Hopefully @Jbeck doesn’t mind. This is his before from DIrac. My comment I left him at AVSForum was his speaker response looks great. Sub level relative to speaker level just looks very low. My thought was redo the levels of speakers and sub and either raise the sub level so Dirac can do cuts instead of boosts. I also mentioned if you like the sound of your speakers without correction, then after you get the sub level fixed try just using Dirac below 500hz..
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
NAD M33 Streaming Amplifier Review
- Thread starter amirm
- Start date
Sorry to hear that.
When you set the levels for the subs/mains did you adjust the levels... ie raise the sub level so the before looks different?
My understanding is when Dirac is enabled, volume will be lower as it needs some headroom for boosting frequencies. I think it allocates up to 10db. So with Dirac enabled my understanding is you will need to raise your volume up to be the same volume as without Dirac enabled.
It seems you’re having multiple issues at one time which makes it difficult to diagnose the root cause. Based on your comments, there appear to be the following issues:
1. Dirac Live is not operating correctly and crashing. What computer are you using and what OS? What version of Dirac Live? I’m using 3.0.13 on a MacBook Air M1 with 16GB RAM and 1TB SSD. I’ve used both the supplied microphone and also a MiniDSP UMIK-1 pointing upwards with the appropriate correction files. Between the two microphones, I only notice a difference in the upper frequency range. Sometimes the M1 switches the input to the built in microphone so I used the M33 USB input for the calibration to be safe.
2. Subwoofer output has some issues. Suggest taking the subwoofer out of the loop for now until you resolve the other issues.
3. Soundstage is narrow. Suggest limiting the correction bandwidth to <500Hz to narrow down the potential problems.
4. Low volume output. Are the tone controls on? I notice that the tone controls being activated drops the volume significantly to give the amplifier headroom. Also what source is being used when the volume is low?
If after narrowing down to the essentials doesn’t pinpoint the problem, I’d suggest reaching out to Dirac Live as they seem to be quite responsive to client inquiries. NAD is usually pretty good but I’ve experienced slow responses due to covid. Good luck and let us know if you find the root cause.
pogo
Major Contributor
- Joined
- Sep 4, 2020
- Messages
- 1,303
- Likes
- 424
With DIRAC LIVE in operation, I have the following experience:
As soon as I want to improve my listening experience with DL, I get a lifeless and cloudy sound (even with my corrected problem area of 20-100Hz only). According to the DL manufacturer, there should be no deterioration! Interestingly, I am not alone with this experience using the Nad M33 & DL, as you can read in other forums.
I am assuming that the reproduction potential of high-quality loudspeakers in connection with the fast response of the M33 amplifier cannot be fully captured by measuring microphones (i used a UMIK-1 and the original supplied measuring microphone). The DIRAC LIVE corrected result sounds now like a mainstream loudspeaker which is unacceptable to my ears. Other users have improved their listening experience using speakers which do not sound like e.g. an Avantgarde Acoustic or a large B&W with suitable equipment.
And that could be the solution.
See below:
Earthworks FAQ - 02 / Technical
'Why do you make microphones that exceed 20kHz?'
Link: Earthworks FAQ
With such a microphone (type M50, …), the happening in the room can really be captured correctly, i.e. deliver optimal data for the correction calculation.
I will do my next attempt with a UMIK-2 instead of my UMIK-1. An improvement should be audible here.
In the meantime i correct with a self-generated flat curve and adjust it with the help of REW. With this solution I don't get a lifeless and cloudy sound.
Last edited:
Vasr
Major Contributor
- Joined
- Jun 27, 2020
- Messages
- 1,409
- Likes
- 1,926
Irrelevant assumption for purposes of this measurement. The microphone for these measurements isn't capturing some quality of music reproduction. It is a very simple decibel level measurement at each frequency played during the measurement sweep. Its accuracy is sufficiently high for any electronics and ensured by the vendor providing a calibrated mic and a sweep speed that is within the capabilities of the mic supplied. In addition, given the type of smoothing and averaging done to detect room modes and correct for them, the variances in SQ between electronics and speakers is irrelevant unless there are significant changes in the frequency response. But it will still be accurate enough for each speaker/electronics.
Assuming your observations are real (and you listen to both before and after at the same volume level as engaging Dirac drops the volume and a change in volume can have significant impact on perceived sound quality), there may be something else going on in the DSP processing that may be contributing to it. You can gather more data by doing a REW sweep with and without Dirac and see what comes out. if the differences are that obvious by ear, more likely than not there will be some visible and significant changes betwen the two measurements for each speaker (and another with both speakers together).
That will allow you to further investigate where the problem might be.
I don't know the UI on the NAD but there should be a way to turn on Dirac without any filters installed. See if this results in any audible difference. If it does then there is something going on in the DSP processing itself rather than anything from the Dirac generated filters.
With DIRAC LIVE in operation, I have the following experience:
As soon as I want to improve my listening experience with DL, I get a lifeless and cloudy sound (even with my corrected problem area of 20-100Hz only). According to the DL manufacturer, there should be no deterioration! Interestingly, I am not alone with this experience using the Nad M33 & DL, as you can read in other forums.
In the meantime i correct with a self-generated flat curve and adjust it with the help of REW. With this solution I don't get a lifeless and cloudy sound.
All that matters is that it sounds good to you, but I want to ask if you’ve tried other curves besides flat at your listening position?
The research from Harman shows that even trained listeners prefer some room gain in the lower octaves. NAD provides a house curve which follows this research.
When I first setup DL, I used a flat curve and found it lifeless, but then used the +8db Harman and NAD curves and found it much more to my liking.
Perhaps you can try the Harman or NAD curves and report back.
pogo
Major Contributor
- Joined
- Sep 4, 2020
- Messages
- 1,303
- Likes
- 424
The same lifeless and cloudy sound result with Harman or NAD curves. It has nothing to do with whether certain frequency ranges are reproduced louder or quieter!
Since DL tries to correct the impulse response and not just the frequency response, I have made the following measurements in the past:
These clearly show that the impulse response is corrected even above the desired correction range of 50-200Hz, in my case it is worsened.
I was able to suppress the impulse response correction using my self-generated flat curve with a quasi-optimal impulse response. The result (correction range of 50-100Hz) is light years better (no lifeless and cloudy sound) and no change above 100Hz can be seen!:
Since DL tries to correct the impulse response and not just the frequency response, I have made the following measurements in the past:
These clearly show that the impulse response is corrected even above the desired correction range of 50-200Hz, in my case it is worsened.
I was able to suppress the impulse response correction using my self-generated flat curve with a quasi-optimal impulse response. The result (correction range of 50-100Hz) is light years better (no lifeless and cloudy sound) and no change above 100Hz can be seen!:
Vasr
Major Contributor
- Joined
- Jun 27, 2020
- Messages
- 1,409
- Likes
- 1,926
I really have trouble understanding exactly what you are saying because of the terminology you are using.
To me, it just seems like the first curve was the result of a up to 600hz (could have been default setting) flat target correction applied (whether it was user error in not limiting to 200 or a bug in software I don't know). The rest of the spectrum is just measurement variance between two successive measurements.
And the second one is about 4db or so attenuation filter applied around 50hz and nothing else. Tonally it would sound much more bassy and boomy than the first one and perhaps the lack of that characteristic in the first is what makes it sound lifeless and cloudy to you.
I don't think the problem is with Dirac or the NAD.
pogo
Major Contributor
- Joined
- Sep 4, 2020
- Messages
- 1,303
- Likes
- 424
In general, I just wanted to highlight the problem that is probably caused by the impulse response correction from DL. It's not about more or less bassy or boomy sound.
The UMIK-1 was used as the input source for DL on the first curve. Only the range from 50-200Hz should be corrected via the correction window. There is also no measurement variance, as can be clearly seen from the second curve, since the upper frequency range is congruent here!
On the second curve, my flat curve (it's not a DL target curve!) was used as the input source for DL to suppress DL's impulse response correction. I can also correct my headphones on the M33 so well.
See also the procedure here:
Link
I've modified this procedure for my purposes.
I measured / generated a neutral frequency response curve (my flat curve) using a mixer mixing the two L / R preouts of the M33 to the left output channel, as described in the procedure previously, and had this signal A/D converted by the mixer (input source for Dirac Live).
The UMIK-1 was used as the input source for DL on the first curve. Only the range from 50-200Hz should be corrected via the correction window. There is also no measurement variance, as can be clearly seen from the second curve, since the upper frequency range is congruent here!
On the second curve, my flat curve (it's not a DL target curve!) was used as the input source for DL to suppress DL's impulse response correction. I can also correct my headphones on the M33 so well.
See also the procedure here:
Link
I've modified this procedure for my purposes.
I measured / generated a neutral frequency response curve (my flat curve) using a mixer mixing the two L / R preouts of the M33 to the left output channel, as described in the procedure previously, and had this signal A/D converted by the mixer (input source for Dirac Live).
Last edited:
Low weight would be the primary objective.
Vasr
Major Contributor
- Joined
- Jun 27, 2020
- Messages
- 1,409
- Likes
- 1,926
Clearly we are talking past each other. So, I will let someone who speaks the same language try to help you.
It's extremely difficult to help with Dirac Live without the full set of information:
1. Image of the microphone placed at the MLP and the general room showing the speakers placement.
2. Screen shot of the MLP sweep from within Dirac Live.
3. Screen shot of the filter screen showing the total measurements and what target curve is being applied.
4. REW measures in the MLP with and without Dirac Live.
The setup that you have is beyond my understanding so I think you'd be best served contacting Dirac Live and explaining what diagnosis you have done and what you're trying to determine.
When I look at my REW measures, it's clear that Dirac Live is only applying correction to the window range I specify, whether that be the default <500Hz or when I narrow it to 30-200Hz. Everything above and below the window remains the same (there is some +/-1db measurement variability).
I'm sorry I'm of no help, but what you're trying to do is beyond my comprehension.
pogo
Major Contributor
- Joined
- Sep 4, 2020
- Messages
- 1,303
- Likes
- 424
Don't get it wrong, because I am very satisfied with my flat curve solution and don't need any help. I have only confirmed the observation of Jbeck, which can also be read in other forums. In the past I asked myself why I get a lifeless and cloudy sound if I only have a range of 50-100Hz corrected by DL. My further research then showed that the impulse response correction still affects the entire frequency range. Then I came up with the idea of the flat curve, which effectively suppresses this correction. The result is terrific for me, as I now have my problem area fully under control and the lifeless and cloudy sound no longer occurs. I have downgraded my DL to a Devialet Sweet Room with better usability
Of course, I also wrote to NAD and have received feedback that they too were measuring with an Umik-1 and, after a long test, were able to roughly understand what I mean . I cannot say how it will go on (F/W update M33, S/W update DL, ...). Next I will try a UMIK-2 as described in post #287.
Not necessarily, that is more of a marketing requirementAnd an ICE that revs to 9,000 rpm
There is a difference between what is actually important to making a fast sports car and what journalists and fans have been lead to believe by marketing departments.
The GT3 is certainly one of the best of them because Porsche do actually know what is important. So do Ferrari and McLaren.
Bugatti, Lamborghini less so.
Hi Craig, as fellow 390DD owner contemplating upgrading to the M33, I thought you’d be interested in knowing the market for the 390DD. I placed mine online for sale here in Asia and it sold within a week!
I bought the 390DD and BluOS MDC for the equivalent of US$2,229 in 2015. I sold it for US$1,160. I’m rather pleased with its resale value. The little grey box cost only US$180/yr in depreciation. Most importantly, the buyer is super happy with the purchase.
I understand what pogo explaining. My issue is also this.
If I select any curve. Harmon or nad
I move the curtain far left so only correcting 30 to 200 hz everything sounds flat. Life less and no sound stage. Even when I don't apply correction above 200hz my max volume is cut by 20db. All I want to do is correct below 200hz but dirac muddied the sound.
I'm trying to learn how to take before and after measurements using rew.
I do know that DL works great on my c658. There is no loss of volume and I can specify correction in the < 200hz range.
Both the m33 and c658 are using my magnipan LRS speakers. So my previous question regarding dipole speakers are not needed.
Now I can't get DL 3.0.13 to run even after using it for days before. Same computer and deleting and new install. Crashes after first measurement. Had to revert to earlier version.
If I select any curve. Harmon or nad
I move the curtain far left so only correcting 30 to 200 hz everything sounds flat. Life less and no sound stage. Even when I don't apply correction above 200hz my max volume is cut by 20db. All I want to do is correct below 200hz but dirac muddied the sound.
I'm trying to learn how to take before and after measurements using rew.
I do know that DL works great on my c658. There is no loss of volume and I can specify correction in the < 200hz range.
Both the m33 and c658 are using my magnipan LRS speakers. So my previous question regarding dipole speakers are not needed.
Now I can't get DL 3.0.13 to run even after using it for days before. Same computer and deleting and new install. Crashes after first measurement. Had to revert to earlier version.
