azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing |,, | Logs:
Degi has quit [Ping timeout: 240 seconds]
Degi has joined #scopehal
promach3 has quit [Ping timeout: 244 seconds]
promach3 has joined #scopehal
_whitelogger has joined #scopehal
electronic_eel has quit [Ping timeout: 240 seconds]
electronic_eel has joined #scopehal
<_whitenotifier-b> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±2]
<_whitenotifier-b> [scopehal] azonenberg 946ef1b - Fixed several bugs in de-embedding and angle interpolation
<_whitenotifier-b> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±1]
<_whitenotifier-b> [scopehal] azonenberg f0a5a3a - Still some issues with the very beginning of de-embedded waveforms. Truncate until we figure out what's causing that.
<_whitenotifier-b> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±1]
<_whitenotifier-b> [scopehal] azonenberg c156978 - Now truncate end of de-embedded waveforms due to boundary condition issues
<lain> azonenberg: nice!
<azonenberg> that's after applying phase correction too, and some bug fixing
<azonenberg> that pulse after de-embedding looks very much like the actual pulse straight into the scope
<Kliment> how does this work?
<Kliment> and why is it called spaghetti?
m4ssi has joined #scopehal
<azonenberg> Kliment: it's called spaghetti because...
<azonenberg> it's a giant pile of my lossiest cables all connected end to end
<azonenberg> Lol
<azonenberg> And the theory of operation is super simple
<azonenberg> You convert the waveform from time to frequency domain
<azonenberg> Rotate each complex point, backwards, by the S-parameter phase of the channel
<azonenberg> Divide each complex point by the S-parameter gain of the channel
<azonenberg> then convert back to time domain
<Kliment> and where are you getting those parameters from?
m4ssi has quit [Remote host closed the connection]
<azonenberg> Kliment: VNA measurement of the spaghetti
<azonenberg> The typical intended use case of this is when you have either field solver simulation or VNA measurements of a test fixture, cable, probe, etc
<azonenberg> and you want to remove that from your measurements and focus on what the signal looked like at the DUT itself
<azonenberg> for example at the 4 GHz bandwidth limit of my new scope, even my fancy Mini-Circuits "18 GHz" cables have 1.18 dB of insertion loss
<azonenberg> You don't care about what the signal looks like at the scope, you want to know what it looked like coming INTO the cable
<Kliment> Yeah makes perfect sense
<Kliment> I just got my first VNA and I'll actually be using this
<Kliment> but not for high speed signals, just antenna tuning in this case
m4ssi has joined #scopehal
m4ssi has quit [Remote host closed the connection]
bvernoux has joined #scopehal
juli964 has joined #scopehal
alexhw has quit [Remote host closed the connection]
alexhw has joined #scopehal
<_whitenotifier-b> [scopehal-apps] pepijndevos commented on issue #146: glscopeclient hangs on Rigol MSO5000 -
<pepijndevos> This Rigol drive continues to prove hilariously broken
<pepijndevos> s/drive/???firmware???/
<miek> i wonder.. how much of their scpi stuff is happening in software on the linux side? could be some serious yak shaving opportunities... :p
<azonenberg> miek: lol
<azonenberg> Soooo one of my longer term plans for glscopeclient is to be able to either import a saved waveform or synthesize one from an IBIS model
<azonenberg> then do channel emulation to see what that waveform looks like through a lossy channel
<azonenberg> and then run filters like eye patterns etc on it to look at the result
<miek> ooh nice. reminds me of the awesome equalisation stuff shown off in a signal path video
<azonenberg> miek: did you see my de-embedding work?
<azonenberg> raw signal as acquired on top, de-embedded at bottom, through
<azonenberg> on a 1 GHz scope, it will be even more impressive with a higher bw scope i think
<miek> i did, very cool
<azonenberg> i'm gonna try running 1000base-X through that channels on my 4 GHz scope when it comes in on weds and probably film a youtube demo
<_whitenotifier-b> [scopehal-apps] pepijndevos commented on issue #146: glscopeclient hangs on Rigol MSO5000 -
<miek> i wonder whether it would make any sense to apply to limits of a scope itself too? like my scope's got a ~300MHz LPF before the ADC, but it rolls off pretty slowly and i can still measure a signal way up above 450MHz as long as it's not really weak
<pepijndevos> azonenberg, ^^^ ReadRawData probably needs an API change to work reliably with Rigol.
<azonenberg> pepijndevos: File a separate ticket against scopehal for this
<azonenberg> there's a few other pending API changes that need to happen
<pepijndevos> alright
<azonenberg> Great work on the debug so far though
<_whitenotifier-b> [scopehal] pepijndevos opened issue #185: Expose actual number of bytes transferred in ReadRawData and possibly other commands -
<Degi> My 70 MHz rigol can measure like 200 MHz lol
<azonenberg> Degi: i imagine there's a fair bit of attenuation?
<Degi> yes
<Degi> I should upgrade it to 350 MHz sometimes... Probably when the warranty runs out
<miek> wait, you haven't upgraded it yet? i thought that was the whole reason to get an mso5000 :p
<Degi> Hm yes
<Degi> Was too lazy for that, apparently since a firmware update you cant just ssh into the scope
m4ssi has joined #scopehal
bvernoux has quit [Ping timeout: 240 seconds]
bvernoux has joined #scopehal
<azonenberg> bvernoux: o/
<bvernoux> hello
<azonenberg> bvernoux: so i'm starting to try and do more advanced SI work with scopehal
yourfate has quit [Quit: Sayōnara, ihr Trottel.]
<bvernoux> SI ?
<azonenberg> signal integrity
<bvernoux> ha great
<Degi> Hmh why is it phase shifted
<azonenberg> Degi: because the cable has nonzero propagation delay
<Degi> Does it correct for the phase shift in the able
<Degi> Nice
<bvernoux> so you apply a s2p reference to do the de-embedding ?
<azonenberg> Yes. This is a full s-parameter model from VNA-ing the cable
<azonenberg> bvernoux: correct
<bvernoux> ha great
<bvernoux> like on VNA ;)
<azonenberg> I plan to support forward transforms soon as well
<bvernoux> very good things
<azonenberg> so you can model what a signal looks like through an arbitrary channel
<azonenberg> the latest item on my list is to try and build something that simulates a PRBS, or an arbitrary waveform coming from a VCD simulation, through an IBIS buffer
<bvernoux> and add also VNA display ;)
<azonenberg> and generates an ideal view of what your waveform would look like in the time domain right at the chip pad
<bvernoux> do you have seen latest Rigol Scope/VNA ?
<azonenberg> After that, then I can apply a S2P for an arbitrary channel
<bvernoux> they are doing all in one now
<Degi> Huh?
<azonenberg> No i havent
<bvernoux> RSA5000N
<azonenberg> bvernoux: ultimately i want a f/oss hyperlynx competitor
<Degi> hyperlynx?
<azonenberg> bvernoux: and it's not full 2 port? i.e. it only gets you 2 of the 4 s-parameters?
<bvernoux> ha yes we definitely need a f/oss hyperlynx competitor
<bvernoux> as such stuff are so expensive ...
<azonenberg> Yeeaeh
<bvernoux> even f/oss competitor will be expensive but open ;)
<azonenberg> To start i'll be using sonnet as the field solver but i still need to be able to go from IBIS and S2P to an eye pattern
<Degi> Heh maybe a fancy UI for that one foss EM solverß
<Degi> ?
<bvernoux> Does OpenEMS have capability to go as far ?
<bvernoux> (when we will have a nice gui to use it ;)
<Degi> hm it can do EM simulations...
yourfate has joined #scopehal
<bvernoux> anyway it is mandatory for next step on RF advanced stuff
<azonenberg> So tonight I'm going to write an IBIS file parser as i havent found a good free C++ library for it
<azonenberg> should be easy enough, the format doesn't look complex
<azonenberg> To start I will only support a generic i/o buffer model and not include things like package parasitics
<azonenberg> Longer term, IBIS package files actually include info like crosstalk between adjacent balls
<bvernoux> ha nice I have never studied IBIS file but they seems very interesting
<azonenberg> Anyway the part i'm looking at right now is that they have I/V curves for each of the P and N fets in the driver
<Degi> huh
<azonenberg> So i'm trying to think how to simulate the time domain waveform of an I/V curve driving a transmission lin
<bvernoux> azonenberg, I see you are planning to buy a new scope 40GSPS per chan
<azonenberg> in other words, as the input digital signal toggles from low to high, i will be applying one of two I/V curves
<bvernoux> what is the reg ?
<bvernoux> ref
<azonenberg> bvernoux: waverunner 8404m-ms, it's arriving wednesday
<bvernoux> ha great
<azonenberg> it's the big brother to my current 1 GHz scope
<azonenberg> same form factor, same probes, etc
<azonenberg> but 20 Gsps interleaved to 40 (not all 40)
<azonenberg> vs the 10 interleaved to 20 i have now
<azonenberg> And 4 GHz instead of 1 GHz
<bvernoux> yes great
<azonenberg> Same sample rate as my current 2 GHz scope which is out for repair
<azonenberg> but double the bw
<bvernoux> you have also 4 probes which match 4GHz BW ?
<bvernoux> I suspect probes are 1GHz max
<bvernoux> else you need 50Ohms probes ...
<azonenberg> My stock lecroy passive probes are 500 MHz
<bvernoux> ha ok
<azonenberg> then i have a 1.5 GHz active probe, four of the pico 1.5 GHz passive probes, my own 2 GHz probes
<azonenberg> and a single 6 GHz pico transmission line probe
<bvernoux> so 4Ghz is definitely using 50Ohms input probes
<azonenberg> I'm talking to lecroy about getting a higher bw active differential probe too, but that's $$$ and a ways out
<bvernoux> yes active diff probe are crazy expensive especially for >1GHz ...
<azonenberg> I'm also eyeing an 8 GHz 40 Gsps (on all channels) lecroy SDA
<azonenberg> i dont have the cash now but if it's available down the road i'd like it
<bvernoux> and why not something like WAVERUNNER 640Zi ?
<bvernoux> they are cheaper ok they are not supported anymore ....
<azonenberg> Because i got a good price on this one and it fits my existing rack mount and even is getting the software options transferred over
<bvernoux> they are also 40GSPS 4GHz or more
<azonenberg> And it's fully supported
<bvernoux> ha ok great
<bvernoux> it is a second hand ?
<azonenberg> Lecroy factory refurb
<azonenberg> Asking price on ebay was $20K, i got it for $13K plus the 1 GHz scope
<bvernoux> ha great
<bvernoux> so it comes calibrated with a 2 year warranty like a new one ?
<azonenberg> 3 year actually i think
<azonenberg> But yes
<bvernoux> ha ok very nice
<azonenberg> full factory cal and like-new warranty
<bvernoux> especially at that price
<azonenberg> That's the advantage of buying straight from them
<azonenberg> my other lecroy scopes were retired rental fleet units
<bvernoux> personally I will not buy something on Ebay for that reason
<azonenberg> TRS-RenTelCo only has a 1 year warranty on their refurb sales
<azonenberg> This one was listed on ebay but i bought off ebay to negotiate the tradein deal
<azonenberg> i have a direct phone/email contact with their sales rep
<bvernoux> especially if that exceed 3KUSD a refurb is the best from the manufacturer to have warranty
<azonenberg> I'm willing to buy a refurb from a big name like TRS
<azonenberg> But not some rando on ebay
<bvernoux> yes there is too much risk to have something broken else ;)
<bvernoux> and to loose lot of money & time
<azonenberg> bvernoux: anyway, so tonight i'm going to write an IBIS parser
<azonenberg> And then a basic simulator
<bvernoux> maybe it will be interesting to integrate that in KiCad ;)
<bvernoux> to be used with SPICE ...
<bvernoux> on my side I'm checking to have samples of CH569;)
<bvernoux> it seems to be like a little revolution for Gigabit/USB 3.0 Host/Device controller ultra compact and maybe very cheap
<bvernoux> integrating a RISC-V ;)
<bvernoux> details are here but no datasheet is available so far
<bvernoux> the amazing things is the package QFN68 or QFN40
<bvernoux> and of course the Built-in super high-speed USB3.0 control and transceiver (built-in PHY), support USB3.0 Host/Device mode, OTG function, support USB3.0 HUB
<bvernoux> even Hub ;)
<bvernoux> also HSPI is amazing
<bvernoux> with builtin SerDes @1.25Gbps ...
<bvernoux> the HSPI => the fastest transfer speed is about 3.8Gbps
<azonenberg> Reading up on transmission line math is making me remember just how much calculus i've forgotten :P
bvernoux has quit [Quit: Leaving]
m4ssi has quit [Remote host closed the connection]