|
|
Post by bestboy on May 5, 2015 12:58:01 GMT -5
So, what's going on with this crossover cross talk issue?
Can it be fixed by firmware? Is it possible to transfer the crossover points of the speakers from the XMC-1 over to the Dirac software in order to use them as lower boundaries for the filter creation?
Or is it maybe possible to create something like Dirac Live LE+ which includes the curtain feature of Dirac Full? This could then be used to apply the required boundary manually?
|
|
|
|
Post by tunatamer on May 5, 2015 13:43:09 GMT -5
So, what's going on with this crossover cross talk issue? Can it be fixed by firmware? Is it possible to transfer the crossover points of the speakers from the XMC-1 over to the Dirac software in order to use them as lower boundaries for the filter creation? Or is it maybe possible to create something like Dirac Live LE+ which includes the curtain feature of Dirac Full? This could then be used to apply the required boundary manually? I'd highly recommend springing for the extra 100 for full if you can afford it. Best money I've spent in recent memory (except for the XMC of course)  |
|
|
|
Post by nickwin on May 6, 2015 9:04:59 GMT -5
Hmm, it's not looking like Emotiva is going to response to this. Until we hear otherwise Im going to assume that what we've seen and what Ansat measured is by design and won't be "fixed".
IMO if this is the case the manual should be updated, but until then I'll say it, if your using a crossover, set the Dirac curtains of your mains to whatever crossover point you use. This will preven multiple filters from being applied to your sub.
Thank Ansat for the workaround.
Nick
|
|
|
|
Post by millst on May 6, 2015 11:09:01 GMT -5
Yeah, I wouldn't expect much support direct from Emotiva here. Yes, they do help periodically, but it's not consistent or thorough. Be persistent and get the right person on the phone. That's the only thing that's really worked for me.
-tm
|
|
|
|
Post by novisnick on May 6, 2015 11:15:39 GMT -5
Yeah, I wouldn't expect much support direct from Emotiva here. Yes, they do help periodically, but it's not consistent or thorough. Be persistent and get the right person on the phone. That's the only thing that's really worked for me. -tm Yes, tech support is on the phone, this is our/their forum. Now and then they check in and help out but this is for fun,,,,,mostly,,,,,tech support is business. Hope this clarifies the point.  |
|
|
|
Post by bestboy on May 6, 2015 13:31:00 GMT -5
But this crossover problem looks like a real issue. If it exists and the crossover redirection to the sub is done after the Dirac filters are applied, then this can lead to severe anomalies. With Dirac enabled for a 7.1 setup people can easily end up with a subwoofer channel that is boosted the full +10 dB over the entire subwoofer band. For the following table I made up some viable (?) filter values for satellite speakers and a default crossover of 80 Hz  |
|
|
|
Post by nickwin on May 6, 2015 15:38:25 GMT -5
Yeah, I wouldn't expect much support direct from Emotiva here. Yes, they do help periodically, but it's not consistent or thorough. Be persistent and get the right person on the phone. That's the only thing that's really worked for me. -tm Yes, tech support is on the phone, this is our/their forum. Now and then they check in and help out but this is for fun,,,,,mostly,,,,,tech support is business. Hope this clarifies the point. I've talked on the phone with one Emotive tech, had email correspondence with another, and sent out 2 or 3 emails to other people at Emotive with no reply no far. The ones I talked to couldn't explain what is happening here. Ive also talked to two Dirac developers. This thread is not the only way I have been going about this. Maybe someone else would have better luck. I think I might be black listed after getting reprimanded by Dan |
|
|
|
Post by snafujg on May 6, 2015 22:47:13 GMT -5
I will say I've always said the test signals should account for the crossover like the full version does when you run Dirac on the PC. I've often thought the issue in this thread would be a problem, but was told by Emotiva it's not.....Short of every xmc owner calling them, though, I don't see this getting looked at. The company line seems to be that a majority of their customers are happy with it as is so they're not going to look into it. And that's not just with this threads issue. I get the impression that's their take on most of the perceived issues posted out here. The microphone cal file, the lack of bass, hot center/surround channel, Dirac full crashing.....Its unfortunate because they have a great product here. All we're trying to do is make it that much better....
|
|
|
|
Post by millst on May 7, 2015 2:23:48 GMT -5
Yeah, it sucks. One of the reasons AVS people are quick to trash Emotiva. Meanwhile, from what I've seen on a thread over there, MiniDSP seems very quick to respond to issues and release updated software.
-tm
|
|
|
|
Post by nickwin on May 7, 2015 8:11:11 GMT -5
In the past Emotive has been good about fixing problems and making improvements when needed as well. At this point Im not even asking if they will change anything, I just want to know how it works.
As far as Emotive looking into it, they shouldn't have too. Ray and maybe Lonnie too know exactly how routed bass is handled on the XMC. They did design it after all. They should be able to say whether this is behaving as designed or if something is wrong.
I really disagree with the "its good enough because lots of people are happy with it" philosophy. Yes, it is good, but that doesn't mean that it doesn't have flaws that if fixed would make it better. We know from bench tests that the hardware is exceptionally capable.
|
|
|
|
Post by nickwin on May 7, 2015 10:09:01 GMT -5
UPDATE: alright, I've got some new info from Ray. He says that the Dirac filters for the mains should not be getting applied to the routed bass. Dirac filters and delay are the last two things done by the DSP, both are after the crossover. Thats really all I have so far. Its looking like we are seeing some inconsistencies and/or unexpected behavior. Does anyone that uses REW still have FW2.0? I'm wondering if this was introduced with 3.0.
|
|
|
|
Post by Kir on May 7, 2015 12:28:17 GMT -5
... Dirac filters and delay are the last two things done by the DSP, both are after the crossover... Inconsistent indeed. All I've been reading before indicated that when Dirac tests run, they ignore the crossover altogether (?). |
|
|
|
Post by geebo on May 7, 2015 12:49:07 GMT -5
... Dirac filters and delay are the last two things done by the DSP, both are after the crossover... Inconsistent indeed. All I've been reading before indicated that when Dirac tests run, they ignore the crossover altogether (?). They are ignored for the purpose of measuring and filter creation. But the resulting filters are applied after the crossover has been. |
|
|
|
Post by raydyo on May 7, 2015 13:20:40 GMT -5
This issue just got pointed out to me via customer service. And it definitely had me in a panic for a while. I don't have the time to review all of the posts. I am focusing entirely on Tony's graphs top of page 2. They seem to point to a serious problem. The only trouble is, I can't duplicate the results. My measurement is attached. I tried to set this up so it would look as close as possible to Tony's plots. As you can plainly see, there is no "crossover crosstalk" in this plot. Either there is some pilot error on Tony's part or there is some subtle difference in setup that causes a not so subtle difference in result. I am using a freshly factory reset unit with the following changes in Dirac preset all speakers 120Hz crossover 12dB/octave sub crossover. These changes were made so that the subwoofer response from the mains would be exactly the same as the hard-wired LFE response. This has no effect on Dirac filters. Note that +10dB LFE is required by Dolby and dts - completely normal. I ran the same plots but with LFE 10dB lower and the subwoofer curves for LFE and Left Front (and all other speakers) overlapped exactly. Tony - if you wouldn't mind - could you repeat your measurements with my settings and see if you still see this issue? If not, start restoring your settings one at a time and see when the problem returns. My source is 7.1 PCM over HDMI. Measured at XMC-1 balanced outputs. What was yours? Thanks -ray Attachments:
|
|
|
|
Post by Kir on May 7, 2015 13:34:50 GMT -5
Inconsistent indeed. All I've been reading before indicated that when Dirac tests run, they ignore the crossover altogether (?). They are ignored for the purpose of measuring and filter creation. But the resulting filters are applied after the crossover has been. If the filters are created ignoring the crossover, but then applied after the crossover, the end result would be wrong. |
|
|
|
Post by geebo on May 7, 2015 13:43:04 GMT -5
They are ignored for the purpose of measuring and filter creation. But the resulting filters are applied after the crossover has been. If the filters are created ignoring the crossover, but then applied after the crossover, the end result would be wrong. Can't comment on that. We've been told that Dirac measures each channel as if it were full range. No crossover or speaker size is taken into account. Crossovers and speaker sizes can then be changed without having to remeasure. So why would the result necessarily be wrong because of that. Theoretically, it shouldn't matter if the crossover and slopes are properly implemented. |
|
edrummereasye
Sensei
"This aggression will not stand, man!"
Posts: 438
|
Post by edrummereasye on May 7, 2015 14:00:54 GMT -5
If the filters are created ignoring the crossover, but then applied after the crossover, the end result would be wrong. Can't comment on that. We've been told that Dirac measures each channel as if it were full range. No crossover or speaker size is taken into account. Crossovers and speaker sizes can then be changed without having to remeasure. So why would the result necessarily be wrong because of that. Theoretically, it shouldn't matter if the crossover and slopes are properly implemented. Right...e.g., on say the L/R fronts...the filter might go all the way down to 20 Hz...but with an 80 Hz xover, it will be applied to the reduced signal below 80...until some point far below 80, where it's essentially being applied to nothing...because all that content was re-routed to the sub(s) before Dirac got hold of it...sounds right to me... The problem would be if the filters were applied _before_ the crossover...then the filter for the L/R fronts would be getting applied to that signal, from 20-20K or whatever....then the post-filter 20-80 Hz signal would get re-directed to the subs, in all their (wrong-)filtered glory... |
|
|
|
Post by nickwin on May 7, 2015 14:04:36 GMT -5
They are ignored for the purpose of measuring and filter creation. But the resulting filters are applied after the crossover has been. If the filters are created ignoring the crossover, but then applied after the crossover, the end result would be wrong. I think your looking at this wrong. Filters after the crossover means thay each speaker gets the filters intended for that speaker. If the filters were before the crossover, then you would get filters from the mains passed to the sub. |
|
|
|
Post by millst on May 7, 2015 14:13:06 GMT -5
The filters are correcting the response of the speaker in the room. They should be applied after the crossover. If you applied the filter before the crossover, then the filters for the mains would be applied to the mains content being redirected to the sub. That's the concern brought up in this thread.
The negative with this approach is that the filters are not addressing the combined response of the sub and mains. However, most agree the benefits outweigh this downside.
The miniDSP nanoAVR originally didn't have the capability to apply crossovers before the filters and was criticized for it. I believe a software update added that in, which syncs up with the rest of the product line.
-tm
|
|
|
|
Post by nickwin on May 7, 2015 14:36:54 GMT -5
This issue just got pointed out to me via customer service. And it definitely had me in a panic for a while. I don't have the time to review all of the posts. I am focusing entirely on Tony's graphs top of page 2. They seem to point to a serious problem. The only trouble is, I can't duplicate the results. My measurement is attached. I tried to set this up so it would look as close as possible to Tony's plots. As you can plainly see, there is no "crossover crosstalk" in this plot. Either there is some pilot error on Tony's part or there is some subtle difference in setup that causes a not so subtle difference in result. I am using a freshly factory reset unit with the following changes in Dirac preset all speakers 120Hz crossover 12dB/octave sub crossover. These changes were made so that the subwoofer response from the mains would be exactly the same as the hard-wired LFE response. This has no effect on Dirac filters. Note that +10dB LFE is required by Dolby and dts - completely normal. I ran the same plots but with LFE 10dB lower and the subwoofer curves for LFE and Left Front (and all other speakers) overlapped exactly. Tony - if you wouldn't mind - could you repeat your measurements with my settings and see if you still see this issue? If not, start restoring your settings one at a time and see when the problem returns. My source is 7.1 PCM over HDMI. Measured at XMC-1 balanced outputs. What was yours? Thanks -ray Thanks for taking the time to look into this and comment Ray! Here are a couple measurements I took. This is an in room speaker measurement, but it seems to the same or some other odd behavior. My AV rack is just to much of a PITA to pull my XMC out and take line level measurements. I did a factory reset, set my speaker size to small and crossovers to 90hz, other than that I didn't change any settings. Purple is LFE and blue is routed bass from my front left. Attachments:
|
|
