my first thought would be that the turbo just cant get enough air through to sustain 17psi as the revs get up and you use more air.
 
if the turbo CAN sustain it, my next thought would be that perhaps your intercooler core is too restrictive to flow the higher mass flow rate of air and the pressure drops.
 
if it's not that then possibly the exhaust manifold/adapter for the new turbo has something wrong, but another possibility..... but have you considered where you're reading boost from? has it been from the same point the whole time or have you changed from teeing into the wastegate line to direct on the manifold or vice versa?
 
has it EVER held 17psi? or is that just a target you think you should be getting?... it's not uncommon for turbos to drop away in the top end as either their compressor can't flow air or they start to choke up the exhaust  and come out of their efficiency bubble.

From experience, that sounds exactly like the type of spiking you get from MBCs.  Are you using a Tee-piece with a ball and spring arrangement for surge control?  Is the controller as close to the turbo as you can safely get it? (in accordance with its instructions).  MBCs are typically slow to respond to fast-spooling turbos, my Turbosmart MBC took around a second or so to regulate the boost on my old car, running at 1bar - It'd spike to 1.3bar then drop back.
 

adisco

just_ace

I wonder if you just getting a spike to 17psi, but stables at 12psi?

Or possibly ign break down in higher rpm?

Wouldnt think boost leak.

I may or may not be.
In my opinion its not because ive started with 10 psi and gradually upped my boost. Holds 17 psi from 2 800rpm to about 4500rpm and than the dropping starts...