It would be great to get any comments on this!
I have other boxes for "regular" production so I see this rack as more for experimentation. Sequences and external clock if used will come from Octatrack. So far I have the MN row, Benjolin and Plague Bearer. My choice of wanted modules is probably more idealogical than practical. I wonder if there are too many sound sources and not enough processing power. Still I can also send in other cv from the Octa via 1u MIDI. I'll look for the Pam's next, then probably the synchrodyne or cycle box with their expanders. It will be a while before I can finish the rack!

The problem I see isn't with a lack of processing. Rather, this build is really short on modulation sources. You only have the Maths, Contour, and Benjolin here, which makes up one complex LFO, one envelope source (really?), and the randomness that is a Benjolin. No VCAs here, either; I'm not counting the Optomix, as that's really a pair of lowpass gates, and neither a proper VCA nor mixer. Your observation about excess audio sources is spot-on, actually, but the excess is really cramping you here.

Things to lose immediately: the DPO and the very discontinued Cyclebox and its expander. You have a full WMD Synchrodyne setup here, which makes both of those rather superfluous. That's pretty much a modular in of itself...but it DOES need more modulation sources to really do what it's capable of. Even so, it's powerful enough that it would admirably make for this build's "core". Also, given that this module pair is 36 hp, you need as much free space as you can snag, since that's quite a chunk of real delete whatever is possible, and shrink whatever's left. The Contour, for ADSR in 8 hp, and you desperately need more EGs. While you might have that module on hand, it's not space-efficient here; you could drop in a Doepfer A-141-4 and have four ADSRs under common CVs in the same space, which strikes me as more effective for experimentation purposes. Again, scrunch EVERYTHING down as best you can. Even if you have modules on hand, they're not going to be useful if they take up huge amounts of space when you're in need of other functions that have to be addressed to optimize this.

Thanks Lugia for taking the time to comment! I don't have the synchro/cyclebox + expanders, or plog yet so there is space to spare and balance with some modulation sources. I thought Pam's was a good start for this but I will look into the EGs, you are right. There is a small 1u VCA in the top row but I had also thought of an Intellijel Linix, then that space could be used for another line out module giving 4 outs from the 7u case.
Do you have any thoughts on logic?

Pam's IS a good choice, don't get me wrong there...but my concern is the lack of both LFOs and EGs. Your best bet is something which can combine these functions; as of late, a lot of people have been gravitating to the Xaoc Zadar, but Intellijel's new Quadrax + Qx expander might be a better choice from the standpoint that there's no menu diving involved, plus there's a lot of interesting functions there. At the same time, though, the Zadar has crazy-long cycle times that run up into the tens of minutes, plus it also fits into 13 hp with its Nin expander, while the Intellijel requires 18 for the Quadrax + its expander. More or less a "what feels better" sort of issue. Do keep the Pam's, a clocking device and trigger sequencer, it's superb.

That stereo tend to see those for dynamic control on outputs, but in a rig like this, I'm wondering how useful it really is. There's other things that could be in there, especially if you swap that buffered mult (yep, there's enough CV destinations that you NEED it here) down to a 3U row and make it fit in 2 hp, which you already have open. With that gone, and after pulling the Stereo VCA and the 2 hp blank, you have 24 hp in the tile row to play with. If it were me, I'd have that row look like this:

Stereo Line In / uMIDI / Transient Modules u4R / Noise Tools / QuadrATT / Stereo Line Out

Now, what's that new u4R thing? Ahhh...another bit to go in between timing and logic. Have a look at it...makes lots of sense.

And speaking of logic, while the Plog is a really nicely-featured module, it's just the gates. You'll need a few more toys, such as comparators (which can generate gate info from CV/mod voltage curves), a diode OR for combining gates (see the Doepfer A-186-1 for an example), pulse counters which allow you to specify a trigger/gate to fire on a specific beat, maybe a derivator (something like Ladik's J-110, which fires pulses based on CV movement direction...crazy useful for tracking EGs and LFOs) or something probabilistic to screw with clocking, like Ladik's S-090. As regards the comparators, though...there's a specific type of these called a window comparator, and these allow you to output several different pulse signals as a voltage curve moves through a user-defined "window" of upper and lower thresholds. Dovemans' SHFT is a good example of these. Basically, the point of logic is to take the timing and sequential info moving around as pulsetrains, and screw with it all, while also generating more pulsetrains via comparators and the like, and then mashing all of these through a Boolean module like the Plog, et al to "smooth" the behavior between pulsetrains. Boolean logic takes a pair of incoming pulses, then subjects them to a logical test to arrive at a pulse output that happens on four specific pulse states: OR, AND, NAND, NOR. Or, more clearly...

OR: Pulses present at either input result in an output gate when both inputs do NOT concur.
AND: Pulses present at either input result in an output gate when both inputs DO concur.

...and the N states are these, but inverse. NOR (also known as XOR), for example, outputs a pulse when neither input has a high state. These tricks can be used to alter sequencer behavior, fire other events within a patch, screw with ongoing activities, etc etc etc. They open the lid on timing within a modular environment. Now, one other not-logic-but-really-it-is thing that fits here would be ratcheting, which is a triggered repetition of a pulse, usually via a sequencer. But with that pulse counter I mentioned previously, you can patch out a single clock step to fire the ratchet/clock multiplier (have a look at Doepfer's A-160-5 for reference) and get a rapid-fire pulse train with X number of pulses in subdivided time whenever that 7th beat gets fired. This is how Tangerine Dream gets those rapid repeated sequencer notes...albeit a bit simpler than you'd have to do it on Moog hardware. And it's also how you wring every bit of timing functionality out of your build!

Thanks for so much detail and suggestions! I thought and made some changes to the plan. Zadar looks amazing for ambient/drone/generative/noise which is the aim of this rack.

quadra is sufficient for regular envelopes when paired with zadar, rather than quadrax
contour is gone - can use ASR from quadra
maths is gone - big call, but quadratt does cv generation/attenuversion, pnw does lfos, plog covers the logic, quadra & zadar do envelopes, not too much missing?
added zeroscope to help me figure out what the hell is going on with my signals, and to tune oscillators (any midi sequences will be coming in from an octatrack). Once I’m more experienced this could be replaced.
buffered mult is now 3u, that’s a great call!

this leaves a nice 36hp at the bottom for synchrodyne+expand in the future - it would be a while before I can afford that anyway, and buying more modulation and EG first will get me more out of what I have already, just as you said.
Logic I will need to look into in much more detail including the u4r. For now I can get a random stepped voltage out of Pam’s and a bunch of triggers from plog.

This also leaves me with some kind of structure L-R of (midi/triggers/gates)-envelopes-oscillators-vcas/filters, which feels better thought out and tidier than before: