<azonenberg>
Multi-bit channels still use software rendering, channels are spaced further than they probably should, and my driver for the ILA in particular has a few bugs to track down
<azonenberg>
And there's still no trigger config yet, right now i hard-code "channel 0 == 1'b1" in the driver
<azonenberg>
the ILA IP fully supports fairly complex triggers, i just don't have the UI for it yet
<azonenberg>
(or driver support, or even APIs in scopehal for some modes)
<_whitenotifier-3>
[scopehal] azonenberg pushed 1 commit to master [+0/-0/±1] https://git.io/Jv2zV
<_whitenotifier-3>
[scopehal] azonenberg 238169f - Clock is not a physical channel (can't use as trigger etc)
<_whitenotifier-3>
[scopehal-apps] azonenberg pushed 1 commit to master [+2/-0/±3] https://git.io/Jv2zo
<_whitenotifier-3>
[scopehal-apps] azonenberg edc17a5 - Setup menu finally does something! Added a trigger submenu with an entry for each scope that pops up a trigger-properties dialog. No actual content yet.
<Famine>
azonenberg, i have a really good wtf story for you.
<Famine>
ordered some 50 ohm patch cables from digi (mid range cinch connectivity cables), they were made from 75 ohm belden 8241. ok not a problem, mistakes happen, called digikey and they said they would look into it. got an email back this morning that "The frequency for most applications is so low that the impedance mismatch has no bearing on the performance."
<azonenberg>
WTF??
<azonenberg>
they are 75 ohm by design and marketed as 50?
<Famine>
yea
<azonenberg>
WTF
<tnt>
50 ohm +- 50%
<azonenberg>
that sounds like some sort of fraud
<azonenberg>
lol
<Famine>
digikey also said: "We would want to continue to list these cable assemblies as 50-ohms because the connectors would be the limiting factor for these assemblies which are only rated for 50 ohms."
<azonenberg>
wut?
<azonenberg>
"only rated for 50 ohms"
<Famine>
"I have confirmed this with the MFG in the past"
<azonenberg>
sounds like it's time to invoke my favorite secret weapon, public shaming
<tnt>
What part # ?
<azonenberg>
tweet-mention them and include a few screenshots of the exchange :p
<Famine>
J6536-ND
<azonenberg>
It says RG-59 as the cable type
<azonenberg>
at least that's honest
<Famine>
azonenberg, i'm tempted to send it to the signal path
<azonenberg>
Mouser correctly lists it as 75 ohms
<Famine>
cinch's own website lists it as a "True 50 to 50 Ohm" so i guess that's kind of true. its a 50 ohm bnc to 50 ohm bnc with a piece of 75 ohm coax in between
<azonenberg>
lolol
<azonenberg>
mouser says it's good out to 4 GHz. Which it probably is, given a 75 ohm matched load
<azonenberg>
the datasheet says "connector impedance 50 ohms"
<Famine>
I honestly don't even know how to reply to the email from digi because it came from the application engineers over there
<azonenberg>
Why does this cable even exist
<azonenberg>
50 ohm connectors on 75 ohm cable should not be a thing
<Famine>
its to use up those lower rated 50 ohm bncs of course
<azonenberg>
Lol
<azonenberg>
here's the other thing though
<azonenberg>
you'd be much better off having a tiny point impedance mismatch at the connector
<azonenberg>
using it in a 75 ohm system with a 50 ohm bnc
<azonenberg>
than having the entire length of the cable be mismatched
<Famine>
i modeled it in LTSpice, right around 10MHz the phase and impedance goes out the window
<azonenberg>
not the least bit surprising
<azonenberg>
anyway, it looks like digikey isnt entirely at fault
<azonenberg>
in that cinch is publishing a datasheet with misleading info in it
<azonenberg>
octopart et al list it as 50 ohm, mouser says 75
<azonenberg>
the truth is that it's a frankenstein abomination, but there's no database field fo that
<Famine>
i was totally cool with digikey, until they said they want to keep selling this thing as 50 ohm because no one else has complained
<azonenberg>
yeah that sounds pretty ridiculous
<Famine>
i'll try sending them another polite email asking "wtf?!" and then publicly shame them
<azonenberg>
here's the thing, it's marketed "surface mount" but is a cylinder
<azonenberg>
Edge launching this is going to require some very clever PCB outline manipulation i think
<Famine>
well, it does mount to *A* surface
<azonenberg>
Lol
<azonenberg>
anyway, my tentative thought is to have a tiny cutout in the tip of the PCB with width roughly equal to the portion of the outer rim that projects below the main body
<azonenberg>
so it won't be able to roll side to side
<azonenberg>
So the tip of the PCB as seen from the top will have a /--__--\ shape
<azonenberg>
Looking at the docs, it appears these are intended to be "surface mounted" in a completely different way (open end straddling a hole in the PCB, pin inserted through the hole)
<azonenberg>
But nobody makes edge launch machine pin sockets so I'm going to abuse this one :p
<Famine>
thats probably the way i'd do it, or you could cut a T slot, the top of the T would be clearance the flange and the rest would be just shy of the main body diameter
<azonenberg>
I thought about that but the problem is, i need the connector to edge launch mount to a 0.5mm coplanar waveguide trace
<Famine>
ah, yea then the pin would be at a different Z height
<azonenberg>
The socket is 1.47mm in diameter
<azonenberg>
So i'm going to have to do a transition at the probe tip (spending a while in sonnet modeling THAT)
<azonenberg>
but the socket will be edge launched to the trace
<azonenberg>
the point is, i dont want to make the trace significantly wider than the socket
<azonenberg>
i want the trace narrower than the socket, with the socket tangent to it
<azonenberg>
and then horizontal alignment provided by the flange
<azonenberg>
(2.36mm diameter)
<Famine>
I just looked up sonnet, it looks really cool
<azonenberg>
Yeah I have L2 Basic and plan to upgrade to silver and/or gold in the coming months depending on how budget goes
<azonenberg>
If anybody in this channel is looking for some basic modeling for an OSH project I'll support that at no cost as long as the time commitment isn't too extreme, btw
<azonenberg>
I'd obviously be willing to consult on a for-profit board too but would expect some compensation for that :p
<azonenberg>
if you're not making money on the board i won't charge though
<azonenberg>
The lite edition of sonnet is pretty crippled for any kind of fun RF work, the lack of the thick metal model in particular makes a lot of tight geometry and/or coplanar waveguides pretty much impossible to simulate
<Famine>
I'm guessing it will be fairly awful, it's mostly hand calculations and 3rd party blog info on oshpark boards / rf design
<azonenberg>
ah ok. Rather different from this probe then
<azonenberg>
Which is now gearing up for the SEVENTH pcb spin
<azonenberg>
although v0.6 is the first round that I actually was able to simulate, and is a huge step forward from the previous ones
<Famine>
i'm probably going to order some test pcbs and then beg/plead/bribe my way to some time on a decent VNA
<azonenberg>
Sounds like a plan. I only have an xaVNA which is a lot better than nothing but hardly a "real" VNA
<azonenberg>
I'm eyeing the Harmon Instruments units but it sounds like the 5 GHz version won't be ready for sale until end of year ish, and who knows how far out the 20 GHz one is