Best TDD ratio?

I was wondering what the best TDD fixed framing ratio to use generally speaking.

I am coming from AirmaxAC gear where a 75/25 split generally performed worse than a 67/33 split due to some pretty big UL starvation issues that would negatively affect TCP ACK causing lower bandwidth with their TDD fixed-framing implementation. Rates would look good and individual user performance would be good, but under heavy load in the evening with most clients active it would actually push less download bandwidth at 75/25 than 67/33 even if the UL total bandwidth was quite low, due to ack framing starvation.

At lower clients counts I think 75/25 would work ok, but if I used TDD framing on Airmax AC I would always do 67/33. Honestly, I moved away from pretty much all TDD fixed-framing on the AirmaxAC line because of this and use their TDD based flexible framing instead. I think their other products like Airfiber and LTU worked better at 75/25 split.

So for ePMP, mainly thinking of the 4000 series, does Cambium have the same issues with UL starvation issues at 75/25 where I would be better with a split of 60/40, 65/35, or 70/30?

I’m using 70/30 on loaded APs and it seems to work well, 3000s though. Tried flexible and it fell apart.
Saying this the package we sell on these specific units only have 5 upload and up to 50 down.

we’ve used 75/25 for years without any issues

We typically use 75/25, and generally have found that best – but I’ll throw this idea into the mix for MU-MIMO APs…

MU-MIMO AP’s will trigger the need to use MU-MIMO only if it’s necessary. In other words, if the bandwidth needs can be satisfied handily in SU mode, then that’s what it’ll prefer. Only when SU mode is inadequate, will MU mode be used.

So (just throwing idea this into the mix for MU-MIMO APs) you can test on a 3000 AP and you’ll be able to see that in some cases, 75/25 won’t need to trigger MU-MIMO very often because the 75% duty cycle is able to keep up with the requirement in SU mode, and yet if you go to 70/30 or 65/35 (for example) then the 65% will be saturated earlier, and will trigger MU-MIMO earlier… in which case you may still have the same effective peak download throughput (due to MU-MIMO gains) and yet also have more upload (due to 35% upload duty cycle)

So - I just wanted to throw that into the options… if you’re deploying MU-MIMO APs then the ideal ratio may potentially benefit (or maybe not depending on your use case) from having a slightly lower DL/UL ratio in order to trigger MU mode more often, and have more overall effective DL/UL capacity.
#YMMV