I understand your frustration, your jumping into an odd OT area right off the bat. Many peoples OT issues don’t affect others because of different workflows.
The pickups are always time stretched and have the interesting ability to change the sequencer bpm. Although your finding them unreliable there’s most likely a logical behavior that’s happening according to your settings and how you operate them.
Qrec determines when the pickup starts recording but it is qpl that quantizes the end of recording->playback. With it on the pickups try to sync with each other and the sequencer in ways I haven’t fully explored. I would make sure the len parameter is set to off for all PU’s.
There’s still hope to figure it out, I’d help but been busy lately and will be a minute till I can get back to the lab. If the PU’s don’t get you there flex machines are an option to look into as they don’t affect the sequencer bpm…
Honestly I suggest giving it a break for a moment without giving up, and have some fun working with other features of the OT mixing some of the loops you recorded and slicing/warping/and adding effects using scenes and plocks and stuff, just to hear some OT goodness to put a smile back on your face, and then get back to it.
I had Rc 50, Rc 300, Rc 505.
Really like looper functionalities of the Rc 505.
Slaved, it’s perfectly in sync with OT.
I sold it because it’s too big and needs a midi foot controller for “full” control.
I can do with already set length tracks in OT, but one day I’ll surely by an other looper.
Rc 202 can be interesting, still waiting THE looper, slaved to Ot.
To do what you are trying to do, you have to be super accurate with timing. If you are a millisecond off then of course it isn’t going to be accurate. Thats not the OT at fault, its human error.
Multiply = [Track + Midi]
so there isn’t away to quantize the loop? thats a fairly basic function of most loopers. Bummerrrrr 
Octatrack is more a super recorder you can use as a “basic” looper.
You can make crazy things with Recorders and Flex, with realtime sample processing.
QRec in Recording Setup 2 is quantizing rec start.
Multiply works well, but can we divide ? 
yes its the QPlay setting as explained
Yes, you can by changing the record length once a PM loop is going.
I agree with @lescour1 in accounting for human error. Especially since you are wanting to vary the time signature. Also you may be fighting a workflow issue. As in, it’s also the OT. I’d suggest thinking in different musical terms to accomplish the same/similar results. Only suggestions, but maybe other recording methods with the OT, midi syncing devices, being okay with having a tempo of 122.3, etc.
Yep, the way the OP wants his pickup setup basically takes all OT assistance away and is reliant on human timing. Sorry to say, but the reason the BPM is deviating from 120 is because your timing is off. Even by a millisecond this can make all the difference. Its not a limitation of OT. Having said that, what you want to achieve is damn near impossible even with the best human timing. Either start by looping and sync the sequencer to the loop, or start with sequencer and quantize your loop playback and record to the sequencer. Those are your options if you want perfect timing.
The OP is used to a “regular” RC looper, like I was.
I was totally upset with PU at first and made all my first recordings with Flex playback.
PU are very confusing, especially for overdub, you have to press 2 different buttons to start/stop overdub. If you press them twice, you can replacd recording or stop playing ! 
After yesterday other tests, I don’t think I would use PU in live conditions.
They are better than Flex for overdub, if you want to layer sounds, better to handle long loops recording. Less clicks.
@crumblecakes, for the best of 2 worlds and more freedom, I would use both OT and RC. If you want to control with your foot, RC 300 is better than RC 50, much more logical, no track select.
RC 505 is more advanced, especially with the V2 update, I’d recommend that one or the RC 202. You can control them with 2 footswitches, or a midi foot controller.
I tried again, it’s weird indeed and not well implemented. Not a human timing problem.
I used a Master Pickup P1 at 120, synced to it.
I set a Pickup P2 with QREC and QPLAY to 16 and record 3x16=48, it is considered as 4 bars instead of 3.
If synced to P1 >Timestretch.
If synced to tempo, tempo change.
Can anybody confirm that stupid behaviour ?
Any workaround ?
It can be resolved by changing the bar number in Attributes but it’s boring, and impossible for a decent live performance.
As super @Open_Mike mentioned and from what I just tried, it seems to be doable to record/play 5/4 with FLEX.
You can overdub with Flex, with only one track. 
The problem can be the begining of the loop, and the fact that you may have to set scale per track.
I’m guessing the root of some of this is a presumption imposed on a loop length that it will be treated as 16 (or multiple/factor) unless predefining the length
i.e. there are certain assumptions laid down when you adopt open ended settings like Rlen Max
There’s a reasonable case as per the OP’s observations to hope that if you set a loop start/end accurately to a clock (even the current internal one) that you may be able to get a loop of 12/16 or 20/16 on the fly
Let’s find a simple example
Track 5 is a 8 beat Static 120bpm loop , mute T5 make T6 a neighbour and hi-pass it so you hear just enough to hear the 1 and count the beats
PUM on track 1 and 2
Example 1 …
T1 : capture 4 beat loop
settings RLen Max
Punch in on 1 and punch out on the 5
If you do it accurately enough you’ll keep very close the 120bpm and you’ll capture a 16len loop
useful enough
Example 2 …
T1 : capture 3 beat loop
settings RLen Max
Punch in on 1 and punch out on the 4
If you do it accurately enough you’ll keep the feel of the beat, but in 3/4 but it’ll be low 80ish bpm and still a loop of length 16 (i.e. not 120bpm and len12)
not what we might have liked
You can dial in a fixed RLEN if you know what you want ahead of time, even turning LEN on main page to off will allow setting fixed RLENs of interesting ratios, say 12/16 and 20/16, so you can keep it exactly at 120bpm …if you set T1 track a fixed length, lock the OT to it (Track + Tempo) and then set T2 to Max to allow on the fly lengths it seems to be okay if you record a 16, BUT, if you try to record another 12, it will seemingly capture 12 but accelerate the play through, by apparently fitting the same 12 count to another shorter length … arghh
So a specific 12 will be captured fine, but an ad hoc 12 can’t in this case, even though it says 12 … the slave with LEN set off is still being shackled to some prior choice in the code, is it a bug, why is it passable on 16 captured on the fly
It’s at these moments you start to wonder what is happening under the cover, what are the rules that are used from time to time, that unfortunately is also the moment at which musical inclination disappears into the cloud … the frustration with an opaque machine
It’s been talked about a few times, but we really need a PUM wiki, where we scientifically go through everything to document the possibilities and difficulties/workarounds etc … it’s frustrating because the flexibility can lead to undesirable outcomes, but there’s no doubt plenty of ways to harness them productively … just not as flexibly on most loopers (though some do impose restrictions on additional loop lengths) … and let’s not forget all the nuances of the OT architecture, there are so many aspects which can have a bearing, so it’s not always obvious that what you you think you see may just be intended behaviour but misunderstood
Perhaps if we were sure about what was not working well we could get some improvements essentially viewing them as bug fixes, though it may involve enhancing current usability when we are now fairly sure that development on the OT is finished … gotta be worth a shot if it makes the current implementation a little more robust/useful … I’ll set up a thread at some point to collate the community knowledge
Ain’t no doubt that the pickup machines are not as immediate as dedicated loopers and that explains why folk try and seemingly give up on them, but let’s try to understand what’s possible and what may or may not be broken
On a different note, how do you overdub with flex?
wow - some really very useful input here. thanks guys and gals.
@avantronica - In the initial example you gave, for the track 5 with a highpass, would you be able to lock the PU master to track 5? You said you could come “very close” to the BPM, but I’m looking to stay locked in on the BPM. I suppose this may not be possible with the PUM.
I feel like QREC and QPLAY could come in use here but my understanding of them is limited. I got some interesting results trying different combinations, but nothing consistent. I mentioned previously that I did get a few moments where I was able to loop odd beat lengths such as 20, by setting QREC to 1, but overdubbing gave me crazy errors and sometimes my tempo would shift.
@sezare56 - looks like you know what I’m going through. The Roland RCs basically bypassed human error. I know with RC loopers, you usually need to have your guide turned on (aka the click) for the looper to sync tempo, could this be the case with the OT? Can you set the OT PUM to follow CLICK? The video below seems to claim to be able to do so. I’m amazed that the OT doesn’t have a basic tempo lock for the PU, given the high intelligence if the machine. Also, I’m gonna second @lescour1 - could you please enlighten us a bit on overdubbing with Flex? Thank you! And thank you for troubleshooting this!
In this video, they mention that the PUM is set to record and loop in sync with the CLICK - does this mean that the PUM is already set to a predetermined length? Or is there a way to simply lock the PUM to time with the CLICK?
I go on to say how to stay locked to the current tempo - if you record with a RLEN set flexibly to Max then the PUM will seek to create a BPM once you’ve captured and set its length manually … so the skill there in setting start/end gets you close to the BPM - if you want it fixed you’d need to predefine a loop length and then lock to it … some scenarios can be achieved, but not necessarily without pre-planning, sadly
why not explain in the least possible steps exactly what it is that illustrates the core of your issue ?
In your case loading an existing sample (even empty) into a PUM, trigger it manually (Trk+Play) and locking seq to that track/tempo could be useful … then you only ever record into slaves
edit: cueing an otherwise muted click track as the master/locked ‘pickup machine’ might be another way to go about it
Here’s a stripped down, simplified example of what I am trying to do:
I am running an amorphous melodic pattern I created it on the fly, from an external synth running into AB. The pattern is semi-random, has no clear 1 nor a clear meter. However it is set at 120 BPM and thus has a clear pulse. I wish to loop this pattern in an improvisational fashion - going by feel, rather than a predetermined meter.
I also wish to then use the effects onboard the OT to manipulate this loop.
My (simplified) issue:
I cannot have my on-the-fly loop change BPM. I need my BPM to stay the same, regardless of my human-error timing, and regardless of my meter. The OT would then be slipped from the BPM of my other gear which is set to 120 BPM. I also play with others who are playing to a click or looping to 120 BPM. MIDI syncing could keep everything playing in-line, however it doesn’t address the root of my problem, which is sticking to a predetermined BPM.
What I am really trying to do is a bit more complex, but if I can get this function to work, I should be set.
I hope I explained this properly. I can go into further detail about my setup and my intentions with the OT but its a mouthful. Thank you for all your help 
I don’t understand why midi sync wouldn’t address the issue of sticking to a predetermined BPM . . .
You’ve used different words to essentially describe what you already conveyed … it’s not a step by step, simple/minimal repeatable example to illustrate the issue … you’d need to lay down constraints if any … i.e. if you don’t illustrate the ways the current options fail you … it’s hard to offer feedback or suggestions
I’d add this though, you’ll probably have drift between an external clock and a free running OT … so it may be best to clock external devices from the OT and work on any of the above suggestions to lock the master pum bpm and develop from that point
You can’t slave the OT to another device and get reliable use of the pickup machines due to the dreaded ‘dub aborted’ issue