Devialet Chat

For a long time, I have had the intention of making measurements that would show me more exactly what SAM does and what influence it has on the audio signal. I wanted to do this for these reasons:

• I really like the idea and concept behind SAM.
  
• In theory, SAM 100% should produce the perfect bass for the respective speaker. However, I heard SAM in several combinations of speakers and Expert PRO amplifiers and it seemed to me that SAM 100% boosts the bass much more than a simple necessary correction.
  
• I noticed that SAM behaves very chaotically and abnormally, at least for pure sine curves in the low bass area (lower than ~35Hz, probably the value depends on the speaker model).
  
• Although SAM is not supposed to impact high frequencies, I have seen user reports claiming the opposite. Some claim that they hear changes in the high frequencies. There were a few cases (very few) that claim that SAM burned their speaker tweeters (information from the speaker service, not just their personal opinion).
  
• I watched the video posted on the internet by Devialet, which shows the process used to generate the SAM profile. I detected several areas in the process that can generate unwanted results and that got me thinking.
  

Finally, today I managed to make measurements for the SAM analysis and I share them with you in this post.


Information about the measurements made and how I made them

• I measured and compared the following SAM settings: OFF, ON 0%, ON 50% and ON 100%.
  
• I used 2 Expert Pro 220 amplifiers and 2 Expert Pro 140 amplifiers, all in mono mode and connected in a Devialet chain.
  
• Each amplifier has one speaker connected, but there are 2 different models of speakers (2 pieces of each model). Obviously, for each speaker I used its SAM profile. Note: For one of the speakers I did not use the profile from the Devialet configurator: this profile is much too "aggressive", so Devialet provided me with a "softer" profile. This is an important note for some of the conclusions of the measurements.
  
• All 4 speakers are located frontally, one next to the other (two by two).
  
• I used REW software for measurements and generating test audio file (48kHz, 2M sweep, 0-24,000 Hz, -12 dBFS, 32 bit-float WAV). Note: I intentionally used a measurement band wider than 20-20,000 Hz. This is an important note for some of the conclusions of the measurements.
  
• I used Roon software for playback the test audio file (bit-perfect mode).
  
• I used the miniDSP UMIK-1 microphone for the measurements.
  
• All measurements were made in absolutely identical conditions, the only variable being the SAM setting.
  

General notes
These measurements reflect the SAM effect in my audio system and in my room. I think it is not correct for someone to consider that SAM effects are the same in other audio systems or in other rooms.
However, I believe that some of the SAM behavioral patterns resulting from these measurements are universally valid (but I'm not sure, obviously).
The general appearance of the curves is not relevant to the SAM effect: it is the result of my room's response. For the SAM effect, only the differences that appear between the curves when the SAM is activated are important.


The result of the measurements

Color legend

• SAM OFF - black color
  
• SAM ON 0% - green color
  
• SAM ON 50% - yellow color
  
• SAM ON 100% - red color
  
• Harman target curve - blue
  

[attachment=5516]


Bass (9-500 Hz)
[attachment=5517]

My observations:

1. SAM 0% behaves well and according to the information I received from Devialet: starting from 30 Hz down it starts to reduce the bass, and below 20 Hz it reduces it massively. This is good.
   
2. SAM 1-100% amplify the bass like this:
   

• 50-300 Hz: Small amplification. This is good.
  
• 20-50 Hz: Stronger amplification that increases as the frequency decreases. This is intentional and partly good, but I don't know if everything is good. The amplification in the 20-30 Hz range is very high (10-15 dB) and I think that this is not beneficial / natural for music.
  
• 10-20 Hz: Very strong amplification (stronger than at 20 Hz): 15-25 dB or even more. Although SAM 0% makes a massive reduction in this range (and this is very good), when SAM % increases above zero, the bass in this area is amplified back, at to much higher values than those at SAM 0% or SAM OFF. This is bad, very bad... I'm sure this is a bug and/or a design mistake. Such amplification generates great stress in the amplifier and in the speakers, and this for frequencies that humans cannot hear.
  

Treble (5,000-24,000 Hz)
[attachment=5518]

My observations:

1. Although the SAM should not affect the treble, it seems that it does.
   
2. Starting from ~7,000 Hz, trebles are slightly amplified. The amplification level increases with the frequency, reaching ~2 dB at the frequency of 20,000 Hz. This is not good. I guess it's a side effect or a bug.
   
3. Unexpectedly, the amplification level is maximum at SAM 0% and decreases as SAM % approaches 100%. But, it never reaches the same level as SAM 0% or SAM OFF.
   
4. At the same time, this amplification continues at frequencies higher than 20,000 Hz. Again, such amplification generates great stress in the amplifier and in the speakers, and this for frequencies that humans cannot hear.
   

I'm not an expert in interpreting measurements made with REW (in fact, I'm a beginner), so I'm curious what more advanced REW users can deduce from these graphs.

Moreover, I would be very interested if someone could help me with a more elevated interpretation. I am referring in particular to the fact that I do not know how to interpret the REW graphs for phase and impulse.
Also, I can provide the original files of the measurements.

There are two things that I've been wanting to check for a long time, but haven't been able to:

1. The phase correction that SAM 0% makes.
   
2. Synchronization between speaker output for mono configurations (dual-mono or Devialet chains with more than 2 amplifiers).
   

Any help in this regard is welcome.

I’m wondering how much of the response you’re seeing is because of the rooms acoustics and speakers behavior and not the behavior of SAM….

I have only used REW to test with my speakers but I noticed that my REW graphs were different when I added just 2 bass traps. The graphs were even more pronounced with the speaker that was closer to a corner.

Just one more thing to consider.

I also bought MiniDSP UMIK-1 and I want to measure the acoustics in my room, but I haven't learned how to use it yet

(21-Jan-2024, 20:50)daniel.avasilichioaei Wrote: [ -> ]For a long time, I have had the intention of making measurements that would show me more exactly what SAM does and what influence it has on the audio signal. I wanted to do this for these reasons:

• I really like the idea and concept behind SAM.
  
• In theory, SAM 100% should produce the perfect bass for the respective speaker. However, I heard SAM in several combinations of speakers and Expert PRO amplifiers and it seemed to me that SAM 100% boosts the bass much more than a simple necessary correction.
  
• I noticed that SAM behaves very chaotically and abnormally, at least for pure sine curves in the low bass area (lower than ~35Hz, probably the value depends on the speaker model).
  
• Although SAM is not supposed to impact high frequencies, I have seen user reports claiming the opposite. Some claim that they hear changes in the high frequencies. There were a few cases (very few) that claim that SAM burned their speaker tweeters (information from the speaker service, not just their personal opinion).
  
• I watched the video posted on the internet by Devialet, which shows the process used to generate the SAM profile. I detected several areas in the process that can generate unwanted results and that got me thinking.
  

Finally, today I managed to make measurements for the SAM analysis and I share them with you in this post.


Information about the measurements made and how I made them

• I measured and compared the following SAM settings: OFF, ON 0%, ON 50% and ON 100%.
  
• I used 2 Expert Pro 220 amplifiers and 2 Expert Pro 140 amplifiers, all in mono mode and connected in a Devialet chain.
  
• Each amplifier has one speaker connected, but there are 2 different models of speakers (2 pieces of each model). Obviously, for each speaker I used its SAM profile. Note: For one of the speakers I did not use the profile from the Devialet configurator: this profile is much too "aggressive", so Devialet provided me with a "softer" profile. This is an important note for some of the conclusions of the measurements.
  
• All 4 speakers are located frontally, one next to the other (two by two).
  
• I used REW software for measurements and generating test audio file (48kHz, 2M sweep, 0-24,000 Hz, -12 dBFS, 32 bit-float WAV). Note: I intentionally used a measurement band wider than 20-20,000 Hz. This is an important note for some of the conclusions of the measurements.
  
• I used Roon software for playback the test audio file (bit-perfect mode).
  
• I used the miniDSP UMIK-1 microphone for the measurements.
  
• All measurements were made in absolutely identical conditions, the only variable being the SAM setting.
  

General notes
These measurements reflect the SAM effect in my audio system and in my room. I think it is not correct for someone to consider that SAM effects are the same in other audio systems or in other rooms.
However, I believe that some of the SAM behavioral patterns resulting from these measurements are universally valid (but I'm not sure, obviously).
The general appearance of the curves is not relevant to the SAM effect: it is the result of my room's response. For the SAM effect, only the differences that appear between the curves when the SAM is activated are important.


The result of the measurements

Color legend

• SAM OFF - black color
  
• SAM ON 0% - green color
  
• SAM ON 50% - yellow color
  
• SAM ON 100% - red color
  
• Harman target curve - blue
  

Bass (9-500 Hz)


My observations:

1. SAM 0% behaves well and according to the information I received from Devialet: starting from 30 Hz down it starts to reduce the bass, and below 20 Hz it reduces it massively. This is good.
   
2. SAM 1-100% amplify the bass like this:
   

• 50-300 Hz: Small amplification. This is good.
  
• 20-50 Hz: Stronger amplification that increases as the frequency decreases. This is intentional and partly good, but I don't know if everything is good. The amplification in the 20-30 Hz range is very high (10-15 dB) and I think that this is not beneficial / natural for music.
  
• 10-20 Hz: Very strong amplification (stronger than at 20 Hz): 15-25 dB or even more. Although SAM 0% makes a massive reduction in this range (and this is very good), when SAM % increases above zero, the bass in this area is amplified back, at to much higher values than those at SAM 0% or SAM OFF. This is bad, very bad... I'm sure this is a bug and/or a design mistake. Such amplification generates great stress in the amplifier and in the speakers, and this for frequencies that humans cannot hear.
  

Treble (5,000-24,000 Hz)


My observations:

1. Although the SAM should not affect the treble, it seems that it does.
   
2. Starting from ~7,000 Hz, trebles are slightly amplified. The amplification level increases with the frequency, reaching ~2 dB at the frequency of 20,000 Hz. This is not good. I guess it's a side effect or a bug.
   
3. Unexpectedly, the amplification level is maximum at SAM 0% and decreases as SAM % approaches 100%. But, it never reaches the same level as SAM 0% or SAM OFF.
   
4. At the same time, this amplification continues at frequencies higher than 20,000 Hz. Again, such amplification generates great stress in the amplifier and in the speakers, and this for frequencies that humans cannot hear.
   

I'm not an expert in interpreting measurements made with REW (in fact, I'm a beginner), so I'm curious what more advanced REW users can deduce from these graphs.

Moreover, I would be very interested if someone could help me with a more elevated interpretation. I am referring in particular to the fact that I do not know how to interpret the REW graphs for phase and impulse.
Also, I can provide the original files of the measurements.

There are two things that I've been wanting to check for a long time, but haven't been able to:

1. The phase correction that SAM 0% makes.
   
2. Synchronization between speaker output for mono configurations (dual-mono or Devialet chains with more than 2 amplifiers).
   

Any help in this regard is welcome.

Hi,
no chance that SAM does not not modify the frequency response, being an equalization. 
Regarding your graphs, to make them more understandable, you should use a 1/12 smoothing (now is 1/48) and limitate the graph to the Schroeder frequency (around 250hz), and provide a full 20hz-20khz spl graph.
The frequency response, is the most important graph when you make a room acoustic analysis, but others are as important as the SPL, spectrogram and decay first, so you should show us both.
Could you please indicate model and brand of the speakers? It seems they go below 20/25 hz and I would like to check the data sheet to see if is the room that adds low frequency.
I tell you immediatly my thought: don’t like SAM, Devialet should have introduced something similar to ARC , Room Perfect or Dirac.

Daniel,

Your technical knowledge and measurement capability far exceeds mine, my smattering of technical knowledge is in a very different field, health and safety, but there is one thing I was taught which still sticks with me today. That is "measurements tell you a lot about what you measure but they tell you nothing about what you don't measure".

You said one of the 2 things you've wanted to check for a long time but haven't been able to is the phase correction that SAM 0% makes. I think phase correction is critical to understanding what SAM does, it's really the only thing that Devialet tells us about what SAM does. What I would like to know is what SAM is doing with phase at all points of the SAM scale from 0% to 100%. We know that SAM delivers maximum bass extension at 100% because Devialet say that BUT how is that bass extension achieved? Is all phase correction applied at the 0% level and what happens at higher levels just the addition of an increasing amount of level boost via an equalisation profile, is what happens above 0% an extension of the application of phase correction to lower frequencies, is what happens a combination of those 2 things, or is their some other "secret sauce" addition added to the mix? We really don't know.

What I do think is that in the absence of a fuller account of how SAM works from Devialet, I don't think we've got any hope of knowing just what is going on without having access to phase measurements at 0% and at least a sampling of higher SAM settings up to 100%. I think knowing what's going on with phase correction over the whole 0-100% setting range there's no hope of knowing what SAM is doing.

In relation to the measurements you made you said "These measurements reflect the SAM effect in my audio system and in my room. I think it is not correct for someone to consider that SAM effects are the same in other audio systems or in other rooms." I think the room is clearly a significant confounding factor in your results. Looking at the peaks and valleys in your 9-500 Hz chart, it's clear that what you're measuring is an in room speaker response rather than simply the effect of SAM. You're making your measurements in the same way you would measure a room for a "room correction" program and the problem with that is that if you move the microphone to a different location and repeat the measurements your results will be different. The same will happen if you leave the microphone at the same location and move the speakers. You aren't measuring what SAM is doing, you're measuring what the speaker/room combination is delivering at the microphone position.

You need to make your measurements at the amplifier's speaker terminals in order to get a result that just shows what SAM is doing rather than what SAM plus whatever the speaker, room, and microphone location are contributing to the results. What SAM does occurs before the speaker, room, and microphone location get involved. Change those things and your results will change but what SAM is doing to the signal going to the speaker does not change with changes to the room, speaker location or microphone location.

And, since Devialet say that SAM is applying phase correction, you must also measure phase response at different SAM levels as well because it's obviously a significant part of what SAM does. That means you can't understand SAM without knowing what it's doing to phase response and whether that changes at different setting levels.

Bear in mind also that as you increase the SAM setting your amplifier volume setting may become important because of the speaker protection aspect of SAM which starts to ease off whatever SAM is doing at higher volume levels in order to protect the speaker.

@gianventu: SAM is not a "room correction" system, it's a system intended to do 2 things, correct woofer phase response below 150 Hz and to extend the speaker's bass response to some degree without risking speaker damage. It's not intended or designed to modify or correct room caused frequency anomalies which is what systems like ARC, Room Perfect, and Dirac are intended to do. SAM can't do what they do and they can't do what SAM does. SAM and those systems are designed to address different problems and Sweet Room, not SAM, is the room correction system Devialet has implemented in order to do the same sort of thing as ARC, Perfect Room, and Dirac do.

(22-Jan-2024, 12:40)David A Wrote: [ -> ]@gianventu:  SAM is not a "room correction" system, it's a system intended to do 2 things, correct woofer phase response below 150 Hz and to extend the speaker's bass response to some degree without risking speaker damage. It's not intended or designed to modify or correct room caused frequency anomalies which is what systems like ARC,  Room Perfect, and Dirac are intended to do. SAM can't do what they do and they can't do what SAM does. SAM and those systems are designed to address different problems and Sweet Room, not SAM, is the room correction system Devialet has implemented in order to do the same sort of thing as ARC, Perfect Room, and Dirac do.

Yes, SAM is not a room correction as well as Sweet Room that is a nine bands parametric equalizer. 
DRC (I use Dirac in my system), is something completely different that also intervenes on the phasing and the gain of the speakers.
Looking at measurements posted by the opener, SAM modifies heavily the frequency response of the low end. I would be curious to take a look to the spectrogram with SAM 100% and without.

I think REW is difficult, I don't have any concept of it.

Interesting thread. This brings me to the question that when doing a measurement for room correction (I am using Focus Fidelity Designer) is it recommended to have SAM at 0, 50 or 100% ?

(23-Jan-2024, 22:27)markush Wrote: [ -> ]Interesting thread. This brings me to the question that when doing a measurement for room correction (I am using Focus Fidelity Designer) is it recommended to have SAM at 0, 50 or 100% ?

I think the room correction should be done with SAM 0% (if you trust the phase correction made by SAM) or with SAM OFF. Only after that I would try SAM >0% and possibly return to finishing the room correction.

Today I wanted to measure the effect of the "Subsonic filter", as well as that of several "High-pass" filters.
Subsonic filter seems to do its job as intended.
However, with the high-pass filter, I had a very unpleasant surprise: it seems not to work at all. I tried several amplifier configurations and several filter settings and none of them have any effect: no difference is heard, and the measurements come out identical. I even tried extreme settings (cut-off frequency = 5,000 Hz) and there is still no difference. It seems that the high-pass filter does not work at all, which surprises me: I know that many users have subwoofers and use the high-pass filter.
Strange situation...