suckout I was getting on my fronts by adding a few ms delay to my subs using Minidsp 2x4 after running Dirac. It would be real nice if you could do this on the XMC itself, but a MiniDSP does the trick for about $110. Attachments:
|
|
|
|
Post by nickwin on Apr 25, 2015 12:32:00 GMT -5
Interestingly here is an old measurment I dug up of preset 1 that seems to have the same behavior. That was with a 80z crossover. Maybe this isn't isolated to only the Dirac preset. Attachments:
|
|
tubby
Emo VIPs
Route 2 in Weekapaug!!!
Posts: 408
|
Post by tubby on Apr 25, 2015 14:29:19 GMT -5
So far Ive only taken speaker measurements from the MLP. My set up is Macbook Pro > HDMI out to XMC1. XMC set to stereo or surround (I get the same results either way). Im using an EMM1 that has been calibrated by Cross Spectrum. Im hoping someone else will test there XMC so I know if its my unit. I'm pretty burned out on taking measurements at the moment. Ive been playing around with Dirac and REW for a few weeks now. Ive got a filter set Im happy with and it sounds really good even though Im still getting the weird sub behavior. Im sure my curiosity will get the best of me at some point and ill try and get some measurements directly off of the XMC. That said Im confident that what Im measuring is real because what Im looking at is far enough below the crossover that speaker interaction shouldn't be an issue. Im wondering if something weird is going on with the with how the XMC applies filters to bass that has been routed from "small" speakers. Maybe the the filters for the main speaker AND the subwoofer are being applied? I got in touch with a developer at Dirac that has knowledge of the XMC, he's taking a look at some of my measurements and is going to let me know if this looks like something Dirac itself could cause or not. Its really looking like this is originating in the XMC though. If anyone else could test there XMC that would be awesome. Id really like to know if this is just my unit or if its more widespread. Here is a graph of my sub response and my front left routed to my sub so you can see what the sub looks like on its own. It's the behavior of the routed bass below 30hz that really has me scratching my head. Iv also noticed that the response of routed bass changes even from sweep to sweep. This doesn't happen with the LFE channel and it definitely doesn't seem right. Purple is sub only(bass from left) and blue is left +sub? Hopefully the Dirac contact can help you out. I would do some testing here but we got a puppy a few weeks ago and there is not a moments peace until very late. If I can I will give it a go tonight or tomorrow night. I was thinking the same thing about the filters doubling up. You could try moving the low curtain on the mains up to 90hz to eliminate any correction below the cross over. That should give you an indication if this is the problem. The suggestion to take the measures is to eliminate as many variables as possible. The only thing you have to worry about in that setup is the pc and the XMC. No question about HVAC Turing on during a sweep or the mic shifting, you sitting in a different spot,etc. I understand not wanting to take anymore measurements though. At some point you just want to sit down and actually listen to something other than test tones. |
|
|
|
Post by nickwin on Apr 25, 2015 15:06:46 GMT -5
Purple is a signal fed to the LFE channel. Blue is left+sub.
|
|
tubby
Emo VIPs
Route 2 in Weekapaug!!!
Posts: 408
|
Post by tubby on Apr 25, 2015 23:16:28 GMT -5
So I ended up being able to take some measurements. I was not able to do a direct line out, as I could not get REW to hear the line in on my laptop. Because everyone is sleeping I did a near field measurement of one of the subs so I could use a lower volume. I had not yet upgraded to 3.0 so took some measurements with the old firmware and then upgraded and tested the new FW as well. I started with preset 2 which I don't use so it has no EQ set up, cross overs were at 80hz, I ran 4 sweeps in each test, FL, C, FR and LFE. Only the sub was turned on. In each set, the response is identical between the LF, C and RF. The LFE is the same as well up until 60hz where the 80hx crossover starts to kick in, The LFE does not start to roll off until later as I have the LFE lowpass up at 120 I believe. The 4 preset 2 and 8 Dirac measurements are overlaid in the attached screen shot. Looks like mine is working as expected in its current state. I don't know when I will be able to get around to upgrading to Dirac full so someone else will have to take the ball from here to test with a Dirac full filter set uploaded. Forgot to mention. Laptop connected by HDMI measurements taken with Umik, XMC in surround mode.  |
|
|
|
Post by nickwin on Apr 26, 2015 8:02:43 GMT -5
So I ended up being able to take some measurements. I was not able to do a direct line out, as I could not get REW to hear the line in on my laptop. Because everyone is sleeping I did a near field measurement of one of the subs so I could use a lower volume. I had not yet upgraded to 3.0 so took some measurements with the old firmware and then upgraded and tested the new FW as well. I started with preset 2 which I don't use so it has no EQ set up, cross overs were at 80hz, I ran 4 sweeps in each test, FL, C, FR and LFE. Only the sub was turned on. In each set, the response is identical between the LF, C and RF. The LFE is the same as well up until 60hz where the 80hx crossover starts to kick in, The LFE does not start to roll off until later as I have the LFE lowpass up at 120 I believe. The 4 preset 2 and 8 Dirac measurements are overlaid in the attached screen shot. Looks like mine is working as expected in its current state. I don't know when I will be able to get around to upgrading to Dirac full so someone else will have to take the ball from here to test with a Dirac full filter set uploaded. Forgot to mention. Laptop connected by HDMI measurements taken with Umik, XMC in surround mode. View AttachmentThanks for taking those measurements. Your different channels line up exactly as they should, just like mine did the other day. Maybe I need to reflash FW 3.0 or even send my unit in to Emotiva. If you could re try this when you get around to installing 3.0 that would be very appreciated. On on side note, do you know why your LFE is rolling off at 60hz while the routed bass is flat to 20? Actually your LFE looks very different than your routed bass below 30hz as well, and that's with the mains off? Maybe this is just a side effect of routing bass? IMO those two responses are different enough in overall shape that they should sound noticeably different. Your lfe plot has most of it's energy focused around 60hz where your routed bass is pretty linear all the way down to 20hz. By 20hz your lfe is down almost 10db while your routed bass remains flat. |
|
tubby
Emo VIPs
Route 2 in Weekapaug!!!
Posts: 408
|
Post by tubby on Apr 26, 2015 9:14:34 GMT -5
I don't think you are reading the graph correctly. Or I did not explain it well. The top group of lines (4 overlaid lines) is preset 2 (LF C RF & LFE). The bottom lines (8 overlaid lines) are dirac fw2.1 and 3.0, again LF C RF & LFE. The LFE response below 60 lines up fine.
I did do the test with 3.0 just not with a Dirac full measurement and load, since your problem seemed to come up after loading a filter set from dirac full I would like to see someone test that for you.
|
|
|
|
Post by nickwin on Apr 26, 2015 9:25:03 GMT -5
I don't think you are reading the graph correctly. Or I did not explain it well. The top group of lines (4 overlaid lines) is preset 2 (LF C RF & LFE). The bottom lines (8 overlaid lines) are dirac fw2.1 and 3.0, again LF C RF & LFE. The LFE response below 60 lines up fine. I did do the test with 3.0 just not with a Dirac full measurement and load, since your problem seemed to come up after loading a filter set from dirac full I would like to see someone test that for you. Gotcha, sorry, late night + hadn't had my coffee yet. Those plots look just like they should. |
|
|
|
Post by nickwin on Apr 26, 2015 9:32:02 GMT -5
I think I'm kind of sorta maybe starting to see the big picture here. I think the reason it was so hard for me to figure out what was going on when I started taking measurements is because I was experiencing 3 distinct problems at the same time.
1) Cancelation at the crossover: SOLVED. This one was definitely on the Dirac side. Its caused by the delay set by Dirac in its filter creation. I got the same huge suckout on my mains every time I ran Dirac. IMO Dirac assumes the hardware that it is implemented in will have a user adjustable delay. The "subwoofer distance tweak" is a pretty common last step when using an ARC system. If you have this problem with the XMC you either need subs with phase knobs or an external DSP, since delay is not adjustable in the Dirac preset.
2) Overall shape of routed bass response is different than LFE bass response. I honestly don't know if this is normal or not but it doesn't seem like it should be.
3) Routed bass response measures different depending on which speaker its routed from. This one might just be my unit or possibly even my computer or REW.
In my experience 2 and 3 get worse the more times I use the same project file to create and upload new filters. I might pull my mic out today and take the same measurements that tubby did just to be positive that Im not getting some kind of crazy speaker interaction. It would be awesome if someone else could measure there unit. When I first started using Dirac I talked to Emotive about this a few times and they didnt really push sending it in or reloading the FW, but if I'm the only one seeing this behavior I think thats an obvious next step.
|
|
|
|
Post by nickwin on Apr 26, 2015 10:37:35 GMT -5
Wow so I figured out why my LFE channel was measuring different than routed bass and its pretty stupid  It was simply the extra 10db on the LFE channel that was making the difference. My noise floor is 25-30db down from even my lowest measurements but it must have been enough to throw them off below 35hz. When I add 10db to the routed bass measurements they measure identical. Go figure. I can't explain the different results I got from that one set of filters, all I can think is that my SPL calibration somehow got off in REW and I was measuring at a higher SPL than I realized for that measurement set. I think this is problem solved. It appears the only real problem was the phase mismatch at the crossover and we have a work around for that. Back to listening. Attachments:
|
|
|
|
Post by socketman on Apr 26, 2015 12:00:50 GMT -5
nick what microphone are you using. If you have a Umik 1 the cal file will have sensitivity value and all your measurements in REW will be consistent.
|
|
|
|
Post by ansat on Apr 26, 2015 13:31:33 GMT -5
results incoming
|
|
It was simply the extra 10db on the LFE channel that was making the difference. My noise floor is 25-30db down from even my lowest measurements but it must have been enough to throw them off below 35hz. When I add 10db to the routed bass measurements they measure identical. Go figure. I can't explain the different results I got from that one set of filters, all I can think is that my SPL calibration somehow got off in REW and I was measuring at a higher SPL than I realized for that measurement set.
