It all depends on how it will be used, something only you know. From what you've been telling us so far, I'd go with the TFX anyway, I don't see much options.
What kind of PSU will it replace?
Application is 24/7, 200w (even lower) constant load, and is a new generation system and an evolution of the old one. Going from 4U / 19inch chassis to 4U / half of 19inch chassis, it's called a shoe-box.
Environment is climatized @ 25°C, but qualification has to pass 45°C ambient temperature in case of climatization failure or absence.
As it is new development in a compact chassis, i am struggling to take in account the hot internal temperature of the chassis. Interest in TFX is that it could also extract heat as it was the base of the FTX standard.
I would simply repeat what I said earlier, in a different way: I bet there's no PSU maker who will give a 10 year warranty for an industrial/commercial application like yours.
In fact, i indeed know this point. We have 1 or 2 year warranty on this type of PSU, and on MTBF, the rule or consensus is the 100kh lifetime (without fan, @25°C, etc).
I wanted an opinion on TFX as my system integrator
is against TFX spec as he had bad experiences in the past with such PSU (talking about old generations i guess, when efficiency/quality was not so good) and as he judges that the TFX spec is for wide public product and not aimed for industrial purposes
. So i'm trying to make my own opinion as i start on this domain (PSU qualification and TFX vs ATX).
the 1U PSUs probably all run their fans at higher speed all the time, because there's so much more internal impedance to airflow than the TFX12V form factor, and servers almost always run at higher loads. iirc, most server techs look for 60~70% PSU load.
Personnally, i tested the ZIPPY P1H-5400V
(1U, 400W, ATX, server model) that my integrator advised me at first, and with only 1 CPU board of 80W load, the fan rises slowly to max speed @ 52dBA noise... in same test configuration, SS-350TGM fan does not activate at all, as temperature is normal and load is under 50%. So i agree, but not on the load.