Adjust SAM Level?

[Confused]

@daniel.avasilichioaei Regarding the clicks and distortions that you mention, have you noticed the "clicks" during normal music playback, or is this something you have observed when using pure test tones or similar? The reason I ask is that in the past I have noticed some strange behaviour and clicks with SAM when testing with very low frequency pure sine wave tones, but I have never actually noticed this with music.

This also begs the question as to why a pure test tone might upset SAM. Does anyone have any ideas or thoughts regarding this?

[daniel.avasilichioaei]

@daniel.avasilichioaei Regarding the clicks and distortions that you mention, have you noticed the "clicks" during normal music playback, or is this something you have observed when using pure test tones or similar?  The reason I ask is that in the past I have noticed some strange behaviour and clicks with SAM when testing with very low frequency pure sine wave tones, but I have never actually noticed this with music.

This also begs the question as to why a pure test tone might upset SAM.  Does anyone have any ideas or thoughts regarding this?

Bug 1 is harder to spot with music, as it manifests in the bass and sub-bass area. However, it is very visible on any track that has these frequencies: bass and sub-bass "explode" from time to time. It is not about room correction: this is seen even in the movement of the speaker membrane. Clicks I don't remember hearing, but I'm not absolutely sure. And I quit testing...

Bug 2 is very easy to spot with music. It's about very heavy distortions. Other users in the forum have also reported it.

Both bugs reproduce even at small volumes (-35 ÷ 25 dB), so it should not be about SAM or speaker protection.

P.S .: If SAM does not treat a simple wave (pure test tones) correctly, what can we expect in case of a very complex wave (such as music)?...

[Confused]

P.S .: If SAM does not treat a simple wave (pure test tones) correctly, what can we expect in case of a very complex wave (such as music)?...

Perhaps counter intuitive, but I can think of reasons why a simple wave / pure tone could cause SAM problems when complex music does not.  For example, the algorithms and mathematics used in SAM might have issues with division by zero errors (or similar) with a simple wave that could not occur with a more complex signal.  This is pure speculation on my part, but it is perhaps possible.

[daniel.avasilichioaei]

P.S .: If SAM does not treat a simple wave (pure test tones) correctly, what can we expect in case of a very complex wave (such as music)?...

Perhaps counter intuitive, but I can think of reasons why a simple wave / pure tone could cause SAM problems when complex music does not.  For example, the algorithms and mathematics used in SAM might have issues with division by zero errors (or similar) with a simple wave that could not occur with a more complex signal.  This is pure speculation on my part, but it is perhaps possible.

I have very big doubts about this possibility when we talk about software. The division with 0 appears both at the simple and complex waves. If SAM had problems with the division with 0, then... it would be sad.
I do not mention the fact that one of the bugs is caused by a particular firmware, and in others it does not manifest, using the same waves.
In contrast, the complexity of the wave makes the defect more difficult to detect, which I have seen in practice.

In all the SAM profiles I saw on the Devialet site, the correction part of the SAM should change the amplitude of the wave to a reasonable extent (about 0 ÷ 30%). The problem is that in reality at some frequencies the change is much larger. Judging by the movement of the speaker membrane, it is consistently exceeded 100% in some cases.

There would be an element with possible impact: I saw that the SAM profile in the configuration file is the same, regardless of the Expert model, and regardless of the stereo or dual-mono configuration mode. However, the amplifier models, as well as the dual-mono configuration, have different performances. For example, I do not know if a SAM profile measured with an Expert Pro 250 stereo mode works correctly on an Expert Pro 140 dual-mono...

   

[Confused]

@daniel.avasilichioaei - Your last post has actually given me a couple of ideas for things I can use to test what SAM is doing, although that will need to wait until another time, I have a busy weekend ahead, followed by a busy time at work, so playing with SAM gets added to my "to do" list for now.  What I can say now is that I agree with you, I do not think there are actual division by zero errors occurring, I was just using this as an example of where simple things can upset a system, whereas more complex things may not.  As to the actual mathematics behind SAM, I do not believe anyone outside of Devialet has any true knowledge, I certainly do not.

Anyway, looking at your post yesterday I spotted something regarding the SAM "before and after" plots.  Below I have posted the Blade and Blade 2 SAM plots as an example of what I noticed.  Now like most things with Devialet, there is no proper explanation of what these graphs show, indeed, the plots have no scale, units or anything else.  So we are left to ourselves to work out what the mean.  The Blade SAM profile was a relatively early one to be done, before the days of "SAM 2".  Some will remember that after firmware 8.0 was released, any new SAM profiles were designated "V8", although this seems to have stopped now.  The Blade 2 profile was later, and originally designated V8.

With the Blade profile, it looks like the "before" graph shows the light coloured line as the signal, and the darker orange line as the actual, which is a little short of amplitude, and a little behind time.  In the "after" plot, SAM miraculously moves the orange actual plot in line with the signal white plot.  

   

With the Blade 2 plot, the convention appears to have changed.  In the before plot, the orange line has greater amplitude, this this looks to me as representing the additional signal SAM needs to produce to get the speaker to align with the signal as shown in the "after" plot.  However anyone interprets this, there is no doubt that the Blade and Blade 2 plots are showing different things.  I also note that the scale of the amplitude is different between the Blade and Blade 2 plots, and is also a little different between other SAM plots.  All a little inconsistent.  Another point to note is that the white "signal" line is slightly different in profile between the Blade and Blade 2 plots.  Why should this be? The cynical side of my mind starts to wonder if these plots are knocked up by someone in the promotional department, but I do hope that is not true. 

   

I am not trying to make a big point here, consider this simply as a piece of Devialet trivia that just happened to get my interest last night.  But maybe there is something behind all this?  It looks to me that the convention of the plots has changed, maybe at the time of SAM2 and V8 profiles.  I have not checked al the profiles for this, and neither could I, I do not know the dates at which any given profile was produced, I just happen to know the history of the Blade and Blade 2 profiles.  Just another miss-fitting piece in the Devialet mystery jigsaw puzzle.

Or maybe there is something I am missing here?  Does anyone else have any ideas about this?

[Jean-Marie]

@daniel.avasilichioaei - Your last post has actually given me a couple of ideas for things I can use to test what SAM is doing, although that will need to wait until another time, I have a busy weekend ahead, followed by a busy time at work, so playing with SAM gets added to my "to do" list for now.  What I can say now is that I agree with you, I do not think there are actual division by zero errors occurring, I was just using this as an example of where simple things can upset a system, whereas more complex things may not.  As to the actual mathematics behind SAM, I do not believe anyone outside of Devialet has any true knowledge, I certainly do not.

Anyway, looking at your post yesterday I spotted something regarding the SAM "before and after" plots.  Below I have posted the Blade and Blade 2 SAM plots as an example of what I noticed.  Now like most things with Devialet, there is no proper explanation of what these graphs show, indeed, the plots have no scale, units or anything else.  So we are left to ourselves to work out what the mean.  The Blade SAM profile was a relatively early one to be done, before the days of "SAM 2".  Some will remember that after firmware 8.0 was released, any new SAM profiles were designated "V8", although this seems to have stopped now.  The Blade 2 profile was later, and originally designated V8.

With the Blade profile, it looks like the "before" graph shows the light coloured line as the signal, and the darker orange line as the actual, which is a little short of amplitude, and a little behind time.  In the "after" plot, SAM miraculously moves the orange actual plot in line with the signal white plot.  



With the Blade 2 plot, the convention appears to have changed.  In the before plot, the orange line has greater amplitude, this this looks to me as representing the additional signal SAM needs to produce to get the speaker to align with the signal as shown in the "after" plot.  However anyone interprets this, there is no doubt that the Blade and Blade 2 plots are showing different things.  I also note that the scale of the amplitude is different between the Blade and Blade 2 plots, and is also a little different between other SAM plots.  All a little inconsistent.  Another point to note is that the white "signal" line is slightly different in profile between the Blade and Blade 2 plots.  Why should this be?  The cynical side of my mind starts to wonder if these plots are knocked up by someone in the promotional department, but I do hope that is not true. 



I am not trying to make a big point here, consider this simply as a piece of Devialet trivia that just happened to get my interest last night.  But maybe there is something behind all this?  It looks to me that the convention of the plots has changed, maybe at the time of SAM2 and V8 profiles.  I have not checked al the profiles for this, and neither could I, I do not know the dates at which any given profile was produced, I just happen to know the history of the Blade and Blade 2 profiles.  Just another miss-fitting piece in the Devialet mystery jigsaw puzzle.

Or maybe there is something I am missing here?  Does anyone else have any ideas about this?

To my knowledge, the convention has not changed between the two. 
My reading of the Blade2 graph vs the Blade graph is that the blade2 is overshooting for this particular piece of music, probably as an effect of resonance. 
SAM is correcting this and not only corrects the timing but also in that situation adapts the electric signal to counter that resonance effect. 

I have observed it myself on a couple of music pieces with my own system where SAM was correcting a resonance and therefore the level of bass was lower with SAM than without. 

Jean-Marie

[David A]

@Jean-Marie ,

Is there any reason you chose to configure a high pass filter in the speaker settings rather than simply activating the subsonic filter?

[daniel.avasilichioaei]

@Confused


I enjoy your decision to take tests for SAM.   I think some elements in the history of my SAM case might help you in your tests.




It all started when I noticed the temporary "explosions" of bass and sub-bass during the auditions. Initially I blamed the absence of room management / correction. However, when I analyzed the movement of the speaker membrane, I saw clearly that the anomaly is coming from here.




At this point I suspected that there is a problem with the SAM profile for my speakers.


First of all, I wanted to understand what he is doing and how SAM works, and what is the process of generating the SAM profile.


This movie helped me a lot: https://www.youtube.com/watch?v=AcLYb4sBhwQ. I strongly recommend you to watch it.




I realized that there are many parameters and factors that can generate an inappropriate profile. Here are some of them:

• Use of a new (unused) speaker, which has a different bass response compared to a speaker that has been running and has been used for a while.
  
• The parameters of the speaker that have been introduced in the software can be incorrect. For example, for my speakers:
  

• Number of bass speakers (see pictures below): In the picture on the Devialet site are presented other speakers, which have one bass speaker, but mine have two. It is possible that in the parameters of the software it was introduced that the speaker has one bass speaker, instead of two.
  
• Bass reflex: my speakers have a "hidden" bass reflex: it is not a visible hole, but it is somewhere under the speaker. It is possible that in the parameters of the software it was introduced that the speaker does not have a bass reflex, instead of introducing that it has.
  

All of the above (plus others) can lead to a SAM profile that boosts the bass too much, at least for some frequencies.


I requested from Devialet the parameters used for the SAM profile from my speakers, but they did not provide this info.




----------------------------------------------------------------------------------------------------------------------------------------------------------------




Finally, if it helps you, here is the test scenario I used.


Use it very carefully, it may damage your speakers!




Configuration and setup:

• FIRMWARE 13.1.3 DOS 2.2.4, and also FIRMWARE 13.0.0 DOS 2.2.2
  
• Configurator: SAM = on
  
• TONE \ BALANCE = 0%
  
• TONE \ BASS = 0.0 dB
  
• TONE \ TREBLE = 0.0 dB
  
• TONE \ SAM = 100%
  

Test scenario:

• Set volume to any value starting from -35 dB up to -30 dB
  
• Use "Online Tone Generator" https://www.szynalski.com/tone-generator/.
  
• Play sound starting from 1 Hz up to 100 Hz, using 1 Hz step.
  
• For each frequency played, use the remote and switch SAM on and off.
  
• Watch the bass speaker movement, and also listen to the sound.
  

Result:


Between ~10 Hz and ~30 Hz, SAM ON is performing a very strong bass sound amplification. This is MUTCH more than a simple sound adjustment to match speakers (or SAM purpose: "SAM lets your system achieve perfect temporal alignment between the recorded signal and the acoustic pressure generated by your loudspeakers").


Inside the 12 ÷ 20 Hz range, things are getting really out of control: Expert PRO is generating clicks and distortions on the average speaker (maybe treble speaker too). The source sound is not containing such frequencies, but Expert PRO is "generating" such sounds, and they are pretty loud.




ELEC FS 409 on Devialet site (3 speakers)


   




ELEC FS 409 in reality (4 speakers)

   

[Jean-Marie]

@Jean-Marie ,

Is there any reason you chose to configure a high pass filter in the speaker settings rather than simply activating the subsonic filter?

David,

The main reason is that subsonic is fixed 20Hz first order while I was able to choose 16Hz second order with the high pass. 

Full disclosure, I had been using the subsonic solution for over a year before that, but I eventually found the high pass solution to be more bulletproof (applies to every input instead of needing to be activated for  each individually), and the rolling off starts lower. 

So both work and the different is really marginal in my case, but a A/B testing on some pipe organ could be notified, hence my final choice. 

Jean-Marie

[Vivialet]

I have these clicks as well on some music if I set SAM to 100% (e.g. Malia & Boris Blank / Turners Ship).
The reason for these clicks is the speaker protection from SAM. It avoids excursions above a given limit which is causing these "clicks". Dialing SAM back to <70% is fixing this for me.
So it's not a bug but a feature.

Regarding high pass: I tried this as well, especially to integrate a subwoofer, but the high pass is already reducing the level above the defined value. e.g. if you set it to 16Hz even at 30 or 40 Hz the bass volume is reduced which sucks out some energy. I have given up on using the high pass.