I'm a little confused by how you phrased the question, but can't you just pop all your edit tracks and only unmute one track's pops for each 'stem' groups (e.g. unmute pop on DX A, FX 1, MX 1, etc). That way you're not doing any weird sends to your submaster and stem prints because of how your audio is already bussed - wherever your edit tracks are going, the pops will follow. Then once the stems are printed with their set of pops from the edit tracks, I send those stem prints directly to a printmaster track and pull down the volume on 2 of the three stems at their head and tail pops (since I know they're time-aligned already) so that the final printmaster pops are -20.
Usually stages want all the edit tracks popped anyways (at least visually by having muted pops). Then when it's time to feed the tracks for printmastering they just unmute the pops which need to be unmuted so that each stem and subsequently the printmaster receives the pops - doing this also helps for troubleshooting to determine latency and other such issues if the pops are misaligned or have gain anomalies on the receiving end of the signal chain when they hit the print.
Trying to feed pops direct from some other isolated location (e.g. not from the edit tracks themselves), as it sounds like you may be trying to do, would defeat the purpose of the pops in my opinion as they are a kind of bible the whole way through the mix process to ensure everything is correctly dialed in.
I guess the way I'm reading about your method from the question sounds much more convoluted than it has to be unless I'm totally missing a key piece of information.
In a nutshell my workflow is:
- Edit Track (ex: FX 1) pop unmuted --> FX Submaster ---> FX Stem (where it gets printed)
- FX Stem (printed pop) ---> Printmaster (where it gets printed again)
The one caveat to the final step is that since DM&E all have a printed pop, when I feed those Stems to the printmaster as a seperate recording pass, I automate the volume down on the pops for the M&E stems so they don't sum into the Printmaster, but the pops still live the the printed Stem tracks.
Maybe you're trying to do your stems and Printmasters at once and that's what I'm missing? I usually do them as a separate passes both for QC and because, at least with ProTools, I recall there's some documented weirdness on signal degradation when recording multiple bussed tracks in a daisy-chain config at the same time. I don't know exactly how Nuendo works, but if you can print to a track via bussing, I'd recommend that over using an export feature. Printmastering is too critical to be left to an auto-pilot export function in my opinion.
If you're doing all stems and printmasters at the same time and you you could try to just flip phase on one of the stem pops - that way your summing to the printmaster would be -20+3-3= -20 instead of -20+3+3 = -15. But I believe the edit track method is more desirable method because of the troubleshooting abilities it allows for along the way.
Sorry for the fragmented response, I guess I'm just not following your core question clearly.