<_whitenotifier-e>
[scopehal-apps] azonenberg 6f349e4 - Fixed potential race condition causing WaveformProcessingThread to hang if OscilloscopeWindow::CloseSession() is called at the wrong time. Fixes #405.
<_whitenotifier-e>
[scopehal-apps] azonenberg 66fbf4d - Updated submodules with new dual socket transport
_whitenotifier-e has quit [Ping timeout: 260 seconds]
massi has joined #scopehal
bvernoux has joined #scopehal
<azonenberg>
Update from chinese fab, making the PT5s out of RO4350B for the active layer would only increase cost to $3/pc @ qty 500, plus 150 tooling / 100 shipping. So $1750 = $3.50/probe in PCB cost
<azonenberg>
So basically if I do use them instead of oshpark, upgrading from fr408 class material to rogers is only a few cents (which makes sense, the boards are so tiny a lot of the cost is processing not materials)
massi has quit [Remote host closed the connection]
massi has joined #scopehal
massi has quit [Remote host closed the connection]
<Degi>
Also it looks nicer
<azonenberg>
Lol rogers does look good with the white and gold
<Degi>
Doesn't it usually have less glass fiber wobbles too? (Or is that more of an only cheap FR4 thing?)
<azonenberg>
I would not expect fiber weave effect to be significant for something this small
<azonenberg>
not unless you were out into mm-wave frequencies
<Degi>
Hmm, it could be a problem if the traces are parallel to the weaves, compared to being at 45° to it
<Degi>
(Like one trace could be on top of a weave while another could be inbetween two, giving them somewhat different impedance)
<azonenberg>
Yes, that is an issue for diff pairs
<azonenberg>
less of an issue for a single ended trace as the impedance shift wouldnt be huge. the big problem is diff pair skew caused by slight velocity factor changes over a long distance
<azonenberg>
a small velocity factor change on a ~5mm long run is not something to worry about
<azonenberg>
and the reflection would likely be insignificant too
<electronic_eel>
is rogers as stable as fr4 mechanically? against like bending. i haven't used it before, just heard it is more complicated to handle
<azonenberg>
RO4350B specifically is a hydrocarbon filled glass fiber laminate
<azonenberg>
it seems pretty stiff to me my limited experience
<azonenberg>
but the other thing is, i would not use it for the whole board
<azonenberg>
it would be the top ~0.17mm or so between signal and ground
<azonenberg>
then i'd use a cheap fr4-style substrate like Shengyi S1000-2 to pad it out to 1.6mm for mechanical support
<electronic_eel>
ah, ok, then you won't have to worry about it being brittle or similar
<azonenberg>
Yeah
<bvernoux>
My main issue is to have a multi GHz scope ;)
<azonenberg>
Lol. well i cant help you with that
<bvernoux>
Rigol MSO5K is pretty limited ;)
<bvernoux>
next step is probably Tek MSO64B but I'm a bit disappointed by the speed with glscopeclient when you have tested it
<bvernoux>
it was "slow" in comparison to other scope especially Lecroy
<azonenberg>
Some folks here have been poking and prodding at tek to fix their performance
<electronic_eel>
that would be a 2 ghz scope with probably decent data extraction speed
<electronic_eel>
the sfp+ interface seems to be connected to the fpga, not the applications processor like on the mso5k
<tnt>
Does anyone make cheap-ish sampling scope nowadays ?
<azonenberg>
tnt: only one i know of that's deecnt is the one from Pico and it's still 5 digits
<azonenberg>
but it gets ~25 GHz BW
<azonenberg>
no experience with it
<azonenberg>
but i've seen it on the website
<azonenberg>
electronic_eel: has anyone reached out to rigol about getting a dev unit or something?
<azonenberg>
that might actually be worth looking into
<bvernoux>
for information the Rigol Support does not reply to my Email now
<bvernoux>
about the upgrade of FW for Rigol MSO5K to enhance it for faster SCPI data ...
<electronic_eel>
azonenberg: i haven't reached out to rigol, as i was waiting for them to release the sample code and docs they promised on the flyer. they explicitly promise "open source" code
<azonenberg>
ah ok
<bvernoux>
they promised they will provide new firmware with fix related to that during months (even year now)
<electronic_eel>
let's hope they still deliver on their promises...
<electronic_eel>
somday
<bvernoux>
So it seems they are probably busy on other scope they are developing like Rigol DS70000 Series
<bvernoux>
It is ridiculous with tons of toy/gadget like different touch screen to display stuff
<bvernoux>
the price is also totally crazy for a Rigol Scope which is only 8bits
<bvernoux>
I imagine if it has the same slowness as MSO5K with 1WFM/s over SCPI ;)
<bvernoux>
it is the bling bling scope which crazy price for the performance
<bvernoux>
The most funny is the ENOB ;)
<bvernoux>
it is a bad 8bits behind the same used for MSO5K (maybe just a bit faster)
<bvernoux>
It is 20GSs
<electronic_eel>
bvernoux: could be that rigol management or chinese government wanted some nice specs that look good on paper. something that compares in some aspects to what tek, keysight or lecroy offer in their higher end models
<bvernoux>
they clearly not compare to Tek MSO64B in all case in price & performances
<bvernoux>
Rigol have just boosted their 8bits ADC++ to 20GSPS and want to play with the big guys with even more expensive prices
<bvernoux>
Who want to pay more than 30KUSD for a Rigol scope ;)
<bvernoux>
especially with those specifications which are clearly ridiculous for the price
<tnt>
wtf ... also the screen rotates the wrong way, my gear is above my head.
<bvernoux>
yes the screen rotate wrong way totally ridiculous
<bvernoux>
It is like they show a big car with gadgets ;)
<electronic_eel>
i don't think they will sell many of these...
<bvernoux>
yes I doubt any company want to pay that for a Rigol scope
<azonenberg>
yeah
<bvernoux>
when others have big reputation and calibration/support especially in EU/USA ...
<azonenberg>
even siglent has a better reputation from what i know
<bvernoux>
yes
<bvernoux>
Also they announce 5GHz BW with 20GSPS scope ;)
<bvernoux>
4points for a waveform is not really something great ;)
<d1b2>
<TiltMeSenpai> ...is that tilting screen motorized???
<d1b2>
<TiltMeSenpai> that seems like it's asking for trouble
<bvernoux>
Yes it is motorized I imagine the worse if you have a book ;)
<bvernoux>
or something a board which block the screen ;)
<bvernoux>
tnt, Tek are more realistic when they announce 5GHz BW (or even 8GHz) it is with 50GSPS ;)
<bvernoux>
at least you have more than 6 to 10 points for the signal to measure especially out of RTSA mode
<bvernoux>
haha MEMS oscillator what a bad idea ;)
<azonenberg>
tnt: yes, mine didnt even link up
<azonenberg>
at gig
<azonenberg>
it flapped at 100M and worked fine at 10M
<bvernoux>
I need to hack my board SerDes to lower it to 600Mbps or 300MBps to see it on my Rigol MSO5K ;)
<electronic_eel>
i dislike these wiznet controllers. don't trust their tcp stack
<bvernoux>
as 1.2Gbps is definitely too much I see only some aliasing ...
<Degi>
Hmm, considering the DS8000-R has a SFP, can't we just wireshark the traffic? It shouldn't be too hard to reverse engineer that
<bvernoux>
electronic_eel, in addition wiznet are tcp/ip stack are not updatable ?
<bvernoux>
-are
<Degi>
tnt: You can use a fast FPGA as a sampling scope, which is better than most oscillscopes < 1000 € in terms of bandwidth, but still not very good (but extremely cheap)
<electronic_eel>
bvernoux: i don't know exactly about if / how you can update them. but it is a blob. so they have to release an update if some bugs are found
<electronic_eel>
Degi: i think you could wireshark the traffic. but you would need some example application that makes full use of it to have something to sniff off. and i'm not sure if they provided that. haven't seen one
<Degi>
Oh, they have a port but nothing to use it? O.o
<Degi>
I thought they had at least some desktop app
<bvernoux>
Degi, maybe it is a web app ;)
<Degi>
Maybe it is VNC
<electronic_eel>
Degi: yeah, it is like only in the datasheet / brochure, but not on their download page
<tnt>
Degi: I have a 500 MHz scope. But I was thinking how to do eye diagrams for stuff like pcie gen3 or usb-3 10G ....
<bvernoux>
tnt, I have exactly same issue with 500MHz scope to do eye diagram on 5Gbps stuff is hard ...
<Degi>
tnt: Hmm, maybe usable up to 2 GHz or so with the better FPGAs, so pcie gen 2 might juuust be doable, my hopes aren't big for anything faster
<Degi>
Though it might be possible to build a sampling scope with the frontend out of discrete parts, which might work to a few GHz for relatively cheap-ish
<bvernoux>
It does USB 3.1 10Gbps HDMI and so on with eye diagram ;)
<bvernoux>
But it is not really an oscilloscope
<bvernoux>
But internally it is definitely an RTSA
<bvernoux>
It cost 15KUSD ;)
<Degi>
What is a RTSA?
<bvernoux>
Signal integrity testing of data lines to 12.8 Gbps per channel
<bvernoux>
Real-time spectral analysis
<Degi>
I mean for signal integrity testing you can probably build a BER tester from two SERDES and an attenuator to see how much link budget you have until the BER increases
<bvernoux>
Piscoscope call that SXRTO
<bvernoux>
it is just sampling scope if you prefer for eye diagram with very low trace length
<bvernoux>
Degi, will be nice to have such signal integrity tool for less than 1KUSD ;)
<bvernoux>
up to 10GBps
<bvernoux>
even something lower cost up to 5GBps
<Degi>
Hmm
<Degi>
I might attempt one soonish, lets see if I can get the frontend under 100 €
<Degi>
Only problem is that I have very little to verify it
<bvernoux>
Yes big problem is to validate it ;)
<bvernoux>
as It requires expensive scope
<bvernoux>
Also it requires a very low jitter clock with option like GPSDO
<bvernoux>
Great latest KiCad 6.99 daily have fixed the 3D Rendering bug/regression
<bvernoux>
especially the raytracing
<bvernoux>
both are fixed now in fact
<Degi>
Actually does it? I think I could just try to measure the risetime of avalanche breakdown of a transistor
<Degi>
Nice!
<Degi>
Sometimes I have a bug where the window stops rendering and I have to reopen it (in eeschema mostly), I wonder if thats fixed too
<bvernoux>
They have done some HUGE optimizations for PCB too
<bvernoux>
for DRC
<bvernoux>
Jeff Young have done tons of Performance Improvements on KiCad 6.99
<bvernoux>
with tons of bugfix
<bvernoux>
I'm testing latest nightly from today => kicad-nightly-r26965.f218da6ea5-x86_64.exe
<bvernoux>
ERC on schema is also crazy fast now
<azonenberg>
bvernoux: i'm just upgrading myself to v6 now, but on stable. i think 6.01 or something now
<azonenberg>
havent done a design on it yet
<bvernoux>
I test only the 6.99 nightly on my design on backup for test purpose ;)
<bvernoux>
Just to see what have changed and if annoying bug are fixed
<bvernoux>
@azonenberg, I'm using KiCad 6.02 it work fine and I have built lot of design with it without any issue
<bvernoux>
the only issue was related to a regression with the 3D rendering so it is not really an issue with schematic or board design ;)
<bvernoux>
@azonenberg, I have validated also the USB 3.0 SS Impedance matching which was perfect ;)
<bvernoux>
@azonenberg, reaching 330MB/s with my old PC over USB 3.0 SS and libusb (just basic bulk transfer of 4K buffers)
<bvernoux>
The most complex for USB 3.0 SS is to trace the states/stuff to optimize it ;)
<bvernoux>
any trace over an UART just breaking everything as timing are very tight for lot of stuff on USB 3.0 SS
<bvernoux>
I'm working to trace internal stuff over SerDes @1.2Gbps but next step is to use a 32bits Bus to reach about 4Gbps data trace over an other USB 3.0 dedicated port ;)
<_whitenotifier-1>
[scopehal-apps] azonenberg dd2c179 - Updated scopehal to not search all of /usr for plugins. Fixes #393.
<_whitenotifier-1>
[scopehal-apps] azonenberg closed issue #393: glscopeclient looks in /usr/bin for plugins when installed system-wide and tries to dlopen every binary on the system - https://github.com/azonenberg/scopehal-apps/issues/393
<azonenberg>
@louis any interest in taking care of a few build system issues?
<azonenberg>
@louis the next one is https://github.com/azonenberg/scopehal-apps/issues/180, basically getting cppcheck and clang-analyzer tested and integrated into the build. this is partially done but Bird|otherbox hasn't touched it in years
<azonenberg>
So i'm not sure what's left to get it niecly polished
<azonenberg>
https://github.com/azonenberg/scopehal-apps/issues/250, getting the existing unit tests to run during CI builds (there's not a lot of tests, but we should at least be running them so as we add more tests they'll be included in CI)