That may have been my post from awhile back:
Dressing a metal orifice
In any case, a few comments are in order. While drift and creep as defined by posters above are accurate descriptions, manufacturers think of them both as creep, and only a few accept any creep at all. Those that do, generally expect lockup within 60 sec.
But the origin of the phenomenon is identical: air leakage past an o-ring pressing up against an imperfection in the sealing surface. Either the o-ring seals (drift), or it doesn't (creep).
"Drift" turns into lockup when the o-ring (in its static position) deforms enough to fill the scratch or imperfection. If it doesn't, the reg continues to "creep."
While the comment that "the IP resets itself with each breath" is accurate, the bigger problem is a reg with creep that sits pressurized for an hour's boat ride out to a dive site, and blows a high pressure seal just as you arrive, because the IP has gotten so high.
I'm surprised no one has referenced couv's checklist, so I'll do it here:
Regulator Inspection and Checklist (Rev-7)
So to answer the question of "which" IP you use when looking at the manufacturer's specification, it is the lockup value, whether that occurs immediately, in a few minutes, or 12 hours later. And yes, when I have a reg with "drift", I leave it on a tank overnight. If the lockup IP the next morning exceeds manufacturer spec, then a knife edge or piston shaft may need polishing. But if you can't get it within spec, you need to consider whether or not the reg should be dived.
Assuming you can get it to max IP within spec, if there's what's been called drift, then your working IP is the instantaneous IP that the second stage sees between breaths. Not such a big deal for a balanced second, but for a downstream valve you have a problem.
If you tune for an instant IP of 130, the second stage will probably freeflow when the reg is unattended and drifts to 145. If you tune for 145, then a downstream second may breathe much harder at 130.
This is the conundrum we all face when using old gear. Little of the drift described above is permissible with most manufacturers. Atomic allows drift over an unspecified time from 125 to the max IP of 145. Scubapro allows 8 psi over 45 sec per Rene Dupre, tho that's not in the manual. Etc, etc. But many DIY'ers dive gear that drifts or creeps and at some point there is a hazard. It's unlikely that a seal will blow from excessive creep during a dive, because each breath resets the process. But losing a seal when the reg is pressurized but not being used is a much more likely occurrence. Did you bring a spare reg set on the boat? And if you are diving a reg with drift, how did you tune your second stage?
For me, there's enough inexpensive gear out there on eBay that can be refurbished, that an unrepairable slow drift much more than 5 psi in my first stage is grounds to throw it away.
You can polish a knife edge or a piston shaft. But if the imperfection is on the inside of the reg body because some tech didn't use a brass bullet to reinsert the piston, or was clumsy with a steel pick, then there's not much you can do. If someone cheats with extra lube to fill the scratch and the reg appears to function just fine, beware using it a few hundred breaths later, when most of that lube has wiped away. If you're not checking with your IP gauge on a regular basis, whether you're a service technician or not, then you're playing with fire with old gear. EVERY first stage should lock up. If it doesn't, I wouldn't dive it. Don't accept creep. How much "drift" to accept is a personal decision. My tolerance is pretty low.
Use couv's checklist.
