03:43 <_whitenotifier-4> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±2] https://git.io/JmRs2

03:43 <_whitenotifier-4> [scopehal] azonenberg 8f5a8e4 - FlowGraphNode: free dangling refs during shutdown in base class, not just in Filter class. Prevents triggers from holding onto refs to their inputs.

15:19 <azonenberg> Tested a few more AKL-PT2s. Yield now stands at 3 of 6 good or 50%

15:20 <azonenberg> The "bad" ones all failed due to -3 dB BW being too low,. There's also a rogue's gallery of small defects and inconsistencies i want to work on

15:25 <azonenberg> as well as A68-001 which failed due to the recycled SMA apparently being damaged

18:56 <azonenberg> Found time to do a few more probes while waiting for a meeting for work. Eight tested, five passed and three failed

18:57 <azonenberg> The passing probes are pretty consistent for the most part

18:57 <azonenberg> there's maybe a +/- 0.35 dB spread in the most variable parts (about 3.5 GHz)

18:58 <azonenberg> and quite a bit tighter for the rest of the range

18:58 <azonenberg> 62% yield still isn't *great* but i can tolerate it i think

18:58 <azonenberg> bigger question is what's causing the variation

19:49 <azonenberg> monochroma: https://www.antikernel.net/temp/pt2-s21-range2.png

19:49 <azonenberg> s21 flatness of the five satisfactory units i've tested so far

19:51 <azonenberg> A69-005 is quite different from the rest (note the lack of a 4 GHz dip)

19:51 <azonenberg> i want to know why

19:51 <azonenberg> also most of them have a dip around 3.3 GHz and that's bigger than i'd like too

19:59 <monochroma> azonenberg: huuuuh

20:26 <azonenberg> I think i may have an answer

20:28 <monochroma> demons? in my copper?

20:30 <azonenberg> It's more likely than you think

20:32 <azonenberg> https://www.antikernel.net/temp/a68-001.jpg

20:33 <azonenberg> so a68-001 is the recycled SMA, note the solder flowing into the center cavity and what appears to be chipping or melting of the ptfe dielectric

20:34 <azonenberg> more interesting is https://www.antikernel.net/temp/a69-005.jpg

20:44 <azonenberg> that little burr actually seems to have improved the response a bit

20:44 <azonenberg> which makes me wonder if maybe my launch is a touch too inductive and that extra C helped

20:46 <miek> i would set up one of the bad ones on a live vna sweep and just.. poke at it, see what happens :p

20:46 <monochroma> yeahhh

20:53 <azonenberg> First step is to finish binning them

20:53 <azonenberg> Then investigate the bad ones in more depth