Home Theater Forum and Systems banner

1 - 13 of 13 Posts

·
Registered
Joined
·
34 Posts
Discussion Starter #1
I've had this NAD amp for a couple of years now. I'm also fairly new to REW, but I think I'm past being a newbie. By way of a bit of background, most of my use (stereo) of the system having this amp involves a Raspberry Pi with a HiFiBerry Digi+ board running Pi MusicBox. Over my LAN I retrieve FLAC files from my NAS and stream lower-res music from the Internet, including Spotify. From the Digi+ I go TOSLink to a MiniDSP NanoDigi and then via SPDIF coax to the NAD. Thus, I stay in the digital domain right up to the bitter end when we must go analog to the speakers. It's a kind of expensive amp, but I chose it so I wouldn't need to have a separate DAC and I liked the idea of remaining in the digital domain for as long as possible. Also, theC390DD is module upgrade-able/update-able with 3 slots, one of which has a USB input card from a computer and 2 are empty. This all works well for me, and although I don't usually use the USB input, it does work. This is reminiscent of the old integrated vs separates issue of past years and I went with integrated. The jury is still out on whether I think I made the right move.

This NAD amp has built-in rudimentary EQ capability with 6 non-parametric bands, 40/60/90/120/180/240 hz each with -12db to +4 db range in 2 dB steps. For each frequency, there is also a Narrow/Wide toggle setting which supposedly sets Q, although I've not seen any documentation giving specifics. I have just updated the amp's firmware to V.2.88

One time when I must use the USB input is when I work with the amazing REW. I use a UMIK mic and load its calibration file. Both the mic and my NAD amp are plugged into my laptop when I work with REW and I can do sweeps and measurements etc. without difficulty. After experimenting for a while, I decided to get serious about EQ'ing my system. Having read Wayne's article here about a "room curve", I decided I would EQ for my sub(*) using the settings in the amp an then in a subsequent step, EQ for room curves using the nanoDigi which has 5 parametric settings and for which REW knows how to create and export filters.

But first, I thought I would experiment a bit with the NAD EQ settings using REW to better understand what effect they were having on the sounds coming from my speakers. Now I am no expert here and I could be missing something or getting some terminology wrong. Please correct me if so. But after some experiments I find myself doubting whether the NAD is actually doing what it claims.

I include here 2 REW measurement sessions.

In one of these, I show 12 curves, 2 wide and 2 narrw for each of the frequencies 40/120/240; 1 at -12 dB and 1 at +4dB. In each case the otter 5 are set to zero.
Test EQs 40 120 and 240 HI and LO.jpg
I have a big room node at about 130 hz, so I tried the second test where I swept a narrower range and worked with just the 180 and 240 hz filters.
Test 160 to 260.jpg
In both graphs, I also did a sweep with all EQs at zero for a baseline.

Here's where I need help. I cannot make sense of these measurements. Every change of a parameter did have a noticeable effect, and boosts or cuts did seem to boost/cut. But the frequencies don't seem to make any sense. The curves more or less move up and down in SPL, but otherwise look too similar.

I tried to upload the REW data files for these pics, but failed several times with this message:

Code:
Request Time-out

Server timeout waiting for the HTTP request from the client.
I will provide on request.


I intend to send NAD a message about this with a link to this thread to get their opinion/advice. I'll keep you posted. Any comments from others here at the shack would be much appreciated.

(*) I am not actually EQ'ing just my sub because I cannot easily get to it alone since my amp is all-digital. I would need a separate DAC solely for the purpose of sending a signal to the sub. I don't want to spend that money. I have a pair of Axiom M80's and I'm trying to EQ the low frequencies without separating. I have my amp crossing over at 75 hz at the moment. I intend to pursue this further, but I'm kinda hung up on this current problem ... I can't trust my amp to be doing what it claims to do.
Thanks,

Tim in Owen Sound, Canada

P.S. Can you tell I'm having fun? That I'm a bit obsessive? That I'm retired?
 

·
Premium Member
Joined
·
6,711 Posts
If you have the line out option on the NAD you could use a soundcard analog input to measure the NAD's outputs directly to see what the filters are doing, failing that rig up an attenuator on the speaker outputs.
 

·
Registered
Joined
·
34 Posts
Discussion Starter #3
If you have the line out option on the NAD you could use a soundcard analog input to measure the NAD's outputs directly to see what the filters are doing, failing that rig up an attenuator on the speaker outputs.
Thanks John. The NAD has no analog either in or out. It does have one each of SPDIF and TOSlink outputs that I believe mirror the corresponding inputs.

My central point with my post is that in playing with the EQ parms of the NAD, I am not getting the results I expected, not by hearing but by measuring. When I ask for a 12 dB cut at 80 hz, I expect to be able to see it when compared to a graph with +/- o dB at the same frequency. I would also expect to see some difference in the spread of the measurement between Narrow and Wide settings.

Instead, all my readings seem to be variations on the same pattern. That pattern itself seems almost random.

In making my post, I had hoped to upload the REW measurement files also so you and others could load them into REW yourself, but I got errors trying to upload. The images of the graphs are too busy to be of much use by themselves. It's hard to see which plot is which and the legends are truncated. I will try uploading again in a few minutes in a new reply to the main thread.

While I'm replying to you John, may I say how impressed I am with your work building REW. I taught college programming for 25 years prior to retiring, including Java programming. There has to be a lot of code in REW judging by the sophistication of the human-computer interface and what are obviously pretty heavy calculations behind that. The fact that it's all running with such speed on a Java virtual machine makes it even more impressive. Do you have any post(s) here at HTS that speak of how REW is designed and built. For example, what do you use for a development IDE? Perhaps Eclipse?

Cheers!
Tim
 

·
Registered
Joined
·
34 Posts
Discussion Starter #4 (Edited)
A quick additional note. Because the NAD amplifier is not one that REW supports as a DSP, I must manually transcribe REW's calculated filter parameters over to the amp. There is only one set of parameters, so it's not easy to do an A/B comparison .. takes too long to switch. The EQ capability of the amp is pretty rudimentary. It is not parametric. For that reason, I thought I might be able to use REW's calculated Q values at the fixed frequencies of the amp to determine whether I should specify Narrow or Wide. The fact that I could find no documentation giving numeric meanings to Narrow and Wide was another reason for conducting my experiments.

When I bought the amp, the description on NAD's website said that the capability to do room EQ was forthcoming. I was thinking something along the lies of Audyssey. What materialized instead was a set of 6 FLAC files and documentation instructing how to set the EQ parameters of the amp using a sound meter and listening. Here's a quote from a paragraph in the documentation:

"The objective of this Room Equalization Setup is to try and subtract standing waves or “Room Boom”, we are not trying to add more bass response to speakers that lack deep bass extension. Most speaker and
room combinations usually have a roll off between 60Hz or 40Hz. What Room EQ will accomplish is the
‘unmasking’ of lower frequencies. Bass will now sound ‘tighter’ and better defined. The ‘Pitch’ of the
bass notes should be clearer after the equalization. Don’t be surprised if most frequencies do not require
any adjustment. There are usually only one or two frequencies where standing waves occur for a given
room."​

I must say however that with all my adjustments, I never really felt I was achieving a better equalization of low frequencies . That's the main reason I chose to conduct these experiments ... I needed to see what effect changes at specific frequencies had on the measurements. And I'm not very pleased with the findings, assuming I am understanding correctly. I have sent a message to NAD asking that they view this thread and get back to me.

For anyone interested in loading my REW measurements, here are links to the 2 files in my Dropbox.

https://www.dropbox.com/s/9xqn21901xotse1/Tests 160 to 260 EQ 180 and EQ 240 HI and LO.mdat?dl=0

https://www.dropbox.com/s/q7yxen7o3i5e0vr/EQs 40 120 and 240 HI and LO.mdat?dl=0

I tried a couple of different options for uploading the .mdat files without success.

Merry Christmas All !!

Tim P:eek:lddude:
 

·
Premium Member
Joined
·
6,711 Posts
If the mdat is very large the upload may time out or fail. You could try zipping it, or save a few of the measurements in a separate mdat.

If you put a 20 dB (say) attenuator on the speaker outputs you could connect that to a soundcard input, just be careful with the volume setting.

There's quite a lot of code :). Several hundred class files and about 6 MB of Java source (many of my class files really are much larger than they probably should be). I picked Java all those years ago because I thought it would be interesting to learn an OO language and the cross platform support was attractive. I use netbeans.
 

·
Premium Member
Joined
·
6,711 Posts
There is a lot of low frequency distortion in those measurements and the measurement levels, if they are correct, are very high. Have you selected the input the UMIK is connected to in REW's soundcard preferences? That is needed for REW to access the volume control and compensate for volume setting when calculating SPL. If those really are the levels you are using I'd suggest going much lower, around the 75 dB range, and use a longer sweep - 256k gives good results for most purposes.

What speaker/sub setup are you using for measurement?
 

·
Premium Member
Joined
·
1,562 Posts
Hi Tim,
I looked a the one file with the wide enough freq range to see the filter shape. I did not look at all the measurements, just the 'All EQ at Zero', '120Hz -12dB Wide' and '120Hz -12dB Narrow'.

With all the room modes and noise in the signal at the LP it is not easy to see the filter impact clearly without some settings changes. I did 'A/B math' to compare each of the 2 filters to the reference 'All EQ at Zero' measurement. I also applied some frequency dependent smoothing to reduce the impact of the room.

Below you can see the results the 'Ref' trace is Red, the 'Wide' trace is Green and the 'Narrow' trace is Blue.

My observations are:
> The 40Hz -12dB filter was also left active for these 120Hz filter measurements. I am not sure if it is the 40Hz Wide or narrow 40Hz filter that is active, but it has a slight impact on the 120Hz filter shape so we must take that into account for the analysis.
> Both The -12dB 120Hz Wide and Narrow filters are indeed -12dB SPL compared to the 'Ref'.
> The 120Hz 'Wide' filter is indeed a wider impact (Lower Q) than the 120Hz 'Narrow' filter.

It looks like the NAD 120Hz filters are working as advertised.

Capture.PNG
 

·
Registered
Joined
·
34 Posts
Discussion Starter #9 (Edited)
There's quite a lot of code :). Several hundred class files and about 6 MB of Java source (many of my class files really are much larger than they probably should be). I picked Java all those years ago because I thought it would be interesting to learn an OO language and the cross platform support was attractive. I use netbeans.
I think Java has lived up to expectations for cross platform support. It might interest you to know (assuming you don't already) of the Scala programming language. Before I retired in 2008, I began to feel that Java was starting to become a bit "weedy" for want of a better word ... I never much liked C/C++ syntax and very useful concepts like 'generics' were being added but for compatibility required what I felt was cumbersome syntax. One day I searched for other programming languages that compiled to Java bytecode and Scala turned up. Scala has the advantage that it can use existing Java classes. But it also has its own Java-like but cleaner syntax and it supports functional programming. For example, it has type inference so you don't have to declare counter to be an int. Simply assign counter an integer value and Scala will infer that counter is an int. But Scala is a strongly typed, compiled language so once you have caused Scala to infer a type for a variable, it will prevent you from assigning a value of a different type; just as though you had declared it.
 

·
Registered
Joined
·
34 Posts
Discussion Starter #10 (Edited)
Thanks to everyone for your very helpful and informative observations. You have given me food for thought and further learning and restored most of my joy in owing the NAD C390DD. I plan to look more carefully at SPL calibration and levels for testing. Also RTA. And jtalden's massaging of my REW data is very impressive! You took a scrambled mess and turned out a simple graph that even showed I must have been a bit careless in changing EQ values between samples.

I will also let NAD know that my confidence in their Room EQ settings is (quasi-)restored. I will try to find out from them the (fuzzy) Q value that distiguishes Wide from Narrow.

Hopefully as I do these things, I will begin to get the improvements in my system that I seek to get me closer to hearing Diana Krall live in my listening room.

Cheers!
Tim

P.S. John, since I suspect you will read this, IMHO it would be a great addition to REW if the output device could be a digital stream. If I had this on my system, I would envisage
  1. Set a REW preference that results in an ip address that REW will stream to when I press "Start Measuring".
  2. Perhaps one could specify a format via choice: FLAC, ALAC, WAV etc
  3. In my system that opens streams for processing, I would ask to stream from REW's ip address.
  4. When all systems were 'go', I would click Start Measuring and hear the test signal through my speakers.
This would be no easy task, but I believe it would make REW even more appealing to those who see streamed audio as the/a way of the future.

A (perhaps) simpler alternative for sweeps or other finite-duration tests would be to save the sweep in a FLAC (for e.g.) file. Then an added option under Measure would be to Start/Stop a Listen function. The user would click this and then play the sweep file through their system, then click Stop Listen.

Of course with a new generation going vinyl, perhaps this is a moot point. Wait till those people have played their vinyl recordings enough times that listening starts to feel like they're sitting round a crackling fire. :rubeyes:
 

·
Registered
Joined
·
34 Posts
Discussion Starter #11
I did 'A/B math' to compare each of the 2 filters to the reference 'All EQ at Zero' measurement. I also applied some frequency dependent smoothing to reduce the impact of the room.
jtalden, when you have a moment, if you could point me somewhere or by brief reply, I would like to know how you did your magic.

Thanks,
Tim
 

·
Premium Member
Joined
·
1,562 Posts
Tim,
General Information regarding 'Trace Arithmetic' is found in 'REW Help' under; Help/Graph Panel/All SPL/. The gear shaped button on the 'All SPL' panel opens the 'Controls' panel which provides access to the 'Trace Arithmetic' operations.

I also applied a 'Frequency Dependent Window' Using 'IR Window' (tools menu button) to help reduce the room influence on the traces. An REW Help search for 'IR window' will locate general info.

If you have more detailed questions regarding how or why these settings were chosen, I will answer them. I am not sure which parts you understand and which parts aren't clear. I didn't save the modified mdat file but can also recreate it and post it for you if that is helpful.
 

·
Registered
Joined
·
34 Posts
Discussion Starter #13
1 - 13 of 13 Posts
Top