azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | https://github.com/azonenberg/scopehal-apps, https://github.com/azonenberg/scopehal, https://github.com/azonenberg/scopehal-docs | Logs: https://freenode.irclog.whitequark.org/scopehal
<_whitenotifier-f> [scopehal-apps] kench opened pull request #238: Add Travis CI configuration - https://git.io/JTBz6
<azonenberg> https://twitter.com/azonenberg/status/1318347778892857344 initial test results from the AKL-PT2 prototype
<azonenberg> The enclosure is just a little too snug, i pinched the pcb a bit trying to get it in which might be the source of some of the imperfections (but probably not related to the huge hump)
<azonenberg> The resonance seems about the right length to be a reflection off the SMA launch caused by it not being super well matched
<azonenberg> Which isnt surprising, i did the stackup design based off a pretty rough estimate of the polyimide properties
<_whitenotifier-f> [scopehal-apps] kench synchronize pull request #238: Add Travis CI configuration - https://git.io/JTBz6
<_whitenotifier-f> [scopehal-apps] kench synchronize pull request #238: Add Travis CI configuration - https://git.io/JTBz6
<Degi> Huh, OpenEMS looks pretty neat and looks like it'd support 3D structures pretty neatly.
juli966 has quit [Quit: Nettalk6 - www.ntalk.de]
<miek> azonenberg: can you use time domain mode on your VNA? iirc it has it but was buggy/unstable?
Degi has quit [Ping timeout: 272 seconds]
Degi has joined #scopehal
alexhw has quit [Remote host closed the connection]
Pretzel4Ever has joined #scopehal
Pretzel4Life has quit [Ping timeout: 240 seconds]
electronic_eel has quit [Ping timeout: 265 seconds]
electronic_eel has joined #scopehal
<azonenberg> miek: so basically the issue is, it requires using harmonic rather than linear point spacing
<azonenberg> Which i don't like doing, because it's less useful for stuff like de-embedding
<azonenberg> and any time i change the number of points, start/stop of the sweep, etc i need to recalibrate (duh) but it also does some incredibly slow internal operation that takes about 30 minutes
<azonenberg> i have no clue what in the world is so wrong with the firmware/software to need that
<azonenberg> but the progress bar barely looks to move, the entire app locks up because they don't do it in a background thread
<azonenberg> so generally i try to set it up once, keep the software running for weeks at a time, and always have 10001 points linearly spaced from 300 kHz to 6 GHz
<azonenberg> and just never touch anything because it's so painful to reconfigure
Kenley has joined #scopehal
<azonenberg> Kenley: so you're "kench" on github right?
<Kenley> Correct.
<Kenley> I'm not really much of an IRC person these days.
<azonenberg> Invites sent
<_whitenotifier-f> [scopehal-apps] azonenberg opened issue #239: Continuous integration setup - https://git.io/JTByv
<_whitenotifier-f> [scopehal-apps] azonenberg labeled issue #239: Continuous integration setup - https://git.io/JTByv
<_whitenotifier-f> [scopehal] azonenberg added user kench - https://git.io/JTBRg
<_whitenotifier-f> [scopehal-apps] azonenberg added user kench - https://git.io/JTBRg
<_whitenotifier-f> [scopehal-apps] azonenberg opened issue #240: Add static analysis to CI builds - https://git.io/JTByU
<_whitenotifier-f> [scopehal-apps] azonenberg labeled issue #240: Add static analysis to CI builds - https://git.io/JTByU
<Kenley> I've been told that GitHub Actions is the new hotness for continuous integration, but I don't know much about it.
<azonenberg> Well, sounds like something to look into
<azonenberg> I guess the first step will be for me to look at and merge your travis cmake stuff then enable travis on the repo
<azonenberg> i'll do that in a few mins
<azonenberg> meanwhile, you work on getting a windows travis builder working?
<Kenley> There's a experimental Windows Travis CI configuration, but it OOMs.
<Kenley> Windows support on Travis is not great.
<azonenberg> If you want to explore other CI tools instead of travis that works too
<Kenley> GitHub Actions is a reskinned version of Azure Pipelines and it is what the OBS Studio folks use for Windows builds.
<Kenley> I assume it works better because of first-party support.
<azonenberg> Want to try it out on your fork and see if it works better?
<Kenley> Will do.
<azonenberg> then i'll hold off on merging your travis stuff in case we decide not to use it?
<Kenley> OK.
<azonenberg> I dont want to have one ci tool for linux and one for windows, that's just dumb
<azonenberg> so let's find one that we can make work for both
<_whitenotifier-f> [scopehal-apps] azonenberg assigned issue #239: Continuous integration setup - https://git.io/JTByv
<_whitenotifier-f> [scopehal-apps] kench edited pull request #238: Add continuous integration configuration - https://git.io/JTBz6
<_whitenotifier-f> [scopehal-apps] kench synchronize pull request #238: Add continuous integration configuration - https://git.io/JTBz6
<_whitenotifier-f> [scopehal-apps] kench synchronize pull request #238: Add continuous integration configuration - https://git.io/JTBz6
<_whitenotifier-f> [scopehal-apps] kench synchronize pull request #238: Add continuous integration configuration - https://git.io/JTBz6
<_whitenotifier-f> [scopehal-apps] kench synchronize pull request #238: Add continuous integration configuration - https://git.io/JTBz6
Kenley has quit [Read error: Connection reset by peer]
<azonenberg> lain Kliment kc8apf: So i just finished another test of the AKL-PT2 prototype
<azonenberg> it turns out mounting is actually extremely critical to get good performance out of it
<azonenberg> In particular, you cannot mount it flat on the PCB because you get capacitive coupling through the polyimide substrate to the signal trace and ground plane from the flip chip resistors. It's only 100um thick so the coupling is quite strong
<azonenberg> When i glued a nylon washer under the board body to space it up a bit, the response flattened out massively. Long term plan is to use an 0.8mm or so FR4 stiffener to accomplish the same goal, but preinstalled from the fab
<azonenberg> Red = S21 when soldered flat on the board
<azonenberg> Navy = S21 when installed as seen in the photo
<azonenberg> Blue = somewhat simplified sim not modeling losses in the full 150mm of transmission line
<azonenberg> I suspect i can get at least that good, probably better, if i move to immAg for the production batch
<azonenberg> Interestingly, lecroy hints at this same effect in a slightly obscure paragraph in the wavelink probe manual
<Kliment> azonenberg: it's flex
<Kliment> azonenberg: solder it then fold it over
<azonenberg> Kliment: I'm putting a fr4 stiffener on the other end of the PCB for the SMA connector to provide mechanical rigidity
<azonenberg> once i do that, adding a second one where the washer is in this photo is a piece of cake
<azonenberg> i am gonna be doing a new board spin with immersion silver anyway because this long line has a lot of loss which i think is at least in part due to the enig
<azonenberg> And at least one respin of the enclosure with mechanical improvements
<azonenberg> The Dymax UV urethane i'm using actually seems to work really well for holding probes on
<azonenberg> it's UV cure so if you screw up applying you can easily rework before curing
<azonenberg> once cured it's strong, but it's weak enough that if you apply enough shear force you can shear it off the board without damage or residue
<Kliment> yeah it's a lovely glue
<azonenberg> you've used the same stuff?
<azonenberg> Dymax 9-911?
<Kliment> I think I've used the loctite equivalent
<azonenberg> ah ok
<Kliment> same properties
<azonenberg> the dymax stuff is also blue/longwave UV curable
<azonenberg> so a standard gel nail curing system will cure it
<azonenberg> no need for hard uv-c etc
<Kliment> yeah they come with a little UV pen
<azonenberg> ah ok yeah mine did not, it just came in a giant syringe clearly meant for automated dispensing
<azonenberg> so i repackaged in smaller 10cc syringes because i couldn't find any opaque 1cc ones
<azonenberg> Proposed next rev AKL-PT2 enclosure
<azonenberg> The shell around the SMA will probably get rounded, mostly looking for comments/thoughts on the strain relief area itself
<azonenberg> i also want to add some chamfers where the barrel meets the thick SMA area
<tnt> I was actually wondering why you had cuts in both directions. It can only flex one way right ? given the flex orientation.
<azonenberg> tnt: it can bend a little bit the other way
<azonenberg> not well, but it can
<azonenberg> I might shrink the cutouts to make it stiffer
<azonenberg> among other things the flex isnt rigidly constrained inside the sleeve
<azonenberg> it can twist a little bit, and move around
promach3 has quit [Quit: killed]
bluecmd[m] has quit [Quit: killed]
jevinskie[m] has quit [Quit: killed]
jevinskie[m] has joined #scopehal
<_whitenotifier-f> [starshipraider] azonenberg pushed 1 commit to master [+0/-0/±1] https://git.io/JTRZ8
<_whitenotifier-f> [starshipraider] azonenberg fa6c196 - v0.2 AKL-PT2 shell
<azonenberg> https://www.antikernel.net/temp/strainrelief-v2-2.png and with some more chamfering etc
promach3 has joined #scopehal
bluecmd[m] has joined #scopehal
<_whitenotifier-f> [starshipraider] azonenberg pushed 1 commit to master [+0/-0/±1] https://git.io/JTRCY
<_whitenotifier-f> [starshipraider] azonenberg 79bec73 - Updated to pass Shapeways DRC
<_whitenotifier-f> [starshipraider] azonenberg pushed 2 commits to master [+3/-0/±1] https://git.io/JTRWi
<_whitenotifier-f> [starshipraider] azonenberg bf57bc0 - Simulations for AKL-PT2
<_whitenotifier-f> [starshipraider] azonenberg f2c9128 - Updated ipcores
alexhw has joined #scopehal
alexhw has quit [Client Quit]
alexhw has joined #scopehal
juli966 has joined #scopehal
bvernoux has joined #scopehal
<azonenberg> bvernoux: did you see the new AKL-PT2 test results?
<azonenberg> with and without the plastic spacer?
<lain> azonenberg: oh that makes a lot of sense, and very interesting @ flat flex coupling
<azonenberg> Yeah
<azonenberg> So the AKL-PT3 design as it stands is a problem
<azonenberg> I need to figure out how to make it work in a tilted form factor