Home Theater Forum and Systems banner

REW plots comparing Dirac PC software compared to the XMC-1's Dirac software

5K views 8 replies 3 participants last post by  jtalden 
#1 ·
I posted the response below at Emotiva's website, but I thought I'd post here as I'm looking for more technical experience for my learning benefit (I'm an EE although audio isn't my concentration). I am interested in improving my system even more, but I suspect I crossed the diminishing returns threshold a long time ago and this is the last improvement I'll see for a very long time! It was very neat to get the improvement I did.

Dirac Live full software ($750) versus Emotiva XMC-1 Dirac Live "Full" upgrade ($99): Dirac Live $750 software wins?

What I did: I used the XMC-1's Preset 1 speaker preset and manually set level and time aligned my subs (prior to using Dirac Live trial software on my PC). This involved taking 20 measurements and picking the best front speaker/sub time alignment (see REW Preset 1 file at link above) by adjusting crossover slope in the XMC-1 and subwoofer distance. I then used HDMI to connect my HTPC to the XMC-1 and let Dirac calibrate the left channel (left speaker crossed at 80Hz to dual mono subs), right channel (right speaker crossed at 80Hz to dual mono subs), etc... I used an Steinberg UR22 mkII and an Earthworks M30 mic for REW and for Dirac. However, the XMC-1's on-board Dirac was calibrated using an UMIK-1 USB mic (which I no longer have).

System details: https://drive.google.com/open?id=0B1J0a4OV_WGLUVZBNUJBR29LTEU
All speakers crossed at 80Hz with 12dB slope to dual JL Audio F212 subs. REW files are also available for download above. Note there is a 200Hz room issue that I've never been able to fix (causes the one phase shift on right channel and a dip on the left channel with no phase issue). My room is very well treated though, just a FROG with short sloped ceilings.

It's crazy how one algorithm is targeting the low end more (PC Dirac Live trial software), while the XMC-1 Dirac Live "Full" software targets the high end more. One would definitely want the low-end targeted more, and the high-end left alone as much as possible, so I think the PC version of Dirac is actually better. For the low frequencies better phase coherence, lower 3dB point and a sweeter looking waterfall is a nice improvement!

Really makes one wonder how perfect the XMC-1 would be if we could adjust the sub channel delay following calibration (like the $750 Dirac software will do) or if the XMC-1 applied the crossover upstream of Dirac...

Potentially, I could try to tackle the 200Hz issue in REW by exporting filters to the XMC-1's Preset 1, then rerun Dirac on my PC. I could also setup the optimum gain structure for levels (e.g. speaker trim) prior to running Dirac on the PC. I don't think I've seen a better waterfall plot anywhere though, so I'm tempted to enjoy the bass and leave things alone for a while.

I'll have to enjoy my Dirac trial (and even better bass) since I'm unsure of paying the $750 for the software!?! Plus I'd have to let my PC software decode HD audio formats instead of using bitstreamed output to the XMC-1. I'm using KODI now to do this and it sounds excellent, but when atmos/dts-x comes to the XMC-1, I will have to figure out something else... It is neat that the software version doesn't downsample HD audio tracks to 48kHz (like the XMC-1 does when Dirac speaker preset is selected), and it even uses 32 bits to do the math (and my PC's processor) so there's less truncation. This bass improvement, while minor, was probably worth the $750, and I think it subjectively sounds clearer (but I think it's my mind playing tricks on me although it could be the gain structure too as one would think the XMC-1 would have more headroom with Dirac disengaged).

OLD Results (blue) using Dirac speaker mode on the XMC-1 Versus using Preset 1 on the XMC-1 with Dirac software (purple) are attached.
 

Attachments

See less See more
6
#2 ·
Very nice! They both look excellent to me. From a measurements perspective, this is outstanding.

If you are still taking measurements, I would be interested to see 2 additional measurements. Both measurements taken using one of the 2 mains (FL or FR) with the SWs disabled. The mic to be placed on the TW axis and about 0.75m meters from the TW. One of the 2 measurements would be with Dirac filter active (either Dirac filter) and the other with the Dirac filter disabled, so basically; 'Dirac on' Vs 'Dirac off'.

The reason I am interested is just for my own understanding. There is lots of expert advice that indicates the best EQ scheme is not to EQ above the Schroeder Frequency. Many seem to be critical of PEQ solutions in the MR and higher frequencies and yet many also seem to accept Dirac and some other full range EQ solutions. In my limited experience the range >800 Hz is not really impacted very differently whether it is EQed or not. The 200-800 Hz range however is very different for many of us and I suspect Dirac has made some pretty significant adjustments in this range. These 2 measurements would let me get a sense of the how aggressive your Dirac EQ actually is. Your room is obviously very well treated so the correction may be much less than I expect. That is what I want to get better idea of. I am assuming you are very please with the sound of voices with both these 2 Dirac solutions and that they both seem significantly improved Vs Dirac disabled.

Thanks for sharing your results. It's a really impressive setup.
 
#5 · (Edited)
Thanks! You helped me before on teaching me about time alignment, so I will definitely upload the measurements in the next few days. Too hot right now to turn off the air conditioner though! I assume TW = tweeter?

I assume you are trying to see if Dirac is correcting any loudspeaker design issues. I think Dirac is for sure correcting some phase issues that seem to occur around/at the main speaker's crossover points. No doubt this will show up at a close distance (which would rule out the reflections off furniture, etc...).
One thing to note on the Dirac on versus off, the Dirac off measurements at the MLP are the bottom ones in this REW file: https://drive.google.com/open?id=0B1J0a4OV_WGLQUNKZGpOWWpmSjQ

For others, note that Audyssey and Dirac (and probably others) can actually correct time domain issues and are far beyond the parametric EQ these acoustic professionals are thinking of.

I would say from a cost doesn't matter standpoint, if I were to actively triamp my floor standing speakers (the right way by taking the crossovers out of my speaker cabinets), this would be the best way to get rid of these speaker crossover anomalies (in lieu of using Dirac above 300Hz or so). However, while I bought some Ashly XR1001's from ebay to play with (analog crossovers, but maybe a nice DSP later to adjust time delays like an Electrovoice DC-One or a Yamaha SP2060), I don't have room for three amps per speaker right now. I do want to triamp them; they are technically 3.5 way speakers, but I was going to just make them 3 way speakers crossed to my sub as they are now. If I did this next week (have spare 2 channel amps for TW and MR), I could take the center channel out (to make room) and just go with a phantom center. What do you think? My Atmos amps are just sitting there until the XMC-1 Atmos upgrade comes next year...

Here is the spec sheet for my speakers: http://images.klipsch.com/p39f_-_Spec_Sheet_635738653350358000.pdf

Below are images of the MLP measurements for the front right with the manually time aligned 80Hz 12dB/octave crossover to my subs, with no Dirac. I think these will give you a great glimpse into what the PC Dirac Live trial software is doing in an ultra well treated room. Note that the crossovers listed on the spec sheet seem to align with the two phase issues at 700Hz and 2.5kHz. I'm assuming you went to my google drive and already noticed this; definitely you are very wise for wanting to see measurements closer to the speakers to verify these crossover anomalies.
 

Attachments

#3 ·
I can't say much to the plots since I'm mainly a 12v audio guy but the comment from jtalden about EQing above the Schroeder Frequency (800Hz) is intriguing. I'd like to know more about this, maybe with a 12v spin on it if you don't mind, maybe in another post so I don't thread hack!

As for the OP, bravo for having two of those JL beasts! I'm super jealous since I've heard one in a demo and was very impressed.
 
#6 ·
Yes, I meant the tweeter. Any mic height between the midrange axis and the tweeter axis if fine for this purpose. I usually split the difference. Most manufacturers and reviewers seem to use the tweeter axis.
I think the smoothness of the SPL and house curve are far more important than the phase. My substandard hearing can still easily detect relatively small changes in that regard. I am not able to hear any difference due to removal of the overall phase rotation. Many others indicate they can. I did think I could tell a noticeable improvement when the XO handoff provided close phase tracking in my 2 kHz XO. So possibly close phase tracking of the 2 drivers is significant at least for midrange XOs. I would need to specifically ABX test to be sure as this was a more subtle effect for me and may only be my imagination. I haven't seen any experts specifically stress this. It does show up in several speaker manufacturers advertising as a sales point, but I don't know if it is mostly just a sales pitch, or if it actual improves sound quality. I think there is a case that the horizontal dispersion uniformity would be significantly improved and that is now being mentioned as an important factor. I try to provide close phase tracking for all XOs, but I would expect the midrange XO to be the more important than SW or super-tweeter XOs. It appears the Dirac filter is very able to correct any XO range phase inconsistencies at the LP. I'm unsure if it can effectively fix the horizontal off axis dispersion as well if the passive XO does not already provide close phase tracking. I'm thinking it can't based on my limited understanding.
 
#8 · (Edited)
I was reading Elliot's website at westhost the week before last, but it is down now, hopefully it comes back up. He listed other benefits to bi-amping and tri-amping such as improved damping factor, less interference of back EMF from the woofer affecting tweeter and mid, less IMD although this is only audible for low frequencies and not as bad if you have a good amp, etc...

I am not sure I can hear the difference in phase for HF and MF. Keep in mind though, my speakers are supposed to have very nice crossovers internally, and they appear to be. I suspect since this was Klipsch's flagship that took them years to design that even the changing impedance of the woofers are accounted for in the crossover design.

The Dirac Audio Processor (DAP) software has an on/off button for the Dirac filter; it even keeps volume level the same. Honestly, the only difference I hear is in the low-end when I click this on/off button. The mid-range sounds great no matter if the filter is on or off. I remember when you helped me get my front and sub's phase to track better at the sub crossover last year. I heard an immediate improvement and I'm still grateful! Definitely the bass seemed to blend better with the main speakers and sounded more life-like.
 
#9 ·
Yes, I wasn't thinking straight when I indicated that the SW XO phase tracking may not be not as important as the midrange XO. It is probably more important. I too can easily hear differences when the timing is changed.

It's been several years since I actually made any changes to my bass range. Even then I only clearly remember testing the tradeoff between close phase tracking through the XO Range (smoothly increasing GD) Vs allowing the phase to cross at the XO point (providing less overall GD). To do that the SWs delay is reduced by 1/2 WL and the polarity of the SWs is reversed. This alternate alignment still provides phase agreement at the XO point and excellent SPL support throughout the XO range. The bass bandpass arrival time is more in line with the MR bandpass arrival time (less overall GD). In the XO range though, the phase delay increases to 90° at the XO frequency and then falls back near 0°. I could easily hear a difference, but both sounded very good. In some ways the alternate timing sounded more dynamic/impressive in my room. I did settle on the close phase tracking option as the more conventional solution and eventually added a rePhase phase correction filter to remove the overall phase rotation. That change with the rePhase filter is the one that I could not clearly detect.

I haven't actually used a truly mistimed SW XO setting in a long time. Is difficult to compare those situations because it results in significant SPL variation in the XO range. Thus, there are expected sound quality differences. If we use different EQ settings to bring the SPL back in line then; is any sound difference due to the mistiming, or the different more aggressive EQ used? Comparisons that way are problematic and I have no recent experience to draw from. It's best to just avoid those situations.

I have also seen longer lists of why active DSP XO is superior. It sure suits my situation, but that is more for experimentation flexibility not necessarily sound quality. We need to remember that a well designed passive XO is not the normal fair. The overall cost can be significant for a good design. There are plenty of well designed, highly regarded speakers with passive XOs. I have been very impressed with several of them. If both methods are implemented well I would not necessary expect a clear sonic advantage for a given situation. To me it would be a very difficult, if not impossible subject, to address in an objective way.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top