Isotek Sigmas EVO3 mains filtering conditions findings

I have been building audio-stuff for many years and my current system (Horns, SET, Tube DAC etc) is while I continue to tweak a lot, I hardly expect to make huge steps so I started to build several line filters on my workbench, limited bandwidth spectrum analyser and alpha labs EMI meter attached and having filter designs including common/differential mode sections.
Found the best combination, avoiding any unwanted effects of resonance/ringing of the filters themselves and I was technically quite pleased.
BUT, having of the shelf components and I was not looking for another project to cover a year, I treated myself to one of the references in the market, the ISOTEK EVO3 SIGMAS. Technically I was convinced they know what they are doing, the design was rather straight forward, the components...some very tailor made and specific and the build quality stellar. If I wouldn't hear an immediate difference it would fit my strategy to rule out every element in the audio chain that could cause a negative contribution.
To illustrate where I am coming from (EE background), if I don't hear a difference I will choose the cheap option in components. For choices like this I have a couple of relays that can be switched (blindly by a second person) from the listening position while playing a 20sec track on repeat of Diana Krall's Live in Paris singing "Case of you". In this way I was awed by the sound improvement of having good capacitors in my loudspeaker crossover and then moving to impregnated DTAC technical rational thinking needed a redesign here (micro vibrations in caps?)
Anyhow, back to he Isotek.
It is not as easy to switch back and forth while listening to music in this case, and I always doubt people when they did comparisons that took a couple of minutes switching or that were not done as a blind test.
The best I could do was done. The Isotek and a regular very cheap extension cord/socket/strip connected to my audio-group and I had to re-plug all mains plugs of 2 SET amps, DAC and CD player. Phase marked and plugged in correctly.

Before listening the mains noise on the wall socket was 300mV and on the output of the isotek 140mV (I was not impressed by that since my home-brew filters showed reduction factors of 6-10x).

Since currently I am finishing the RAAL lazy ribbon tweeter housings, the tweeters are not in the system and the loudspeaker will be limited to 8-10kHz (yes very silly, but then again nothing to loose when doing a test anyhow)
Listening to Krall, without EVO3, sounded a bit as if the tweeters were present (?????). With EVO3, balanced, truncated for the highest frequencies but very balanced, warm and detailed.
Without filter, certain moments in music are aggressive...Krall shouting at 2:30 in Case of You, but with filter this is smooth. This is the exact area in perception where I cramp listening to certain old Jazz with a trumpet a bit too hot in the recording.
This is not just a bump or dip in the frequency curve difference.

Played Boz Scaggs But Beautiful CD and listened to the amazing difference!

Now, the challenge to explain or discussion...without the regular explanation that audio is amplified by modulating mains supply voltage/current (we all now)....what is the mechanism here and above all, how would a couple of 100mV noise reduction have a very very significant big impact on the sound (and how would this be quantified???)

I am amazed, and delighted.

I sent some questions to ISOTEK regarding the relative small noise reduction measured, no answers yet. I am tempted to add some filtercomponents to the EVO3, but for now this is just a wild idea ;-)



Very happy with the Sigmas here.  Happy enough to splurge on the Synchro to go with it; it didn't take it to the next level, but it's good to know it's there.
Isotek does good stuff. I have their Aquarius power conditioner and very happy with it.

Maybe one day I will upgrade to Sigmas. They tend to get pretty expensive once you climb their “ladder”
Any of you compared the Equi=Core 1800 to the IsoTek?