Since unlike quartz, ceramic arctube, isn't impermeable to neon, I wonder why lamp manufacturers don't uses neon-argon starting gas in ceramic MH lamps, to reduces their starting voltage (This can also be a good replacement to the Kr85 radioactive gas).

When it is permeable, the Neon will seep into the outer, "contaminate" the intended hard vacuum there and so cause the risk of discharge there. The only way to really keep the vacuum is to not use the Neon at all. Then you need some other method to assist the ignition...

So how MV retrofit HPS lamps, have a neon-argon filling, if they have a hard vaccum in the outer?

Do they really have hard vacuum there?
The retrofit's are of two kinds: One uses Penning mixture you have mentioned, those ignite directly on the mains voltage, so do not need anything extra. So the arrangement in the outer could be made so, the pure Neon there does not break down.

The other type uses some sort of internal starter, generating high voltage pulse (thermal contact with heater, glowbottle,...), same as the "selfstarting" HPS do (from these "selfstarting" the retrofits differ just by the arc voltage/current rating, to match the MV ballast characteristic). And these do not need the Penning mixture, so no Neon problems. But even with the Neon, the peak voltage is below 1kV, so still not attainable even with the Neon in the outer (just needs larger gaps and rounder edges)


But the CMH all need high voltage pulse for ignition (and even about twice as high compare to the HPS), so it is way more sensitive to anything in the outer, what may cause a discharge there, so influence the breakdown voltage.