Ok, re-reading through the entire thread again, I suspect there are two things going on here:
The System:
1. The OP is listening to a system simply with an odd configuration. From what's been posted, it isn't trunked (DSD+ shows as CB96), it isn't simulcast (there appear to be multiple frequencies in use but I've seen no mention of multiple transmissions on the same frequency) and the odd data bursts might be due to something not yet understood (eg AVL etc). I think DSD+ would be the best tool to further investigate. So I don't believe the oddities of this NX96 system are the root cause of why the 436 isn't functioning as expected.
NX96 on a 436:
2. Several people are posting issues with missed NX96 transmissions. I have access to multiple 436's, a TRX1 and several NX48/NX96/DMR handhelds, so I've now spent a few hours running a number of side-by-side tests. All tests are on simplex Conventional channels, results are consistent across VHF and UHF, and between 436's. The 436's are both early models, both have self-mod C1 caps fitted, neither have had clock issues [yet !]. I've found:
a. Across all tests, FM works better than NFM (most noticeable on NX96, some improvement on DMR, little difference between NFM and FM on NX48)
b. I too can confirm AUTO Digital Threshold mode decode of NX96 is inconsistent.I too get the 'missed calls' behavior (it's like the NX96 signal doesn't exist, or the 436 is asleep). Frustratingly I can't pin point a set of circumstances where I can consistently replicate the problem. I have even seen several tests where two identically configured 436's behave differently on the same signal - one decodes, the other is asleep.
c. For me, Digital Threshold Mode Manual 10 gives best results for NX96. Others might (and have) find a different threshold works best. It rarely ever misses a call but will display higher ERR rates vs an AUTO setting (assuming the AUTO setting detects/decodes the signal in the first place). NFM seems to react slower than FM in this setting.
d. An incorrect Manual setting will render the 436 completely deaf to NX96, NX48 or DMR (eg setting Manual 12 vs 10 for NX96 consistently makes it deaf). Same issue for NX48 if the Manual setting is too far from the correct threshold.
e. For me, Auto Digital Threshold mode (noted working at 10) gives best results for NX48 and DMR. In contrast to the previous Firmware release, I used to have to run Manual 5 to have any chance of some simplex DMR decoding - even then it was hit and miss. Now I completely rely on an Auto Threshold setting, and DMR simplex is vastly improved as a result. I can even run repeatable tests where Manually setting one 436 to 10 won't work while the other 436 sat next to it is happily the decoding the same signal on Auto (showing 10).
3. So in conclusion:
a. I do believe there is a digital threshold related issue with NX96 decoding. It has similar traits to the DMR simplex problems when the DMR upgrade was first released. But fixing it might impact performance on DMR/NX48 (and possibly other formats).
b. FM is better than NFM all round. In theory that shouldn't be the case, but that's what I've found.
c. For NX48 and DMR Simplex I would recommend Auto Digital Threshold.
d. For NX96 I would INITIALLY recommend Auto Digital Threshold, then use FUNC+Volume to note the best threshold value, and then add this Manually to the System.
e. For Searching for mixed modes I would recommend Auto Digital Threshold, but be aware you might miss some NX96.
HTH
Jim