1 no negative effect on the sound quality
2 Yes the level is lowered by 11dB, therefore yes, you should use TRS cable here to connect the Kemper, not XLR-cable.
3 the preamps are designed for Microphone-LEVEL signals and also for line level.

Make sure in Totalmix input channel Mic 1 the input gain is fully closed zero. Use the gain knob carefully to ramp-up your signal when really needed.

From page 38 user manual :

The two combo jacks also allow for the usage of mono and stereo TRS jacks. The TRS inputs
have a fixed level attenuation of 11 dB. Based on the adjustable amplification from +0 dB up to
+65 dB, the sensitivity is +21 dBu down to –44 dBu, referenced to full scale of the AD-converter.
Therefore the TRS inputs are true full level Line inputs, and the unit can also be used as Line
amplifier.

I've used Spdif for several years, there's 9 samples of latency between the DI (unprocessed) and the STACK (processed) side of the SPDIF signal. On top of that, the reamp loop adds around 500 samples (10ms) of latency, which bring various phase issues that have to be compensated.

In SPDIF mode, the DI output signal is too hot. In analog mode you can choose several DI output modes.

After tedious testings, for some reason, using analog output into the Fireface gives a slightly better sound, or should i say a more suitable and satisfying sound to my ears (Slightly more silky high end, slightly more tight and rounder low end). And of course, no reamp latency and no phase issue between the orignal stack processed signal and the reamped one.

Trust me, there's NO WAY i'll ever return to SPDIF using this device.

I ran the same test using a VIRUS TI 2 keyboard with native Asio USB audio VS TRS jack analog output to fireface input. The sound is silghtly different, and more satisfying to my ears.

I have no explanation, nor i am trying to launch any debate. I swear there's a difference, as tiny it is, it is still very satisfying to me.

I don't see the point in being condescending and rude.

On the kemper the SPDIF DI out or the analog DI output carry the unprocessed signal straight from the guitar.

When you connect the kemper using SPDIF the signal carries the D.I raw unprocessed signal on the right and the processed signal on the left. You then get your processed take and the unprocessed signal on separate tracks.

At 48khz/24bits there's a slight latency of 9 samples between the two signals, the processed signal has a slight latency due to the kemper internal processing time.

About reamping : There will always be a "roundtrip delay" for the way from DAW to Kemper and back. But most DAWs do have tools on board to compensate this as it is absolutely necessary to avoid phasing issues. At 48khz / 24 bits there is a latency roughly around 10ms or 500 samples.

In rock music layering 2 differents guitar sounds to get a more consistent rythm guitar tone is almost a common rule. It's commonly used for solo sounds too...

So when you reamp your orignal DI take, the result is not aligned / in phase with your first original processed take. Unless you compensate for the reamp latency by using the kemper as an outboard gear / external effect in your daw. Your daw will then send a sample to the " external effect " and measure the time it takes to come back thus detecting the latency and adding compensation accordingly.

A second way to do it : You could ignore the first original take and reamp the DI a second time with the same sound / patch used in the original and both reamps would then be aligned in this case. But there are lot's of different takes and sounds, this is a huge time loss.

A third way : You record the way you want, then using plugins like InPhase you can align every tracks and then avoid phase issues.
You can even align everything by hand using phase inversion on one of the track and moving the second until they almost cancel each other.

Playing guitar is all about finger control, you can't predict how a simulated amp/hp will react to the way you play unless you play and record the direct processed signal. This is why you record both the DI and the processed sound a the same time.

No offense taken
Thanks anyway for the valuable help.

I didn't want it to sound like a refusal.
In fact i used the Kemper with SPDIF for a lots of years.
Going away from SPDIF to traditional analog TRS / TS connexion between the kemper and the Fireface was also for convenience :
- It gives me more control over the " DI " signal. In analog mode you have 3 different DI settings unavailable to the SPDIF DI output mode which tends to be realy hot.
Because of that, back then while reamping i had to lower the recorded DI track volume, it was clipping the kemper input, i had to tweak the volume until getting the amount of gain and reach by ears the original sound of the kemper patches we made from scratches.
- We ran several test using different audio interfaces (RME, Antelope and UAD) and the same kemper patches, for some reasons the " analogue cables way " added a little something very satisfying to the sound while comparing to the SPDIF counterparts at the same volume.
Maybe preamps and converters on the above interfaces degrade the sound in some way which have some unexpected impact and/or add some unexpected color ? I have absolutely no idea.
We then did the same tests with a Virus TI 2 Keyboard and got the exact same feeling.

It's all subjective and it might be beyond logical, i realy don't want to sound stubborn or anything but we were all unanimous about the results. I have no explanation, i am sorry for that.

(If i find some time tomorrow or later, i'll upload some samples somewhere so you can listen to the difference.)