<Famine>
azonenberg, i see what you mean about the free version of sonnet being super limited
_whitelogger has joined #scopehal
m4ssi has joined #scopehal
<azonenberg>
Famine: yeah
<azonenberg>
L2 Basic has i think 4x the ram cap plus unlocks the thick metal model etc, it runs on Linux, and it's also like two versions newer and has a much nicer UI
<azonenberg>
I plan to upgrade to L2 Silver and then Gold down the road, just can't justify the cash at the moment
<azonenberg>
The SMA match i linked earlier is about middle-of-the-road capacity for L2 Basic. You can go bigger but not much
<azonenberg>
L3 Gold lets you go up to 2GB RAM which is actually enough for serious work, plus it adds support for another signal layer so you can do a signal-ground-signal sandwich
<azonenberg>
then as much as i'd love pro, at least in the near term, the budget will be difficult to justify
<azonenberg>
They do have nice upgrade plans though. If you buy a higher edition while still having an active support contract for the lower one, they only charge you the delta in price
<azonenberg>
and add an extra year of support
<azonenberg>
i.e. if i buy L2 Silver before December, it costs me ~half price and i get support until december 2021
<tnt>
They don't offer a "per-hour" license right ?
<azonenberg>
They have a free 30-day trial, and from what i'm told if you ask sales nicely they will give you an additional 30-day block for a reasonable price
<azonenberg>
No hourly options. That's generally where you hire someone to do the analysis for you and the software comes included
<azonenberg>
Which is why i was making the offer to do exactly that at no cost for OSH projects - because it's probably the only way they'd be able to get it done
<azonenberg>
I mean, given the complexity of the software and the capabilities, it's very reasonably priced and their support is absolutely first rate
<azonenberg>
I encountered a segfault on christmas eve that was blocking my analysis. Emailed support at 9 in the morning not expecting much
<azonenberg>
by 10 i had a reply from the support guy with a workaround
<azonenberg>
by 11 i had an email from one of the actual developers with the root cause identified, a better workaround, and was told the fix would be in the next service release
<azonenberg>
and as much as i would love openems to be usable for this sort of stuff, it's just not. The UI is nonexistent
<azonenberg>
Also, some recent change is breaking debug builds
<azonenberg>
asan reports a heap buffer overflow in FunctionGenerator::FunctionGenerator()
<azonenberg>
doing some debugging, it's during the write *of the vtable*
<azonenberg>
writing 8 bytes at offset 656 into a 658-bit LeCroyVICPOscilloscope object
<azonenberg>
Other drivers appear unaffected
<azonenberg>
i looked at disassembly and confirmed the bug is real, it truly is overrunning the buffer
<azonenberg>
it's not a gcc bug, clang does the same thing
<azonenberg>
I'm at a bit of a loss right now as to how to proceed
<azonenberg>
658-byte*
<_whitenotifier-3>
[scopehal] azonenberg pushed 1 commit to master [+0/-0/±1] https://git.io/JvaAo
<_whitenotifier-3>
[scopehal] azonenberg e6be84b - LeCroyOscilloscope should not inherit base classes virtually
<azonenberg>
so apparently having that be virtual inheritance borked EVERYTHING
<azonenberg>
i still don't understand why, the guys over in ##c++ say it's wrong
<azonenberg>
oook so i figured out the segfault in the multi scope use case
<azonenberg>
the fix will be nontrivial since it relates to the history window
<azonenberg>
basically the history currently assumes one trigger -> one set of waveforms
<azonenberg>
when you have multiple scopes, and triggers that may not yet be perfectly aligned
<azonenberg>
it's not obvious how to decide what trigger events mean what
<azonenberg>
I'm going to add a temporary workaround, but it's not a proper fix
m4ssi has quit [Read error: Connection reset by peer]
m4ssi has joined #scopehal
<azonenberg>
I think short term i'll have one history window per scope
<azonenberg>
that's probably the sanest option
<azonenberg>
but properly aligning things when you click on history in one scope and want to have the other one revert to the synchronized waveform will have to come later
<_whitenotifier-3>
[scopehal-apps] azonenberg pushed 1 commit to master [+0/-0/±6] https://git.io/Jvaja
<_whitenotifier-3>
[scopehal-apps] azonenberg 8e6f7b1 - Fixed several crashes and bugs around multi-scope support. Now have one history window per scope.
<_whitenotifier-3>
[scopehal-apps] azonenberg pushed 1 commit to master [+0/-0/±1] https://git.io/Jvajh
<_whitenotifier-3>
[scopehal-apps] azonenberg b4a9e94 - Minor rendering tweaks to digital channels
<azonenberg>
one history window per scope, not yet fully synced
<azonenberg>
still needs work but it's progress
<lain>
:3
maartenBE has quit [Ping timeout: 255 seconds]
maartenBE has joined #scopehal
<monochroma>
cool!
<_whitenotifier-3>
[scopehal-apps] azonenberg pushed 1 commit to master [+0/-0/±4] https://git.io/JvVfB
<_whitenotifier-3>
[scopehal-apps] azonenberg 828cc89 - Major improvements to packing density of digital waveforms. Fixes #74.
<_whitenotifier-3>
[scopehal-apps] azonenberg closed issue #74: Digital channels should be drawn smaller than analog in mixed-signal WaveformGroup's - https://git.io/Jvgh4
m4ssi has quit [Remote host closed the connection]
<azonenberg>
One TODO is to figure out why the GMII decode looks slightly offset from the other signals
<tnt>
azonenberg: how do you sync multiple scopes ? Just make sure they trigger at the same time ?
<azonenberg>
Right now i'm acquiring from both scopes with no explicit sync
<azonenberg>
Longer term plan is to have a sort of setup wizard that prompts you to make one the "primary", that's the one your trigger is generated from
<azonenberg>
then connect trigger-out from the primary to each of your one or more secondary scopes
<azonenberg>
on an ext trig input or otherwise unused channel
<azonenberg>
Once that's wired up, touch a probe from the primary and each secondary scope, one at a time, to the same signal
<azonenberg>
and it will adjust trigger offset to deskew them
<azonenberg>
ideally every scope would be on a common refclk but for shorter captures you can get away without that
<azonenberg>
with million-point captures you might start to get skew if you 100 MHz ILA and 10 Gsps scope are not exactly 100:1 sample rate
<azonenberg>
There's lots of details to work out, for example what to do if trigger is set up wrong and one of the secondaries either fails to trigger, or triggers when the primary did not
<azonenberg>
and how to make history work sanely across all instruments
<tnt>
Yeah, none of my scopes have refclk inputs :)
<azonenberg>
Pin headers for probe respin came in. Good thing i didn't redo the pcb yet
<azonenberg>
ED10173-ND photo is wrong, the flange is solid and the open end of the connector is at the skinny end
<azonenberg>
looking at the vendor catalog I actually want the mill-max 0339 series, not the 5200 series. So ED11455-ND should actually do what i want
<azonenberg>
1.37mm tube diameter, 1.72mm flange
bvernoux has joined #scopehal
<azonenberg>
~4.8mm long
<azonenberg>
So i think THIS is my new probe tip socket. Finally
* azonenberg
orders 25
<bvernoux>
hello
_whitelogger has joined #scopehal
<azonenberg>
it will have a SMA going into a 50 ohm precision load (two parallel 100 ohm Vishay FC series, flat to ~10 GHz) with a female socket that accepts my probe tip edge launched right in front of the resistors
<azonenberg>
i might make a second copy of the same layout without the load as well
<bvernoux>
Do you want I do measurement on actual version ?
<bvernoux>
with my VNA
<azonenberg>
then a SMA-SMA through trace as well
<bvernoux>
up to 6GHz
<bvernoux>
before to do the final one
<azonenberg>
My plan is to make the characterization board first
<azonenberg>
on RO4350B at multech with sonnet sims to match all of the connectors etc
<azonenberg>
send a copy of that to you and the harmon instruments guy and have you both do some high bw vna measurements
<azonenberg>
i'll also do some pulse response tests with a 40ps risetime pulse gen and monochroma's 5 GHz lecroy
<bvernoux>
yes nice
<bvernoux>
yes will be interesting that harmon instruments test up to 20GHz
<bvernoux>
to check the limit
<azonenberg>
I only simulated my probe to 10 GHz
<azonenberg>
DC to C-band is my realistic goal, DC to X-band would be awesome
<bvernoux>
yes it is the most interesting up to 10GHz (even just 6GHz for > 90% of guys)
<electronic_eel>
IIRC there was this reflection on the last trace I saw here posted in the channel
<electronic_eel>
was that some mismatch at the scope end, reflecting back to the probe, reflecting back to the scope and then shown on the trace?
<azonenberg>
I believe that was a mismatch on the characterization board
<azonenberg>
But i'm actively investigating that
<azonenberg>
i want the new characterization board to get better measurements of that
<electronic_eel>
ah, so you plan to get rid of it with the better simulated/rogers char board
<azonenberg>
It is also possible it was a mismatch at the scope side. monochroma's scope did not seem to see it
<bvernoux>
I have done some minor modifications on my TRL board
<bvernoux>
a v0.2 is WIP ;)
<bvernoux>
maybe I will buy a new SA working up to 26.5GHz with TG 8GHz
_whitelogger has quit [Ping timeout: 256 seconds]
_whitelogger has joined #scopehal
<bvernoux>
a fun things is nanovna 2 specs are amazing for the price ;)
<bvernoux>
up to 3GHz for < 100USD
<bvernoux>
with something like 60dB (even 80dB on some freq) dynamic range
<azonenberg>
Specs for the 760Zi-A btw... 6 GHz on the 50 ohm ProLink inputs (only 3.5 GHz on ProBus inputs, each channel has one of each and you can mux)
<azonenberg>
53ps 20-80% rise time
<bvernoux>
yes I have missed it on Ebay :(
<azonenberg>
same sample rate as my hdo9204, 20 Gsps on 4ch / 40 Gsps on 2ch
<azonenberg>
This isnt on ebay
<bvernoux>
for 9KUSD
<bvernoux>
the version without A
<azonenberg>
it's a rental fleet unit that's also up for sale
<bvernoux>
I was thinking it was a fake auction for the price