API power supply - regulator latch-up

Just built a self contained PSU based on Baumann's design. All's well except for the +48V sometimes seems to "latch-up" at around 21-24V on power-up. The frustrating thing is that it does it about 50% of the time - other times are just fine - bang on 48V.
I've tried powering up with no load, 10mA load and up to 100mA load - and the same applies! - about 50% success at getting 48V - on one or two occasions it has hovered around 24V for approx 5 to 10 secs and then rampped up to 48V. When it does work - loaded 100mA and it has been rock solid - 2-3mV ripple!
I'm suspecting crappy brand of regulator - but it'll be a pain replacing it.
Just wondered if anybody has come across this before - particularly on this type of circuit - and if ther is a remedy!
Cheers as always

just lower C16 from 47uF to 4.7uF/63V (or leave it out). Maybe this was a typo in baumanns schemo.
http://www.thediypill.phx.com.br/forumfiles/psu312sch_v1.pdf

Spec-sheet of National's Lm-317 says, this cap is for improved ripple rejection and should not be larger than 10uF (preferebly solid tantalum).
http://cache.national.com/ds/LM/LM117.pdf

I had the same problem in a similar schemo. The supply-voltage dropped from 48V by several Volts when switching on phantom (messured supply- not mic-side) to 42V and 36V. When I kicked this cap off, Voltage was stable.

Thanks Harpo :grin: :guinness:
Will give it a go tonight