azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | https://github.com/azonenberg/scopehal-cmake, https://github.com/azonenberg/scopehal-apps, https://github.com/azonenberg/scopehal | Logs: https://freenode.irclog.whitequark.org/scopehal
<marcos__> yeah let me know if I can help. When I run it remotely it also hangs, but not instantly like in the LAN
<azonenberg> Interesting. Probably something that happens after X number of waveforms and over remote the higher latency keeps it alive for a bit
<azonenberg> I'm just busy with lots of other stuff, idk if you saw some of my tweets about it but i'm at the tail end of a major renovation of my house/lab
marcos__ has quit [Ping timeout: 245 seconds]
m4ssi has joined #scopehal
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
marcos__ has joined #scopehal
<marcos__> not really a twitter a user. I hope the renovation is coming along nicely. Summertime is coming, makes it easier to work in the house
<marcos__> in general I'm busy with this project https://hackaday.io/project/164932-axiom-100kw-motor-controller
<marcos__> its the reason I got a fancy scope
marcos__ has quit [Ping timeout: 244 seconds]
futarisIRCcloud has joined #scopehal
marcos__ has joined #scopehal
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
m4ssi has quit [Remote host closed the connection]
marcos__ has quit [Ping timeout: 244 seconds]
bvernoux has joined #scopehal
<bvernoux> hi
<azonenberg_work> o/ bvernoux
<bvernoux> hello
<azonenberg_work> So i did a test with a 200R resistor and the signal looks perhaps a bit cleaner
<azonenberg_work> but still not what i want
<azonenberg_work> (on the probe)
<azonenberg_work> Do you want me to send over the stuff i have now to you, or debug more?
<bvernoux> do you have used special RF resistor ?
<azonenberg_work> vishay FC series flip chip
<bvernoux> ok
<bvernoux> I can continue in // the test
<bvernoux> if you find something better I will do the modifications
<azonenberg_work> it's slow progress because the house project is eating up much of my time and i still dont have a fully set-up lab
<azonenberg_work> I'll pack things up and mail to you shortly then
<bvernoux> yes no problem there is no hurry
<bvernoux> I'm waiting my calibration kit for the VNA to do serious measurements ;)
<azonenberg_work> remind me what you wanted as far as probes go? i have two bare probe pcbs and one 3d printed enclosure right now, plus the other boards we talked about
<bvernoux> as it is a bit a joke to use random short/open/load ;)
<azonenberg_work> did you want one or both assembled?
<bvernoux> sold 5usd on Ebay ;)
<azonenberg_work> just use a piece of wire and a 50 ohm carbon resistor from an old tv
<bvernoux> do like you can
<azonenberg_work> that should be good enough right?
<azonenberg_work> put them on a breadboard with a couple of dip switches to mux between them
<bvernoux> I will check what I need to buy again to assemble the other boards too
<bvernoux> I have wrote a document for that ;)
<azonenberg_work> lol
<bvernoux> as there is lot of stuff for different boards ;)
<azonenberg_work> yeah theres a fair bit of stuff
<bvernoux> I have lost so much time yesterday soldering 3 boards with awfull qfn ;)
<bvernoux> next time I will let do that job to manufacturer they are really better ;)
<bvernoux> anyway because of that (because of soldering bridges) I have done a cleanup on my PCB routing to avoid that ;)
<bvernoux> also my stencil height was too big and there was too much solder paste and so bridges on QFN ...
<azonenberg_work> this is latest result with my probe vs direct sma feedthrough
<azonenberg_work> while not perfect, it's quite good IMO
<bvernoux> it seems not too bad
<azonenberg_work> the main artifact i see in the time domain view is that little hump
<azonenberg_work> at the center of the long "0" period on C4
<bvernoux> it is always with same pattern with 10Gbit ETH ?
<azonenberg_work> This is 1G, not 10G
<bvernoux> ha ok
<azonenberg_work> and it's an 8b/10b idle sequence
<azonenberg_work> sec
<azonenberg_work> There yo ugo
<azonenberg_work> 10G is 64/66b so all LFSR scrambled even for idles
<azonenberg_work> my test signal generator is an artix7 gtp so max of 6.6 Gbps, and my scope is only 1 GHz bw so nyquist at 2 Gbps
<azonenberg_work> test pattern now is 1.25 Gbps
<azonenberg_work> The probe i'm using now is configured roughly the same as a Pico 916
<azonenberg_work> in terms of attenuation/impedance
<azonenberg_work> pico's version claims to be good out to 5 GHz
<azonenberg_work> bvernoux: also idea: if we can measure s-parameters on my probes with your VNA
<azonenberg_work> Might it be possible to use an FFT+iFFT to flatten out the response in post?
<bvernoux> yes I will try your probe to measure s-parameters with my VNA
<bvernoux> it was the idea to use it for that too
<bvernoux> It is why I want to charaterize it up to 6GHz ;)
<azonenberg_work> :)
<bvernoux> 30KHz to 6GHz with an extract of data to s1p
<azonenberg_work> awesome
<bvernoux> I'm still searching a good tool to visualize Touchstone S1p/S2p files
<bvernoux> which also support marker and math computation for capacitance...
<bvernoux> anyway I'm working on a gui tool with Qt to do that and retrieve directly VNA data ;)
<bvernoux> to->tool
<bvernoux> I can already retrieve all data like form1, form4 ... ;)
<bvernoux> now need to convert them to s1p or even s2p for full 2ports measurements (s11, s21, s12, s22)
<bvernoux> I'm planning to also add my Signal Generator to manage it from the gui ;)
<bvernoux> all are GPIB
<azonenberg_work> I want to integrate a signal generator controller API/UI with glscopeclient
<azonenberg_work> because i need to control scopes that have integrated signal gens
<azonenberg_work> and most cannot support two separate connections (one for scope, one for siggen) at once
<azonenberg_work> So i need to do both over the same socket
<bvernoux> yes will be very nice
<azonenberg_work> short term, scopehal has an API for siggens already (though it does not currently support AWG, only sine/square and a few other shapes)
<azonenberg_work> But there's no UI to invoke it
<azonenberg_work> But if you add your generator to scopehal with that API it will work once i make a UI
marcos__ has joined #scopehal
<bvernoux> yep nice feature
<bvernoux> I still do not really know how to interface in a clean way with scopehal ...
<bvernoux> especially as my instruments are USB or USB GPIB so far
<bvernoux> USB GPIB is mainly using VISA dll
<azonenberg> Yeah i havent tried integrating with visa yet
<azonenberg> it's all been custom drivers
<azonenberg> mostly because existing SDKs were either nonexistent, or awful
<azonenberg_work> bvernoux: Thoughts on respinning the probe design with a bunch of 200R resistors in series?
<azonenberg_work> say, five end to end for 1K total
<azonenberg_work> this should be much flatter than using a 1K because less pad-to-pad capacitance
<azonenberg_work> but i'm not sure how this would work because it would no longer be a lumped component
<bvernoux> yes will be interesting to simulate it
<bvernoux> if you have the characteristics of the res
<azonenberg_work> tl;dr is, 26.2 fF pad to pad shunt cap
<azonenberg_work> then 1.89 pH in series with the nominal resistance
<bvernoux> ok perfect
<azonenberg_work> the problem is when you put a bunch of them end to end the assemblage is no longer electrically short
<bvernoux> I will try to simulate it and then I could check with reality ;)
<azonenberg_work> if you want i can hold off on sending you any boards for a bit
<azonenberg_work> respin the probe design with five resistor footprints in series
<azonenberg_work> and test that way
<bvernoux> yes like you prefer
<bvernoux> like that I will characterize both version
<azonenberg_work> because the current one has what i'd consider OK-ish signal quality, but 250R input impedance is a little light
<bvernoux> this one and the other one
<azonenberg_work> Yeah that makes sense
<bvernoux> and anyway in future I will buy other resistor ... for other check
<azonenberg_work> ok so i'll work on that tonight
<bvernoux> perfect
<azonenberg_work> will be 2-3 weeks before i get the pcbs back
<bvernoux> I wil also charaterize the 16x probes if you are interested
<bvernoux> to check if they match well 50Ohms ... with all S1 params
<bvernoux> ...
<bvernoux> to see the differences
<bvernoux> 16x flex probes
futarisIRCcloud has joined #scopehal
<bvernoux> it will be interesting to check how they are up to 6GHz ;)
<bvernoux> I shall receive also some converter to check U.FL ;)
<bvernoux> IIRC they are robust UFL connectors which works reliably more than 20connections like usual cheap UFL
<azonenberg> oh? i havent seen any
<azonenberg> even from hirose
<azonenberg> my research indicated that mmcx was the best compormise between size and longevity
<bvernoux> they come from Dr Kirkby ;)
<bvernoux> I have tried to check what is it as I have some boards with UFL anyway I always prefer SMA
<bvernoux> yes mmcx are also nice for fast plug
<monochroma> azonenberg: u.fl is still super tiny surface mount
<azonenberg> yes but u.fl dies quickly
<azonenberg> which is why i went with mmcx
<bvernoux> but IIRC mmcx are often rated to only 500 mating cycles
<azonenberg> 500 >> 10
<azonenberg> :p
<bvernoux> pairing cycles ;)
<bvernoux> there is also some fast SMA plug
<bvernoux> but i do not know if they are good ...
<bvernoux> or something limited to 1Ghz max ...
<bvernoux> do not laugh I have 7mm connectors on my VNA ;) called APC-7 too
<bvernoux> ultra expensive and very special ;)
<bvernoux> advantage over SMA is they have very low VSWR ...
<azonenberg_work> oh also i would need to respin the enclosure too
<monochroma> oooo SMP connector, "Performance up to 40GHz" interesting for a non wave guide interface: https://www.samtec.com/rf/components/50-ohm/smp
<azonenberg_work> or make the probe tip longer
<azonenberg_work> because the current enclosure doesnt have enough internal space for anything but the PCB
<azonenberg_work> a resistor wont fit
<azonenberg_work> i think what the pico probes do is a printed NiCr resistor on the pcb itself
<monochroma> clearly you need to switch to a ceramic substrate
<azonenberg_work> lol
<azonenberg_work> substrate isnt the issue, its the resistors
<monochroma> yeah, but i usually see the printed resistors on ceramic substrates
<monochroma> i want one of those sample cards on page 9
<azonenberg_work> lol
<azonenberg_work> who doesnt
<azonenberg_work> i think i'm gonna do some qucs modeling for five of those 200R resistors in series with 0.5mm between them
<azonenberg_work> and see what it does
<bvernoux> thanks to share the qucs project
<bvernoux> I will check it
<bvernoux> in order we use the same
<azonenberg_work> bvernoux: any ideas on how to cancel the resonance from the probe tip?
<azonenberg_work> thats an unavoidable unterminated stub
<azonenberg_work> so we'll have to compensate somehow?
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
sorear has quit [Ping timeout: 276 seconds]
sorear has joined #scopehal
<bvernoux> azonenberg_work, to be checked with qucs simulation to check different things
<bvernoux> maybe a small cap
<azonenberg_work> how small? like an 0201 passive or two parallel pcb traces? :p
<azonenberg_work> i'm running a few sims to experiment now
futarisIRCcloud has joined #scopehal
<bvernoux> it really depends how VNA measurements or simulation will look like in smith chart in fact
<bvernoux> to check what need to be corrected
<bvernoux> by iteration
<bvernoux> but I imagine something like a very light RF filter
<bvernoux> maybe just with a rf capacitor < 5pF ...
<bvernoux> I have some Johanson RF capacitor for that
<bvernoux> used to tune RF antenna but could be useful for probe ...
<bvernoux> a bit like normal 10Mohm passive probe but here we have 50 Ohms ...
<bvernoux> this paper is good
<bvernoux> to damp the resonance they speak about adding res on top of probe
<azonenberg> bvernoux: so its funny actually, if i put the five resistors in series
<azonenberg> early qucs simulations suggest that this will actually damp it out
<azonenberg> a little bit of peaking on the rising edge but it looks quite nice
<bvernoux> also maybe 450ohms will be the best
<bvernoux> to have 10x
<azonenberg> We can do 4*100 + 50
<azonenberg> having 5 footprints gives us flexibility
<bvernoux> capacitor on tip is not a good idea as we want the small as possitive capacitor ;)
<bvernoux> possible
<azonenberg> to start i was going to do 5*200 or so
<bvernoux> they clearly show to add resistor Rtip improve the things
<bvernoux> even if in diagram it is for active probe and after 4GHz...
<azonenberg_work> bvernoux: i'm talking about a ~5mm long stub though
<azonenberg_work> i cant really put the resistor any closer to the dut unless its a solder-in probe
<bvernoux> hmm yes
<bvernoux> so it is hard to test that case
<bvernoux> this extract of book is interesting
<bvernoux> with 450ohms probe tip frequency range is limited to 500MHz max too
<bvernoux> to be checked with qucs to see if we reproduce that
<azonenberg_work> I think they're saying using a cheap carbon resistor or something
<azonenberg_work> the picoconnect 914 is rated for 4 GHz / 8 Gbps with a 440 ohm input impedance
<azonenberg_work> but that is likely not a lumped 0402 passive
<bvernoux> it will be interesting to charaterize the PicoConnect 914 probe but anyway that will do not help to know what materials they have used ...
<bvernoux> as for a passive probe > 1GHz is amazing
<bvernoux> I will check if I can buy one ;)
<bvernoux> 0.4pF Tip
<bvernoux> we could check what we have with a goo accuracy on the passive probe
<bvernoux> with VNA
<azonenberg_work> Yeah
<azonenberg_work> So my next focus i think is going to be respinning the current GCPW probe design with five resistors instead of one
<azonenberg_work> and then probably tweaking the enclosure a bit
<bvernoux> i'm reading the spec of PicoCOnnect 900 Series and it is clear to obtain similar perf we need 200Ohm to 1KOhm impedance with max 0.4pF
<bvernoux> like them
<bvernoux> could be interesting to test their tip
<bvernoux> they are cheap
<bvernoux> but I doubt that change lot of things anyway I push that in a corner for next time I buy things