azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | | Logs:
<azonenberg> Yeah
deltab has quit [Ping timeout: 246 seconds]
deltab has joined #scopehal
Degi_ has joined #scopehal
Degi has quit [Ping timeout: 252 seconds]
Degi_ is now known as Degi
deltab has quit [Ping timeout: 240 seconds]
deltab has joined #scopehal
anuejn has quit [Ping timeout: 246 seconds]
anuejn has joined #scopehal
Tost has joined #scopehal
m4ssi has joined #scopehal
m4ssi has quit [Remote host closed the connection]
<azonenberg> So i did a digikey order the other day with four HMCAD1520s on it
<azonenberg> among other stuff i needed
<azonenberg> I already have an artix7 on hand, so now I just need to find some time to spend on engineering to make a single channel ZENNECK prototype
<azonenberg> but the ADCs are secured which was a concern in this supply environment
<azonenberg> also @david.lenfesty: I've got some stuff keeping me busy this week but plan to look at the PD board a bit this weekend probably
<azonenberg> What I might do soonish is try to design an active probe board that mates with it
<azonenberg> some kind of simple placeholder with two fixed pins instead of variable pitch inputs or something
<azonenberg> just to validate performance without long solder-in tails attached, and test the PD device-side things
<azonenberg> Intent is to be an intermediary between the final AKL-AD3 with adjustable pin spacing, and the various prototypes I have now
m4ssi has joined #scopehal
m4ssi has quit [Remote host closed the connection]
sam210723 has quit [Quit: Leaving]
sam210723 has joined #scopehal
<agg> azonenberg: i'm having a great time using glscopeclient to grab 8b10b data off a project and do clock recovery and decode it, just wanted to say thanks :)
<azonenberg> agg: :D
<azonenberg> raw 8b10b or something else on upper protocol layers?
<azonenberg> We currently have decodes for 1000baseX and PCIe gen1/2, SATA is planned
<azonenberg> And building your own custom decode is pretty easy, you can look at the 1000baseX and SATA decodes for reference
<azonenberg> if the raw 8b10b symbols is sufficient, of course, then good for you
<agg> it's a custom protocol on top of 8b10b
<agg> I'll investigate a custom decoder, but for now it's easy enough to just look at the raw symbols
<agg> (and also good to actually check the raw symbols are all working as expected, heh)
<azonenberg> Yeah this is one of the reasons glscopeclient was built to make it easy to do plugins
<azonenberg> So you can write a decode for some in-house protocol in a plugin and easily use it with a stock glscopeclient tree
<agg> yea, seems like it will be very slick
<azonenberg> what scope are you using out of curiosity? seems like most of our userbase is on the lower end side of things, siglents and rigols etc
<azonenberg> 8b10b is typically use on faster stuff
<agg> this is a very slow 50MHz 8b10b, scope is a 1GHz keysight msox3104t
<azonenberg> 50 MHz 8b10b??
<azonenberg> now you've got my attention lol
<agg> like I said, custom protocol :p
<azonenberg> i mean there's nothing *wrong* with that
<azonenberg> i've just never seen CDR used on anything that slow before
<agg> in practice it's synchronous I just didn't have enough probes to capture the clock too
<azonenberg> ah ok
<azonenberg> well, legitimate strategy
<azonenberg> i've done single ended probing of diff pairs with glscopeclient routinely
<agg> (though I do want to do cdr on it in the future actually)
<azonenberg> if you can get a clean enough signal to decode, and aren't trying to do signal integrity measurements, it's a perfectly valid way of doing stuff
<agg> yea, i got the pcbite sp200 probe set today so trying it out, quite the luxury to not have to hold shit in place
<agg> makes doing single ended measurements of each end and then subtracting in glscopeclient feasible
<azonenberg> Did you see my review of the sp200 and notes on the grounding?
<agg> I did a while back when you published it, I was going to review it later today too
<azonenberg> tl;dr solder a little bit of copper wire to the EMI can or the gold plated area on the back of the probe
<azonenberg> and you can make something along the lines of a spring ground
<azonenberg> its a little fiddly but greatly improves performance
<agg> at the moment I have one of their sp10 probes connected with the 1" jumper to the ground pin on the sp200
<agg> so the total length is definitely not ideal but a lot better than the long cable to a pin header or something
<agg> shame they don't do a version with like 0402-sized double spring tips
<azonenberg> I've been talking to them about better options
<azonenberg> the AKL-PT3 is an unreleased prototype of an AKL-PT1 style transmission line probe in pcbite form factor
<agg> now you have my attention, hah
<azonenberg> pretty flat to 2 GHz then starts peaking, totally usable to 4+ GHz with a de-embed to flatten the response
<azonenberg> I need to reach out to them again
<agg> given my scope goes up to 1GHz I thankfully can not worry so much about probes yet
<azonenberg> around christmas they sent me some mounting brackets and stuff
<azonenberg> to see what i'd come up with
<azonenberg> i sent them some early photos and prototype test results, tried to set up a chat to discuss
<azonenberg> and never heard back
<azonenberg> i keep forgetting to follow up
<agg> is my 8b10b, not doing SI measurements but definitely looks like it should be a clean decode
<azonenberg> sorry i meant AKL-PT4
<azonenberg> the PT3 was a failed experimen
<agg> (and at 50M you'd hope so, lol)
<azonenberg> Yeah lol
<azonenberg> I want to make an active differential probe in the same basic form factor but with two pogo pins and adjustable spacing too
<azonenberg> but that's several models out
<agg> yea, that would be sweet
<azonenberg> Pulse response on my 4 GHz scope with the short wire ground
<azonenberg> and with the long alligator clip ground
<azonenberg> still not awful, compared to the horrid ringing of the sp200
<azonenberg> S21 of the AKL-PT4 in red with short ground and blue with alligator clip
<azonenberg> input impedance of SP200 (pink w/ stock ground, cyan w/ short wire ground) vs AKL-PT4 (red w/ alligator clip, blue w/ short wire ground)
<azonenberg> starting at 26 MHz the PT4 has lower loading
<azonenberg> 1.25 Gbps single ended eye patterns through the probe with both ground leads
<agg> Looks good enough to decode even with the long ground
<azonenberg> Yep
<azonenberg> Transmission line probes are far more tolerant of bad grounding than capacitive divider probes like the SP200
<azonenberg> So you can get better performance with a good ground, or acceptable performance with a far crappier ground
<azonenberg> Anyway, i'm not actively working on the PT4 right now as i have other stuff going on, but it's in the pipeline
<azonenberg> I expect the PT1 to launch first as the PT4 will need the same tip compensation network as the PT1 to be useful
<azonenberg> and the PT4 is more mechanically challenging to get right as far as pogo grounds etc go
<agg> yea, makes sense
<azonenberg> But it's coming, as is a ~4 GHz active differential probe in the same approximate form factor
<azonenberg> That is also several projects down the pipe
<agg> I'm looking forward to seeing them come out :) certainly way more inside my budget than keysight active probes
<azonenberg> Yep
<azonenberg> Right now the differential probe roadmap is:
<azonenberg> * AKL-AD1: 4 GHz amplifier, paired with solder in passive tip (AKL-PD1). This is currently blocked on EMC problems I haven't had time to work on (the tip is a very good wifi antenna and the amplifier CMRR is poor)
<azonenberg> * AKL-AD2: 6-10 GHz amplifier, also to be paired with the AKL-PD1. This will require some much more complicated RF design and is also on hold for a while
<agg> Would any of them be useful for small amplitude stuff like measuring psu noise? Don't really need much bandwidth but amplifier in the probe seems useful to get over my scopes relatively noisy input
<agg> Hah, they quickly go beyond the bandwidth of my rf test gear :p
<azonenberg> * AKL-AD3: handheld browser probe, same amplifier circuit as the AKL-AD1 but with a compass-style variable spacing needle pair directly attached to it
<azonenberg> this one will be much more comapct and have less stuff to pick up EMI than the AD1/PD1 combo, so should be easier to get working
<azonenberg> i have two folks on twitter brainstorming mechanical designs now
<azonenberg> as far as PSU noise, for that we might want a different type of architecture
<azonenberg> something with higher gain and lower attenuation
<azonenberg> and high offset capability
<azonenberg> most likely something along the lines of the LeCroy RP4030 or the Keysight/Tek equivalents
<agg> yea, the fancy power rail probes are all very high offset capability, but I wonder if you could get away with just ac coupling for a cheaper simpler option
<azonenberg> those probes are all very similar: AC coupled 50 ohm unity gain active buffer
<azonenberg> then a high offset ~50K ohm DC path
<azonenberg> Designing one of them is on my agenda, but i haven't put any time into it yet
<miek> TSP has a teardown of the keysight equivalent and yeah, the architecture is just like that
<azonenberg> Yeah i know
<azonenberg> I just need to actually try building one :p
<azonenberg> I also want to get an RP4030 to benchmark my design against
<azonenberg> I've been trying to get one for a while, Jon over at LeCroy sales thinks he can get me a refurb from their sales team (former demo unit)
<azonenberg> so i've been holding off on buying a new one to save the $900
<azonenberg> But i wanted one for actual use too so if he can't get me a refurb by summertime i'll likely just buy one new
vup has quit [Ping timeout: 245 seconds]
vup has joined #scopehal
juli969 has joined #scopehal
Tost has quit [Ping timeout: 240 seconds]
bgamari has quit [Ping timeout: 260 seconds]
bgamari has joined #scopehal
m4ssi has joined #scopehal
Tost has joined #scopehal
m4ssi has quit [Remote host closed the connection]
<_whitenotifier-5> [scopehal] azonenberg pushed 1 commit to master [+2/-0/±3]
<_whitenotifier-5> [scopehal] azonenberg 77ed28d - Initial work on Intel eSPI decoder. Can detect x1 or x4 mode, identify opcodes, and pick out some addresses. Full packet parsing nowhere near done. See #407.
<_whitenotifier-5> [scopehal] azonenberg pushed 2 commits to master [+0/-0/±2]
<_whitenotifier-5> [scopehal] azonenberg 72e7815 - MockOscilloscope: correctly handle CSV files with blank lines or #-prefixed comments at start. Fixes #408.
<_whitenotifier-5> [scopehal] azonenberg bdfc6f9 - MockOscilloscope: fixed bug causing waveforms imported from CSV to not be correctly centered in the vertical axis
<_whitenotifier-5> [scopehal] azonenberg closed issue #408: CSV import: don't choke on files with header rows at start -
<_whitenotifier-5> [scopehal-apps] azonenberg pushed 1 commit to master [+0/-0/±1]
<_whitenotifier-5> [scopehal-apps] azonenberg 27f71ad - Updated to latest scopehal
juli969 has quit [Quit: Nettalk6 -]
<_whitenotifier-5> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±1]
<_whitenotifier-5> [scopehal] azonenberg a24f963 - MockOscilloscope: extract device name/serial from Digilent WaveForms CSVs. Fixes #409.
<_whitenotifier-5> [scopehal] azonenberg closed issue #409: CSV import: load name and serial number from Digilent WaveForms CSVs -
apo has quit [Quit: apo]
apo has joined #scopehal
Tost has quit [Ping timeout: 260 seconds]