azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | https://github.com/azonenberg/scopehal-apps | Logs: https://libera.irclog.whitequark.org/scopehal
nelgau has joined #scopehal
nelgau has quit [Ping timeout: 240 seconds]
Degi_ has joined #scopehal
Degi has quit [Ping timeout: 256 seconds]
Degi_ is now known as Degi
<agg> azonenberg: I enjoyed your probe video, thanks for posting it! do you think a 50r-tip resistive probe might be a reasonable low cost option for probing a 1v1 supply? I don't mind the 11mA extra load, and seems it would give usefully less attenuation
<agg> Obviously no offset control so can't get to super low ranges on the scope which might limit the utility
<agg> But I was imagining just soldering a 50r smd resistor and coax to a spare decoupling cap spot
<azonenberg> Yes, that is a possibility. There's a lot of tradeoffs between nothing and a really nice probe that may be good enough for a specific application
<azonenberg> The other option is to try a 0.1 uF cap
<azonenberg> which will AC couple and lose really low frequency transients as well as DC supply droop
<azonenberg> but give you great high frequency performance
<azonenberg> The dedicated power rail probe is essentially this plus a separate low bandwidth DC path
<azonenberg> so if you don't actually *need* the DC path it might be good enough
<azonenberg> Long term I plan to build an open hardware power rail probe
<azonenberg> but i have my hands full with other kinds of probe in the meantime
<agg> I'm trying to debug what causes the whole rail to collapse to 0 for a few ms if I turn on enough ffs in the fpga
<agg> So having the DC content is probably going to be helpful
<agg> But yea, I'll probably try adding a 0.1u and see how it compares
<agg> Would you do just a 0.1u into the coax, no resistor on the pcb side?
<agg> (so that divider is just c on top and r in the scope?)
<azonenberg> Correct
<azonenberg> So this eliminates all DC loading on the PSU and acts like a 50 ohm 1x probe
<agg> Nice, thanks!
nelgau has joined #scopehal
nelgau has quit [Ping timeout: 256 seconds]
nelgau has joined #scopehal
nelgau has quit [Remote host closed the connection]
nelgau has joined #scopehal
miek has quit [Ping timeout: 264 seconds]
<azonenberg> also I'm still trying to find good leads for the AKL-PT5. I'm tentatively using SMPM connectors because they're push-on, small, and good to double digit GHz which is the BW range I'm targeting
<azonenberg> but most SMPM cables are $$$$ and a lot of them are semi-rigid or just really stiff
<azonenberg> I found one from sv microwave that looks nice and is nice and skinny (0.047"), less than half the diameter of the T-Flex 405 my current test cable uses
<azonenberg> But i'm concerned about losses (it's a foot long, not sure if available shorter - i guess i can always measure and de-embed if it's only a few dB)
<azonenberg> And it's also very much not cheap
<azonenberg> i lose a lot of the benefit of a low cost probe head if it has to be used with a $133 cable
miek has joined #scopehal
<azonenberg> So I'm actually contemplating using a probe head with a small amount of high frequency peaking to help compensate for cable loss
nelgau has quit [Remote host closed the connection]
bgamari has quit [Ping timeout: 240 seconds]
bgamari has joined #scopehal
bgamari has quit [Ping timeout: 240 seconds]
bgamari_ has joined #scopehal
bgamari_ has quit [Ping timeout: 240 seconds]
bgamari has joined #scopehal
<agg> well, the resistive probe soldered to the 1v1 rail is incredible in terms of seeing the power rail
<agg> but also the extra 100R load seems to prevent the rail dying
<agg> so I guess problem solved, just leave the probe and scope attached :P
<azonenberg> lol
<agg> for example you can really see the smps going from light-load 100kHZ to actually-working 1.5MHz https://imgur.com/a/1SskBRU
<azonenberg> This is why the dedicated power rail probes are 50K ohms. Much less invasive to the control loop
<agg> at 1v1 the scope's internal offset is enough to see the power rail even at 2mV/V scale
<azonenberg> And yeah the huge offset range of the powrer rail probes is more important for looking at things like 3.3V
<azonenberg> or especially 5/12/24V intermediate rails
<agg> the only thing on this 1v1 rail is the fpga, which goes from very low static current to currently about 600mA atm quite quickly
<azonenberg> Have you tried the capacitive probe option yet?
<agg> so I wouldn't have thought the extra 11mA makes _that_ much difference, but I wonder if having 11mA more load when it transitions from power-save to full regulation makes a difference
<agg> not yet, got a bunch more playing around with this setup before I wanna resolder it
<azonenberg> (also if you get really stuck and have a reproducible test setup you can send me a board and I'll see what the RP4030 makes of it)
<azonenberg> Or if you just want to see how close your cheap setup got to the expensive option
<azonenberg> In future versions of my probing talk I want to include test results of a bunch of low cost open hardware / DIY options compared to $$$$ commercial probes
<agg> thanks for the offer! atm I'm hoping this won't be a huge issue, not least because it only happens when I put every single flipflop in a chain and run at 400MHz (on this ECP5)
<azonenberg> My expectation is that they won't be quite as good but will be a tiny fraction of the cost
<azonenberg> so much better value per $$ for 80-90% of applications
<agg> yea, exactly
<agg> certainly this setup is performing miles better than the rc probe, either handheld or the sensepeek ones
<agg> i.e. I can actually resolve the 20mVpp ripple in power-save mode and the 3mV or so in regulation, and see the transition and the frequencies...
<agg> it was all just a hash previously
<azonenberg> Yep :D
<azonenberg> and yeah the sensepeek probes are not half bad if you solder up your own low-L ground, they're on par with most other R-C divider probes at that point
<azonenberg> but the stock long ground lead is awful as you can probably imagine
<agg> yea, I have a little wire soldered on, but it's a bit fiddly
<azonenberg> I really want to make both resistive and active differential probes for the platform
<agg> which is a shame as the whole setup otherwise is great at removing fiddliness
<azonenberg> I have a ~3 GHz resistive probe prototype that has been on ice for a couple of years now
<azonenberg> sensepeek sent me some dev material then ghosted me
<azonenberg> i need to reach out to them again but don't want to until I get time to actually spend on it
<agg> yea, it would be very neat for sure
<agg> resistive probe in that form factor seems like it wouldn't need to be mad expensive either
<azonenberg> oh it wouldn't
<agg> how would you handle grounding?
<azonenberg> let me dig up pics of my prototype
<azonenberg> That is one of the unsolved problems. Easy mode is to use the same long ground lead as before
<azonenberg> but thanks to the lower capacitance get much better performance
<azonenberg> i tested a rigid copper wire ground lead which worked great but was super fiddly
<azonenberg> i have some ideas around pogos etc
<agg> when you say long ground lead, I don't recall what mine even came with but I usually use a very short f-f jumper to a sp10 probe
<agg> obviously still something like 4" of ground
<agg> but perhaps a bit better than the croc clip on a handheld rc
<agg> yea, that's pretty similar grounding to what I have atm (although my wire is soldered to the tin)
<azonenberg> This is my prototype resistive probe though
<agg> certainly works but yea, fiddly
<agg> yea
<azonenberg> https://www.antikernel.net/temp/pt4-s21.png S21 response with both ground leads
<agg> that's the resistive or the sp200?
<azonenberg> the good ground is excellent, flat to like half a dB out to 3 GHz then it peaks
<azonenberg> My probe
<azonenberg> sp200 is crap :p
<azonenberg> I want to push it a bit further BW wise and also figure out a more consistent grounding solution
<azonenberg> 1.25 Gbps 8b/10b test pattern through probe with good ground
<azonenberg> https://www.antikernel.net/temp/akl-pt4-flexground-eye.png and with the long flex ground - looks awful but good enough for protocol decoding still
<azonenberg> and this was my first prototype, it's going to get better
<azonenberg> I think 5+ GHz bandwidth in this form factor is very achievable
<agg> nice!
<agg> i remember thinking 1GHz was loads of bw buying this scope, sigh
<azonenberg> lol
<azonenberg> yeah then you start doing fast stuff and no matter how much you have it's not enough
nelgau has joined #scopehal
<azonenberg> i'm already dreaming of having the $27K LeCroy wants to upgrade my 16 GHz scope to 20 GHz lol
<agg> the onboard fft is just hopeless in the face of glscopeclient
<agg> now up to 1A on 1v1, PLL running at 400MHz, SMPS at 1.5MHz
<agg> ecp5 up to 55C so I think it can go a bit harder
<azonenberg> Have you tried the spectrogram view?
<azonenberg> It does smaller FFTs of configurable length so you can see changes in spectral content over time as a 2D plot
<agg> nice!
<azonenberg> great for seeing power rail spectral lines come and go as activity changes
<agg> my capture has no time-varying content at the moment though
<azonenberg> ah ok
<azonenberg> i was thinking for your other one with the big transient
<agg> that would be cool but I don't have the memory depth
<azonenberg> ah ok
<agg> keysight's stingy memory
<azonenberg> yeah it works best with 1e6 points or more
<agg> I am at 1e6 points and 5GS/s atm
<azonenberg> And i have some beautiful spectrogram captures from my waverunner with 1.28e8 samples
<azonenberg> the picoscope can do up to i think 4e9 points
<azonenberg> but is max 1.25 Gsps and 500 MHz BW
<azonenberg> and lacks a power rail probe
<azonenberg> that is in fact one of the reasons i wanted to build one
<azonenberg> i wanted to be able to use the picoscope for lower bw power rail analysis over very long time periods with the 12 bit ADC
<agg> can probably afford to drop the sample rate to get more time capture, but max 4Mpt memory
<agg> yea, doing this over a long time period would be very interesting
<agg> i'll try it out later with actual application gateware
<agg> at the moment this is just thousands of FFs in a row all toggling every clock to drive up power usage and spectral content
<agg> though I'm fairly happy with -40dBm at the clock frequency, pdn can't be completely awful (especially given it's single-sided assembly)
<azonenberg> agg: note that that's the one FFT bin
<azonenberg> you might have some spreading
<azonenberg> if you zoom in you can use dual cursors to integrate total in-band power between them
<agg> aaah very nice
<agg> -38dBm, I'll take it
<azonenberg> I intended it for RF applications looking at total power of a broadband signal
<azonenberg> but you can use it for this kind of stuff too
<azonenberg> I want to improve the UI of the peak tracking logic at some point
<azonenberg> the bubble placement is created from scratch every iteration now
<azonenberg> so they can kinda "vibrate"
<azonenberg> i want it to be stateful to some extent, and if the calculated new position isn't too far from the old one move it less or not at all
<azonenberg> the filter graph also has at least one segfault i haven't caught plus general ui clunkiness
<azonenberg> Lots of little things i need to work on but as the sole UI dev i just don't have the resources
<azonenberg> i keep getting more people contributing scope drivers, and sometimes filters or decodes
<azonenberg> but katharina was the only other person touching UI and i haven't seen her around in months
<agg> I guess the problem is without a scope driver for your scope or a decoder for your protocol you can't use glscopeclient
<agg> less motivation to work on ui :p
<azonenberg> Yeah, that's true
<azonenberg> and i'm not saying we don't need drivers or decodes *too*
<azonenberg> i would just really like to not be the sole ui dev
<azonenberg> that is honestly probably the single biggest thing holding back the project
bvernoux has quit [Quit: Leaving]
bgamari_ has joined #scopehal
bgamari has quit [*.net *.split]
GenTooMan has quit [Ping timeout: 240 seconds]
Bird|otherbox has quit [Ping timeout: 240 seconds]
GenTooMan has joined #scopehal
d1b2 has quit [*.net *.split]
Yamakaja has quit [*.net *.split]
Yamakaja has joined #scopehal
Yamakaja has quit [Changing host]
Yamakaja has joined #scopehal
fridtjof[m] has quit [Ping timeout: 252 seconds]
mikolajw has quit [Ping timeout: 260 seconds]
tiltmesenpai[m] has quit [Ping timeout: 268 seconds]
CarlosEDP has quit [Ping timeout: 240 seconds]
sajattack[m] has quit [Ping timeout: 268 seconds]
d1b2 has joined #scopehal
d1b21 has joined #scopehal
d1b2 has quit [Ping timeout: 256 seconds]
d1b21 is now known as d1b2
Bird|ghosted has joined #scopehal
fridtjof[m] has joined #scopehal
mikolajw has joined #scopehal
sajattack[m] has joined #scopehal
tiltmesenpai[m] has joined #scopehal
mikolajw has quit [Write error: Connection reset by peer]
sajattack[m] has quit [Remote host closed the connection]
fridtjof[m] has quit [Write error: Connection reset by peer]
tiltmesenpai[m] has quit [Remote host closed the connection]
CarlosEDP has joined #scopehal
tiltmesenpai[m] has joined #scopehal
sajattack[m] has joined #scopehal
mikolajw has joined #scopehal
fridtjof[m] has joined #scopehal
tiltmesenpai[m] has quit [Quit: Client limit exceeded: 20000]
CarlosEDP has quit [Quit: Client limit exceeded: 20000]
mikolajw has quit [Quit: Client limit exceeded: 20000]
sajattack[m] has quit [Quit: Client limit exceeded: 20000]
fridtjof[m] has quit [Quit: Client limit exceeded: 20000]
nelgau_ has joined #scopehal
nelgau has quit [Ping timeout: 260 seconds]