Firmware 13.2.0 - DOS 2.2.6 for Expert Pro - Release note

[Vivialet]

Summary: My initial experimentation with Sweet Room filters from REW yield subtle changes for the worse, so point out if I'm doing something wrong.

I had previously taken measurements for Dirac and REW, so had those ready to go.

In REW I played around with rePhase filters, changing the Target Level in .1 dB increments and seeing how REW calculated the EQ filters. I chose what looked best (correcting the biggest flaws with the smallest gain) and wrote that into a Sweet Room file (using the Alternate as a control). I measured the difference in SPL when using each and then level-matched them with a 1 dB main gain so that the comparison was more valid.

I'm surprised by how subtle the difference is, but it is noticeable and not for the better. It almost seems like the processing is degrading the sound quality but the filters aren't doing anything. What I hear is flatter, less dynamic, and farther away. Voices are a notch back on the soundstage, there's a bit less bloom to crescendos, and instruments are a little less realistic. I keep staring at my Sweet Room file trying to find a syntax error (which would mean it isn't used), but the amps show it in use and there is that small difference in sound. I expected a -5.7 dB gain with 2.73 Q to be dramatic. My Dirac filters are much more noticeable. Were my expectations wrong or have I made a technical mistake somewhere?

I would remove filters 1 and 2, 2db is not worth correcting. For filters 3 and 4 I would only reduce by maybe 3 db.
Filters 6 and 7 look ok. Try to increase a little bit under 60 Hz.

Dirac is also increasing the gain which you are not doing at all. Maybe this is the reason that Dirac has a more dramatic effect.

[alandbush]

Summary: My initial experimentation with Sweet Room filters from REW yield subtle changes for the worse, so point out if I'm doing something wrong.

I had previously taken measurements for Dirac and REW, so had those ready to go.

In REW I played around with rePhase filters, changing the Target Level in .1 dB increments and seeing how REW calculated the EQ filters. I chose what looked best (correcting the biggest flaws with the smallest gain) and wrote that into a Sweet Room file (using the Alternate as a control). I measured the difference in SPL when using each and then level-matched them with a 1 dB main gain so that the comparison was more valid.

I'm surprised by how subtle the difference is, but it is noticeable and not for the better. It almost seems like the processing is degrading the sound quality but the filters aren't doing anything. What I hear is flatter, less dynamic, and farther away. Voices are a notch back on the soundstage, there's a bit less bloom to crescendos, and instruments are a little less realistic. I keep staring at my Sweet Room file trying to find a syntax error (which would mean it isn't used), but the amps show it in use and there is that small difference in sound. I expected a -5.7 dB gain with 2.73 Q to be dramatic. My Dirac filters are much more noticeable. Were my expectations wrong or have I made a technical mistake somewhere?

If I understand correctly, you have used previous measurements.  I presume nothing significant has changed subsequently with your environment.
I also note that it is a stereo measurement judging from the same filters applied to each speaker.
I've always assumed that Dirac applied different filters to each speaker.  Is this one of the reasons Dirac sounded better than the Sweet Room filters?

Personally, I have always measured the speakers individually and, I think, most importantly measured post filter application, although I realise this is not a simple task with the Devialet procedure.  REW provides an accurate prediction but real rooms do not always behave as predicted.  I am no longer surprised if the post measurement is subtly different than the prediction, especially if applying gain.

I strongly suggest re-measuring, as painful as that may be, to confirm that the predicted curve is what you are listening to.

Ps - I'm curious to know what happens if you do enter an invalid formatted EQ.txt file. A manually produced file is prone to such error and I would hope that there was some indication that the file was invalid. If not, then a post measurement would seem to be very necessary.

[deviousalet]

I will get around to measuring and creating filters for each speaker, but I wanted to play with Sweet Room and this was easier since I had a measurement of my system as it is.

Interesting that the prediction isn't accurate, which makes sense but does make the cumbersome Devialet system of SD-card updates really annoying. Our amps are Linux computers running web servers, and yet we must use sneaker net to update them. Sigh. I do love the sound, so will shuffle SD cards around to measure and iterate post-filter I guess.

According to Devialet, if there's an error in the eq.txt file it simply won't be loaded and the amp won't show EQ in the display.

[JohnnySix]

I will get around to measuring and creating filters for each speaker, but I wanted to play with Sweet Room and this was easier since I had a measurement of my system as it is.

Interesting that the prediction isn't accurate, which makes sense but does make the cumbersome Devialet system of SD-card updates really annoying. Our amps are Linux computers running web servers, and yet we must use sneaker net to update them. Sigh. I do love the sound, so will shuffle SD cards around to measure and iterate post-filter I guess.

According to Devialet, if there's an error in the eq.txt file it simply won't be loaded and the amp won't show EQ in the display.

I would recommend moving up the Target line until the maximum adjustment comes out around 4dB. I have never trusted the auto-calculated Target. I try for a target that cuts the worst peaks in half, and certainly not BENEATH some of the "valleys".

[JohnnySix]

Looking closer at your graph... I think you are extremely fortunate to have the majority of your variations within 5dB!
Mine were 30dB to start with, reduced to 15dB with GIK room treatments. Unfortunately my room is exactly twice as long as high.

[alandbush]

I will get around to measuring and creating filters for each speaker, but I wanted to play with Sweet Room and this was easier since I had a measurement of my system as it is.

Understood.

Interesting that the prediction isn't accurate...

Oh no, it is accurate but in a theoretical sense.  It is Real rooms that just don't always behave accordingly. 

According to Devialet, if there's an error in the eq.txt file it simply won't be loaded and the amp won't show EQ in the display.

Ah. Well, I suppose that's better than nothing.

Still, I'm sure one of our forumites could develop a simple gui that drastically reduced the chance of errors. 

[Vivialet]

finally I got an answer for my question regarding the main gain. It is like with any other EQ, if you add an EQ with a positive gain you have to reduce the main gain by the same value.
So if you have an EQ with gain +3 you have to set the main gain to -3 to avoid clipping.
They also confirmed that the EQ does not affect the pre-out.

[JohnnySix]

Ah, thanks for the pre-out confirmation. I was going to report that to them. There are other bugs with pre-out, such as applying MAX volume to them when streaming DLNA source! (Only in Volume 'passthrough' mode, I should add)

[Vivialet]

so far in passthrough mode the pre out was not active. I have not tested if that has been changed with the new firmware.

[Delija]

In my case - it's active on master, but not active on companion. I haven't tried new firmware.