azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing |,, | Logs:
<bvernoux> hello
<bvernoux> azonenberg, do you have shared the preliminary data/results on github or somewhere ?
_whitelogger has joined #scopehal
<azonenberg> bvernoux: i'll upload some graphs tomorrow
<bvernoux> great
<azonenberg> actually i cant sleep
<azonenberg> let me get this for you now
<azonenberg> - baseline of the characterization board with no probe and a 50-ohm terminator at the far end
<azonenberg> - v0.2 probe with a single resistor
<azonenberg> - v0.3 probe with five resistors
<azonenberg> - v0.4 probe with five resistors, i think this one is set up for a different attenuation
<azonenberg> interesting that at higher frequencies (past about 2 GHz) the return loss is significantly higher with the v0.4 probe than the bare pcb
<azonenberg> But the actual probe signal is significantly improved
<azonenberg> this is the same plot as before but zoomed in more on the vertical axis
<azonenberg> bvernoux: this is the bare board with the terminator removed
<azonenberg> and these are the v0.4 probe on the DUT with terminator removed
<bvernoux> ha great
<azonenberg> i figure this is better in that it removes the freq response of my cheap amphenol terminator from the graph
<azonenberg> which has over 20 dB of un-flatness :p
<azonenberg> -20 dB at 2.2 GHz, -40 dB at 1.3
<azonenberg> Sooo thoughts?
<bvernoux> I think the series impedance is not very accurate because of Xavna anyway it seems not so bad
<azonenberg> How high do you think this probe is actually usable?
<azonenberg> and yeah i am ignoring the top bar stuff, i'm focused on the s-parameters
<bvernoux> need to check with my VNA t be sure ;)
<bvernoux> to test it up to 6GHz
<bvernoux> but it seems clearly great up to 2GHz
<bvernoux> as i'm not very confident with xavna for everything > 2GHz
<azonenberg> I still have to assemble the v0.5 board in which i tweaked the trace impedance to be too high
<azonenberg> and see if that gives expected results
<bvernoux> do you have wait at least 30mins in order xavna stabilize ?
<azonenberg> no these were fairly quick early measurements
<azonenberg> i dont consider them final at all
<bvernoux> even if it does not take so much time like old VNA it is good for accuracy
<azonenberg> i wanted to do full measurements once i get the other probe rev done
<bvernoux> yes to compare them
<azonenberg> also in the not too distant future i want to buy one of the picoscope probes
<bvernoux> even if value are not accurate it is good for comparison purpose with exactly same setup
<azonenberg> exactly
<azonenberg> at this point i care less about absolute measurements than knowing if one pcb rev is flatter than another
<bvernoux> very interesting anyway
<bvernoux> could you export the s2p files too ?
<azonenberg> i think i have to swap probes and do it both ways
<bvernoux> when you will have time as there is no any hurry
<azonenberg> since its not the full 2-port version
<azonenberg> i didnt see any option to buy that
<bvernoux> yes and invert the probe too in order to check sparams
<bvernoux> the full 2-port is not available me too I was interested by this one
<azonenberg> Once i am done measuring it, though, i want to feed the extracted s2p's into qucs
<azonenberg> and see how close i can get to the actual scope waveforms
<bvernoux> yes very interesting to check vs theory
<bvernoux> I'm disappointed as I have received last part of my 7Ghz full cal kit
<azonenberg> oh, i just realized something i should try
<azonenberg> i need to go to minicircuits and buy a 25 dB 50-ohm attenuator
<bvernoux> but my name was wrong on packet and post does not want to deliver it !!
<azonenberg> :(
<bvernoux> Dr Kirkby has mad a big mistake :(
<azonenberg> Basically i want to see if there are any weird nonlinearities going on in my scope front end
<azonenberg> The signal coming off the probe is about 26 dB weaker than a direct SMA feedthrough
<bvernoux> so I will need to wait he send me again the male load + some options (RP-SMA)
<azonenberg> But i am seeing significantly different results when probe straight through vs with my probe
<bvernoux> 26dB is huge
<bvernoux> 26dB of gain ?
<azonenberg> well with a 20:1 probe
<azonenberg> you're 26 dB below a direct input
<bvernoux> ha yes I have forgotten than point ;)
<azonenberg> So what i want to do is, feed the scope a direct input 26 dB down
<azonenberg> but with nothing else in the way
<azonenberg> And see how that looks compared to my probe
<bvernoux> ha yes great
<bvernoux> and you have similar results compared with probe ?
<azonenberg> well right now i only have 0 dB input vs my probe
<azonenberg> but i am beginning to suspect part of the mismatch is actually lack of flatness in the scope gain stage
<bvernoux> for that a ERASynth Micro could help a lot as the accuracy is very good for the price
<azonenberg> i.e. when i make the scope amplify my probe signal by 26 dB, it introduces distortion
<azonenberg> So it may simply be that a 20:1 probe cannot be as flat as a 1:1 probe
<azonenberg> and thus by trying to make my 20x probe's output look like the 1:1 scope trace, i'm shooting for an impossibility
<bvernoux> yes clearly the scope has an impact
<bvernoux> especially they are not designed to be ultra accurate in amplitude
<azonenberg> By comparing my probe to a 26 dB attenuated input, though
<azonenberg> i should be able to determine how much of the un-flatness is my probe and how much is the scope
<azonenberg> If i can prove that the eye closure is the scope's fault and not my probes i will be a lot happier
<azonenberg> This literally just occurred to me
<bvernoux> anyway with VNA we will have all details
<azonenberg> that i may be seeing frequency response issues of the scope AFE
<azonenberg> yeah my point is, ultimately i care about end to end accuracy of the whole signal chain
<azonenberg> Not just the probe
<bvernoux> as all is corrected it is why I'm impatient to have my full calibration kit as even today I have already very good results with antenna for example
<bvernoux> like TSA400
<bvernoux> i'm vey near the measurement provided with it as it was measured with a VNA
<azonenberg> Hmmm minicircuits doesnt have any 25 dB attenuators
<azonenberg> If i stack a 20 and a 6 i should get equivalent results right? Logarithms add
<bvernoux> yes only 20 IIRC
<bvernoux> else they have an amazing programmable attenuator over Ethernet but quite expensive
<bvernoux> 0.1dB step IIRC
<bvernoux> I plan to buy one in future ;)
<azonenberg> Yeah I'll get a FW-6+ and FW-20+
<bvernoux> I have VAT-20+
<bvernoux> and VAT-10+
<azonenberg> also RCDAT-8000-90
* azonenberg drools
<azonenberg> too bad its $$$$
<bvernoux> minicircuits are quite cheap for the quality
<bvernoux> and they provide the sparams and full specifications for all their stuff ;)
<bvernoux> yes this RCDAT-8000-90 is amazing
<bvernoux> but a bit expensive :(
<azonenberg> also minicircuits sells equalization blocks, innnteresting
<bvernoux> yes they have tons of stuff I really like
<bvernoux> and less expensive then others ;)
<azonenberg> anyway i cant buy any of this for a while
<azonenberg> my credit card got stolen a few days ago so i can only buy things cash until the new card comes in
<azonenberg> which, if i'm lucky, will be tomorrow
<bvernoux> ha bad
<azonenberg> I went around the gas station i suspected with a BTLE sniffer in hopes of finding the skimmer but no lucky
<bvernoux> ha it was stolen at distance ?
<bvernoux> not physically ?
<bvernoux> crazy
<bvernoux> chip with PIN will have avoided that ;)
<bvernoux> as I suspect it is magnetic payment ?
<azonenberg> the gas pumps at this station are chip + zip code
<azonenberg> probably 90% of the customers use the local city's zip code, so that isnt much security
<bvernoux> ha ok great so no USA use more chip
<bvernoux> no->now
<azonenberg> The majority of stuff in the US is chip based now
<azonenberg> but no pin
<bvernoux> ha no pin :(
<azonenberg> in any case most of the skimmers i'm aware of these days live inside the pump between the reader and the rest of the system
<azonenberg> so they probably get cleartext card data
<bvernoux> yes with skimmer it is hard to stop anything
<bvernoux> even when you have the pin as sometimes they either sniff it or record with a camera
<azonenberg> Yep
<azonenberg> most of them have a pinhole camera if a pin is needed
<azonenberg> zip code might provide some security on a highway rest area etc
<azonenberg> closer to a residential neighborhood the vast majority of people will be from the area
<bvernoux> I have seen a video of a guy which plug a skimmer on a payment terminal
<azonenberg> so you can guess easily
<bvernoux> amazing how they guy is fast
<azonenberg> yeeep
<azonenberg> some of them go in the slot and MITM the reader too, i think?
<azonenberg> like between the read head and the mag stripe
<azonenberg> not sure if that's still done
<bvernoux> but anyway it was catched with video
<bvernoux> it was on a payment terminal
<bvernoux> and the mold was perfect
<bvernoux> very thin
<bvernoux> it was only to sniff swip card in fact as for Chip+Pin it is clearly harder/more complexe
<azonenberg> yeah
<azonenberg> anyway i'm slowly settling into the new lab, still have a lot of unpacking to do
<azonenberg> Need to buy some storage cabinets on thursday when i get paid (good lab cabinets are not cheap)
<azonenberg> and i have to shut everything down tomorrow to put in a 6 kVA UPS
<azonenberg> all of the computers and test equipment are fed by a separate panel that the UPS will be powering
<bvernoux> ha great
<bvernoux> lot of things to do I see
<azonenberg> and then the entire wet bench remains pending, that's largely limited by budget
<azonenberg> $1.7K self-contained sink, $450 eyewash, a few assorted accessories
<bvernoux> yes it is expensive to have a good lab
<bvernoux> mine is way simpler just a room ;)
<bvernoux> but instruments take lot of place ...
<azonenberg> Then the fume hood which i cant seem to find the quote for
<bvernoux> ha yes a fume hood is great
<azonenberg> But similar models i see for about $3K from vendors that do publish prices online
<bvernoux> do you plan to also have a place/room for decapping with sulfuric acid ... ?
<azonenberg> Not including the filters which are around $200ish
<azonenberg> The hood will be for decapping and sample prep
<azonenberg> using a mix of nitric and sulfuric acids, solvent cleans, etc
<bvernoux> ha ok in a special room (maybe the one you said you cannot breath ... ?)
<azonenberg> although i might use it for other work in the future, e.g. photoresist processing
<azonenberg> I'm looking at the Air Science PurAir Basic
<bvernoux> I think I will never have lab with sulfuric acid else my wife will kill me ;)
<azonenberg> Lol
<azonenberg> I have proper storage for the stuff already, actual OSHA-approved flammable and corrosive cabinets
<azonenberg> On top of the fume hood the lab has an exhaust fan i can switch on and off as needed, which provides iirc 8 air changes per hour
<azonenberg> Right now i am running it continuously to make up for the lack of air conditioning
<bvernoux> ha great
<azonenberg> once i get A/C working, i will only run it when ventilation is called for and turn it off when not doing anything that could contaminate the air
<azonenberg> I also am looking at picking up a Hakko FA-430 fume extractor for soldering with
<bvernoux> me too air conditioning is next step for my "small lab" and my house as it is impossible to work in july/august ...
<azonenberg> It's 28C in my lab right now
<azonenberg> Looking forward to the winter, which will make it much more comfortable :p
<azonenberg> the several kW of hardware should have no trouble heating it
<bvernoux> today it is rainy and only 25°C ;)
<bvernoux> but anyway anything > 30°C is not good for measurements/work
<azonenberg> Then i just have to worry about getting the heat pump installed prior to the summer
<bvernoux> yes in winter HW heat the room ;)
<azonenberg> Yeah it will be a full mini-split heat pump, but realistically i doubt i will use the heat mode much
<bvernoux> maybe it is not too cold in winter where you live no ?
<bvernoux> here it is very rare to have something less than -5°C outside
<azonenberg> Even 0C is not super common around here
<azonenberg> it freezes occasionally but usually not for a long period of time
<azonenberg> the handful of snow we get in a given year melts within a week typically
<bvernoux> yes like here
<azonenberg> anyway i think i'm starting to get tired again so back to bed :p
<bvernoux> have good night
<bvernoux> bye
_whitelogger has joined #scopehal
bvernoux has quit [Quit: Leaving]