azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | Online hackathon December 19th all day | | Logs:
_whitelogger has joined #scopehal
futarisIRCcloud has joined #scopehal
_whitelogger has joined #scopehal
electronic_eel_ has joined #scopehal
electronic_eel has quit [Ping timeout: 264 seconds]
_whitelogger has joined #scopehal
Degi has quit [Ping timeout: 264 seconds]
Degi has joined #scopehal
balrog has quit [Ping timeout: 260 seconds]
balrog has joined #scopehal
Bird|otherbox has joined #scopehal
Bird|ghosted has quit [Ping timeout: 256 seconds]
Bird|otherbox has quit [Ping timeout: 265 seconds]
Bird|otherbox has joined #scopehal
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
_whitelogger has joined #scopehal
bvernoux has joined #scopehal
Bird|otherbox has quit [Ping timeout: 240 seconds]
Bird|otherbox has joined #scopehal
<azonenberg> So i did a bunch of product searching on digikey. It seems LeCroy had the right idea using SMPM connectors for their active diff probe heads
<azonenberg> They're about the only connector that meets the requirements (small, good to 6+ GHz, push on connection without a nut to screw in, available in SMT right angle)
<azonenberg> BMA, MMCX, and MCX are the main competitors but SMPM seems to win
<azonenberg> the others are only good to ~6 GHz and if i want to aim high, i'd rather use a connector rated for more than that
<azonenberg> the big challenge is that if i'm using two connectors i have to keep them very precisely aligned during soldering so they mate well
<azonenberg> i was looking for something that had smt signal contacts but pth ground/alignment features and havent found anything yet
<miek> SMP is probably worth a look too, that's what keysight uses or their diff probe range
<azonenberg> SMPM is just smaller SMP
<miek> i know, SMP just seems to be more available so you might find the connector style you want
<miek> also, can your fab do cutouts precisely enough to align an edge mount connector?
<azonenberg> edge cutout? i've had trouble with that at oshpark
<azonenberg> twice now i've had to machine the boards after they came in
Bird|otherbox has quit [Ping timeout: 260 seconds]
Bird|otherbox has joined #scopehal
futarisIRCcloud has joined #scopehal
Bird|otherbox has quit [Ping timeout: 264 seconds]
juli966 has joined #scopehal
<azonenberg> ADL5565... *drool*
<azonenberg> this thing is screaming "make a diff probe out of me"
<azonenberg> It's "single supply" but I'd probably set it up to run on +/- 2.5V power
<azonenberg> with Vocm at ground
<azonenberg> it's differential output so i'd terminate one leg and feed the other to the scope
<d1b2> <Elon Musk> too bad they only show cmrr graphs up to 3GHz
<d1b2> <Elon Musk> i've looked at the ADL5565 before, for some other application
<d1b2> <Elon Musk> unfortunately the graph looks like the cmrr will go below 20dB by 6GHz
<azonenberg> d1b2: yeah im still reading the specs
<d1b2> <Elon Musk> d1b2?
<azonenberg> gaah replied to the bot again, lol
<azonenberg> that's the irc-discord bridge bot
<azonenberg> (06:01:19) d1b2: <Elon Musk> too bad they only show cmrr graphs up to 3GHz
<azonenberg> is how your messages look for me
<d1b2> <Elon Musk> oh
<azonenberg> Anyway, i don't have a differential VNA which is kinda the other thing holding me back from doing a diff probe - it would be tricky to fully characterize
<d1b2> <Elon Musk> you can kind of measure cmrr
<azonenberg> i might build one as a proof of concept but i wouldnt feel comfortable offering it for sale if i didnt have the means to properly test and verify
<azonenberg> i meant being able to measure sdd21 etc
<d1b2> <Elon Musk> well you can treat it as a 3 port network
<d1b2> <Elon Musk> many 2 port measurements are sufficient if you have a good 50ohm load
<azonenberg> Yeah thats what i was thinking, it might be possible to massage two separate 2-port measurements into something i can use
<d1b2> <Elon Musk> it can all be derived using signal flow graphs
<azonenberg> Vishay FC0402 50 ohm resistors are... pretty close to ideal
<d1b2> <Elon Musk> i know from experience that up to 3GHz any old 0402 resistor can give you at least 30dB return loss if you tune the pcb layout correctly
<azonenberg> effective resistance is 50 ohms +/- 0.1% at DC and bumps up to 55 ohms at 10 GHz
<azonenberg> then peaks up to around 90 ohms at 30 GHz before falling off
<d1b2> <Elon Musk> i prefer to think in terms of return loss
<d1b2> <Elon Musk> or |S11|
<d1b2> <Elon Musk> there's a imaginary part to the impedance as well
<azonenberg> Yeah. So i just ran a quick simulation. One of these resistors will give you better than -30 dB return loss to 5.6 GHz with no mounting parasitics
<azonenberg> anyway, the other thing i can do for characterizing just the amplifier is use one of the fancy rosenberger sma terminators i bought
<azonenberg> sine the characterization board will have SMA inputs
<azonenberg> then measure s21 from each input to the output
<d1b2> <Elon Musk> sure
<azonenberg> Digikey 1868-1241-ND
<azonenberg> worst case -28.3 dB to 4 GHz, -26.4 to 6 GHz
<d1b2> <Elon Musk> that's good enough
<azonenberg> and still -19.1 or better at 18 GHz
<bvernoux> d1b2, could you remove the <Elon Musk> before each sentence I do not understand why you have such ?
<bvernoux> it is like a robot which speak ;)
Bird|otherbox has joined #scopehal
<azonenberg> bvernoux: It's a bridge to the 1bitsquared discord
<azonenberg> d1b2 is the bridge bot
<bvernoux> ha ok that explain all ;)
<azonenberg> "Elon Musk" is his username on that server
<bvernoux> maybe it is better to avoid bridge ;)
<azonenberg> There's several dozen more people here than the 61 you see on irc
<bvernoux> and use directly IRC ;)
<azonenberg> There were a number of people i wanted to get in on the conversation who refused to use irc
<azonenberg> it was better than nothing :p
<bvernoux> hehe ok
<azonenberg> Meanwhile i refuse to use discord :p
<d1b2> <Elon Musk> i can't get a stable connection on irc
<d1b2> <Elon Musk> when it disconnects i miss messages
<d1b2> <Elon Musk> the good thing about bridging to discord is now i can search the message history
azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | | Logs:
<d1b2> <Elon Musk> freenode doesn't make logs public and there is only one public logger which is run by wq
<bvernoux> I do not see any interest in log ;)
<bvernoux> it is realtime chat ;)
<bvernoux> if you are not here you miss things like in real life ;)
<d1b2> <Elon Musk> more than once i've answered a question i had by using discord search here
<d1b2> <Elon Musk> in the future i can search ADL5565 and find that discussion a while ago
<d1b2> <Elon Musk> and my internet connection is really bad
<d1b2> <Elon Musk> any time i'm on irc you will see join/leave messages from me fill the chat
<apo> bvernoux: logs save lives
<bvernoux> apo, hehe ;)
<apo> you should see a doctor about that eye of course
<apo> yours* wtf
<d1b2> <Elon Musk> if everyone in the discord changes their nickname to Elon Musk you won't be able to tell who is talking from the irc side
<azonenberg> lolol
<_whitenotifier> [scopehal-docs] azonenberg pushed 1 commit to master [+0/-0/±1]
<_whitenotifier> [scopehal-docs] azonenberg 0526b8e - Updates to Agilent driver docs
<_whitenotifier> [scopehal-apps] azonenberg opened issue #304: Add arguments to disable certain feature-detect flags for easier testing of fallback versions of shaders etc -
<_whitenotifier> [scopehal-apps] azonenberg labeled issue #304: Add arguments to disable certain feature-detect flags for easier testing of fallback versions of shaders etc -
<_whitenotifier> [scopehal-apps] azonenberg labeled issue #304: Add arguments to disable certain feature-detect flags for easier testing of fallback versions of shaders etc -
<_whitenotifier> [scopehal-apps] azonenberg pushed 4 commits to master [+3/-3/±12]
<_whitenotifier> [scopehal-apps] azonenberg 20f86e7 - Updated docs
<_whitenotifier> [scopehal-apps] azonenberg b141460 - Shader: can now load and concatenate two files in shaders
<_whitenotifier> [scopehal-apps] azonenberg d464677 - Added --noglint64 flag to force GL_ARB_gpu_shader_int64 off for testing even if supported. See #304
<_whitenotifier> [scopehal-apps] azonenberg 2b37630 - Major refactoring of waveform rendering shaders to remove duplicated code and improve performance
<azonenberg> And by parallelizing fill and a few other things, we now get about a 3x speedup on waveform rendering
<azonenberg> especially important when zoomed out on high memory depths
<monochroma> oooo
<azonenberg> This is on top of refactoring all of the shader code so that i don't have three different copy pasted almost-identical versions of the rendering routine slightly specialized for analog, digital, and histogram paths with and without int64 support
<azonenberg> now it's a couple of different files i concatenate at load time to piece together a shader for my specific configuration
<azonenberg> and nothing is implemented more than once
<azonenberg> I think i'm going to put together a demo of the Tek MSO6 + WaveRunner 8404 soonish. I have some more stuff to finish on the mso6, specifically the function generator, but i don't need that for the demo
<azonenberg> Soooo I'm puzzling over something (lain Kliment miek any ideas?)
<azonenberg> I took a look at the lecroy Dx20 probe tips to see how their active diff probe was designed
<azonenberg> It looks like all three versions of the tip are approximately 2050 ohms series resistance in each leg
<azonenberg> The Dx20-SI has 100 ohm floating axial resistors then 1950 ohm SMT resistors behind them
<azonenberg> the Dx20-SP has 2050 ohm resistors
<azonenberg> and the Dx20-QC has 2K ohm SMT then 56 ohm axial tip resistors
<azonenberg> In each case the layout appears to be a somewhat loosely coupled 100 ohm diffpair on the top side referenced to ground on the bottom side. The ground plane runs up to the SMT resistors and possibly partially under them
<azonenberg> What i'm curious about is, how do the return currents work in this model? there is no connection between the ground plane and the signal contacts
<azonenberg> it's only connected at the amplifier side
<azonenberg> *very* interesting...
<azonenberg> this is a 100 ohm diffpair with a damping resistor in each leg, fed from port 1 (foreground)
<azonenberg> the ground plane runs up to and not past the resistors
<azonenberg> it looks like you get a U-shaped current loop forming in the ground plane coming into and out of the far side
<azonenberg> which is kinda what my intuition said was going to happen. but cool to see it validated
<_whitenotifier> [starshipraider] azonenberg pushed 1 commit to master [+4/-0/±0]
<_whitenotifier> [starshipraider] azonenberg 1f45c07 - Initial version of AKL-AD1 using SMAs
bvernoux has quit [Read error: Connection reset by peer]