<benishor>
I just sent back my sds2104x plus to siglent for fixing
<benishor>
two months after I bought it, it worked, I turned it off, turned it on back again and it was dead like a fish
<benishor>
no sounds, no relay, no video signal, nothing
<benishor>
fusible fuse was ok
<benishor>
I would have opened it myself if out of warranty but why bother
<azonenberg>
Yeah i generally let vendors fix their scopes
<tnt>
Heh, I asked Agilent how much to fix mine and it's 350 eur just for them to look at it so huh ...
<benishor>
was it still in warranty?
<azonenberg>
Yeah we have an old tek, i forget the exact model, at $dayjob
<azonenberg>
it had high offsets on a bunch of the channels, failed calibration
<tnt>
benishor: no, by a long shot
<azonenberg>
they wanted several hundred USD to even look
<azonenberg>
and once they looked, they'd let me know how much MORE it would be to fix it, and if they even had the parts to fix it
<azonenberg>
as it was out of official support and repairs were best-effort baased on whatever they had gathering dust on the shelf
<tnt>
benishor: thing is ... I _know_ what's wrong. I even have the exact part number I need to fix it. But although the part is available, it's reserved only for use by service center and they don't want to sell it to me.
<benishor>
that's fucked up
<benishor>
can't you find it on ebay though?
<azonenberg>
ok that's not so good
<tnt>
Couldn't and definitely not new ( and given it's the PSU and its design mean it's going to die because they literally heat up the main cap to 90C apparently on purpose, I wouldn't want a used one ).
<benishor>
I got a brand new Siglent SSA3021X Plus a month ago and I managed to fry a switch from its frontend by leaving the HF antenna connected to it one stormy night (even though the equipment was powered off)
<benishor>
and it turns out the warranty does not cover this
<benishor>
so I had to void the warranty and open it myself, change the switch and hope for the best
<benishor>
and I made it work for 10 eur
<tnt>
Definitely a more critical fault that mine, but ... working scope is better than no scope even if the calib might be ever so slightly off now.
<_whitenotifier-7>
[scopehal] azonenberg ae1c848 - LeCroyOscilloscope: don't send redundant stop commands if trigger is already stopped. Do not try to change trigger-out pulse width on older (SDA 8Zi era) scopes that lack the necessary variable.
<_whitenotifier-7>
[scopehal] azonenberg fe05f5f - VICPSocketTransport: debug trace output now includes scope hostname for easier debug of multi-scope setups
<_whitenotifier-7>
[scopehal] azonenberg 05d7fc6 - Siglent: SetUseExternalRefclk now silently ignores request if there is no ext ref input
<_whitenotifier-7>
[scopehal] azonenberg 2a770c8 - Siglent: external trigger support for 6000 series (probably 2000/5000 as well but not yet tested)
<_whitenotifier-7>
[scopehal-apps] azonenberg 9d516da - Timeline: draw trigger arrows for all enabled scopes. Not yet draggable.
<_whitenotifier-7>
[scopehal-apps] azonenberg 83d93dc - ScopeSyncWizard: cancel timeout when a waveform actually arrives
<_whitenotifier-7>
[scopehal-apps] azonenberg e378443 - Multiscope: now use trigger phase patching to deskew rather than messing with hardware deskew and trigger offset
<Stephie>
I have an old HP 54123a 23ghz sampling scope/32ps tdr which has a good GPIB interface and would be neat to write a glscopeclient integration for
<Stephie>
33ghz*
<Stephie>
I always thought there was a GPIB integration in scopehal but i guess i was wrong
<Stephie>
looking for a probe to use the thing for eye diagrams, but seems all the antikernel probes are not being sold rn
<azonenberg>
correct. They're all in various stages of R&D
<azonenberg>
you can make one off the designs on github if you're brave (and can find the parts)
<azonenberg>
but none are what i would consider fit for general use yet
<azonenberg>
the PT5 is very close
<azonenberg>
it's just a question of which of two resistors to end up speccing in the production rev
<azonenberg>
and that will depend on test results after i get some of each
<azonenberg>
i ordered said resistors in february and they're currently due next month
<Stephie>
that's quite a lead time
<Stephie>
so a PT5 production run is probably a while away?
<azonenberg>
So, depends on your definition of production
<azonenberg>
I ordered a hundred as that was the MOQ
<azonenberg>
(they're almost $9 each so that was quite the hit to my wallet for passives lol)
<azonenberg>
I will be setting aside some of that hundred for R&D use as most of my solder-in probes are now using the same tip resistor architecture
<azonenberg>
the remainder will be available for probes to be sold
<Stephie>
OK, sounds good
<azonenberg>
I'm also looking at partnering with some folks for producing them in higher volume than is feasible for me to DIY at home, so some technology transfer will probably be required there WRT tooling for assembly and such
<azonenberg>
as well as improving tooling for higher volume
<azonenberg>
the jig i use for trimming resistor leads now is not super optimized and is tricky to use
<azonenberg>
anyway, as soon as i figure out which resistor i am using for the PT5
<azonenberg>
i plan to order resistors for PVT1/PVT2 runs
<azonenberg>
i guess that would be PVT2 only actually
<azonenberg>
PVT1 will use what we already have on order
<azonenberg>
PVT2 likely wont' be until end of year given lead times
<azonenberg>
Let me just run down the whole list of probes, there's quite a few
<azonenberg>
AD1: AD1: solder in 2-part active probe similar to lecroy wavelink style (separate amplifier and tip modules), LMH6401 based. On indefinite hold, may or may not revive it
<azonenberg>
AKL-AD1*
<azonenberg>
AKL-AD2: name reserved for an ADL5580 based ~10 GHz differential probe, but no design work has been done due to the awkward common mode voltage on it
<azonenberg>
AKL-AD3: handheld differential browser probe, ADL559 based. Needs a lot more mechanical work and i am also having reflection issues due to impedance mismatches at amplifier input
<azonenberg>
Plan to revisit after the AD4 is completed as the two are architecturally similar
<azonenberg>
AKL-AD4: solder in differential probe, also ADL5569 based. First board spin had a strap pin set wrong which fried the amplifier
<azonenberg>
i reworked the board but the chip was dead
<Stephie>
oops
<azonenberg>
rather than putting a new chip on the board, i took this as an opportunity to fix several more minor issues and respin with a new chip
<azonenberg>
That's at fab now expected to ship a week from tomorrow
<azonenberg>
tl;dr absolute max was +3.6V
<azonenberg>
I strapped it to logic high = +2.5V
<azonenberg>
but i forgot that the chip was using an offset ground at -2.5V
<azonenberg>
so my strap was actually to +5V
<Stephie>
oops
<azonenberg>
Anyway, so i have no idea on RF performance whatsoever because the amplifier did not survive
<azonenberg>
but i've used the 5569 before and it was quite good except for the reflection off the input flex
<azonenberg>
which i'm eliminating by putting the input right at the amplifier so there's no stub to reflect even if it's a bit mismatched :P
<azonenberg>
once i get that working i'll look at improving matching for the AD3, then figure out mechanical design for it
<azonenberg>
So that's it for the differential probes
<azonenberg>
Then there's the power rail probe, tentatively AKL-AP1, which is currently just a part number with no actual engineering done
<Stephie>
active probes are nice for eliminating stubs yeah
<azonenberg>
basic idea is to be a split architecture like the RP4030, N7020A, etc
<azonenberg>
and well, the AD3 is active too but it has a much longer distance from amplifier to tip
<azonenberg>
due to the jaw design
<azonenberg>
i may revisit the architecture as i found the 3d printed body not springy enough
<azonenberg>
i think it will need an actual metal spring
<azonenberg>
anyway, then there's the high impedance active voltage probe, the AKL-AV1
<GenTooMan>
plastic is plastic it's properties mean it will deform permanently to a certain degree more than spring steel
<azonenberg>
GenTooMan: yes i was hoping to keep it in the elastic deformation regime
<azonenberg>
but i did get some plastic deformation anyway
<Stephie>
what's the AV1 designed for?
<azonenberg>
Probing things that a transmission line probe is not suitable for
<azonenberg>
but are still single ended
<azonenberg>
my two immediate use cases are things like moderate speed signals with pullups
<azonenberg>
and crystal oscillators
<Stephie>
i see
* GenTooMan
always worries about single ended connections but uses them anyhow.
<azonenberg>
It's an R-C divider stage like a normal 10x passive probe, but 5:1 attenuation and much lower capacitance
<azonenberg>
because the amplifier is 2.4 pF vs the ~17 pF of a scope input, your tip C can be proportionally smaller
<azonenberg>
Anyway first rev worked, and had circa 1.5 GHz bandwidth (about what i was aiming for) but there was too much stray capacitance
<Stephie>
it would seem to be that if you have almost 0 stub length then you don't need to worry about matching, or the transmission line properties of what you're probing at all?
<GenTooMan>
yeah can't use high capacitance to check a crystal oscillator.
<azonenberg>
Yes. the challenge is actually getting there :)
<azonenberg>
anyway with the trip trimmer maxed out (250 - 750 fF)
<azonenberg>
i still was nowhere near wherre i needed to be
<azonenberg>
with another 1.2 pF in parallel with the trimmer i got close
<azonenberg>
rise time is not as fast as i want, i need to investigate
<azonenberg>
but i respun the board with a bunch of ground plane cut out around the tip area
<azonenberg>
which should hopefully reduce that stray capacitance and let me use a smaller tip C and get faster rise times on the high-z input
<Stephie>
good luck!
<azonenberg>
We should find out later this week
<azonenberg>
it's at fab expected to ship thursday
<azonenberg>
so hopefully in hand by weekend
<azonenberg>
i have parts in inventory to populate but will have to order more passives to do a third prototype
<azonenberg>
i only bought two of the trimmers in case i needed a different value
<azonenberg>
Next up, the passive probes
<azonenberg>
AKL-PT1: the OG transmission line probe, the one TSP did a review on
<azonenberg>
it's been on ice for over a year due to that 5 GHz peaking
* josuah
just realize that impedance matching also applies to reactance!
<josuah>
get those caps right at both ends!
<azonenberg>
Yep
<azonenberg>
i have a design for a new tip that should improve both mechanical and electrical properties
<azonenberg>
it will be a pogo rather than a rigid pin, and have integrated damping resistance
<azonenberg>
that's in early stage R&D, waiting on some prototypes to evaluate
<josuah>
it makes much sense for PCBs
<azonenberg>
i'm trying to get around a patent from LeCroy who uses carbon fiber
<azonenberg>
so i want to explore putting an 0402 or 0201 passive *in the tip itself*
<azonenberg>
soldered in line with a pogo needle
<azonenberg>
then some epoxy to reinforce
<azonenberg>
No idea if it will work mechanically but electrically the theory is sound
<azonenberg>
if i can pull it off, it should allow a >10 GHz handheld passive probe with very flat response
<josuah>
the tip might not be held by the 0402/0201 solder after all...
<azonenberg>
because the first resistor will be hundreds of microns from the needle point
<azonenberg>
The idea is that you cut the needle
<azonenberg>
solder the resistor between the two halves
<azonenberg>
then add epoxy around it for mechanical support
<josuah>
10GHz! not even sure I have even *seen* any scope going that high from my own eyes
<azonenberg>
so the epoxy bears most of the force
<azonenberg>
I have a 16 GHz scope and am dreaming of making probes that can keep up
<azonenberg>
anyway, that's very researchy and i have never seen this done
<azonenberg>
the highest BW passive probe on the market that i know of is somewhere around 8-9 GHz
<azonenberg>
and that was i think a 5:1
<josuah>
would LeCroy not be ineterested in opening their patent a bit for such a project?
<Stephie>
ive not seen many companies willing to be flexible with patents
<azonenberg>
i haven't reached out to them as they'd almost certainly demand royalties which would go against the spirit of open source
<azonenberg>
i'd rather design a probe that doesn't step on it and/or wait until it expires and copy their design
<GenTooMan>
josuah, in one syllable no, they also can't be that's just how legal idiocy works.
<azonenberg>
and in keeping with the spirit of open source and low cost, integrating bog-standard passives seems better than carbon fiber custom cut and such anyway
<josuah>
Stephie: I thought it would helps preventing them from being ejected of the scene, but yeah... I doubt LeCroy worries too much about Open Source concurrence
<azonenberg>
josuah: my probes are directly competing with theirs and have 1-2 digits less on the price tag
<azonenberg>
they're not going to help :p
<josuah>
azonenberg: sure, something having a patent does not make it best
<azonenberg>
Anyway, moving on, the AKL-PT2 is essentially dead. That's the solder-in flex castellation design
<azonenberg>
it was too mechanically fragile
<azonenberg>
and had poor performanec if flat on the DUT in Pico's testing
<azonenberg>
it had to be mounted at an angle which was hard
<azonenberg>
the PT5 is designed to come in at an angle with the positioner so it's easy to get full BW
<azonenberg>
without any coupling to the DUT inside the resistors
<azonenberg>
The PT3 is also dead. This was basiclaly a PT2 but instead of a long flex coming off, it had a u.fl for you to hang micro coax off
<azonenberg>
the thin polyimide dielectric gave it the same coupling-to-DUT problems that the PT2 had, but worse
<azonenberg>
The AKL-PT4 is on indefinite hold, it was basically a PT1 repackaged in PCBite form factor
<azonenberg>
i have a few mechanical issues to solve, plus sensepeek ghosted me when i tried to reply to the email thread with some initial prototype characterization results
<azonenberg>
i haven't tried to follow up since as i wnated to get the PT1 nice and flat first
<azonenberg>
then revisit the PT4 using whatever innovations i developed for the PT1
<azonenberg>
And finally, the PT5 is substantially done
<azonenberg>
the only question is which of two resistor topologies i go with
<azonenberg>
350 on the PCB plus 100 damping (only version i have tested to date)
<Stephie>
i always looked at the default pcbite probes and thought that there's no way to add a spring ground
<azonenberg>
or 450 damping and nothing on the PCB (which may have hgiher bandwidth but may also have high freq peaking or other unknown effects, i've never got my hands on the resistor to test)
<Stephie>
and given the differences ive seen at 100mhz with a wire ground vs spring, it's a shame
<azonenberg>
correct. i've done a review on xdevs
<Stephie>
ah
<Stephie>
sad.
<azonenberg>
you can actually get several hundred MHz out of a SP200 probe if you solder your own ground to the PCB
<azonenberg>
but it's very hard to make it work mechanically
<azonenberg>
anyway, the idea of the PT4 was that you could get away with comparatively crappy grounding and have decent rf performance because the probe capacitance was so much less
<azonenberg>
but you could also get really good performance if you managed to get a low L ground
<azonenberg>
my prototype of the PT4 hit like 3 GHz BW almost perfectly flat then had some peaking
<azonenberg>
Anyway, long term if i can re-establish a relationship with sensepeek to get the mechanical parts i need for the intergatino
<azonenberg>
integration*
<azonenberg>
i want to explore an active differential probe in that form factor
<azonenberg>
as well as a turning the PT4 into a double needle transmission line probe with a signal and ground contact at variable spacing
<azonenberg>
so you can have two pogos coming down on the pcb instead of one
<azonenberg>
lots of diff probes do this but i have never seen it done on a passive probe
<azonenberg>
(I might also try to do the same on the PT1, TBD)
<azonenberg>
that is a major enough change i'd probably give it a new part number though
<josuah>
building something exceptional out of common parts...
<josuah>
ok, maybe not *only* common parts
<azonenberg>
The tip resistors i'm using are not common parts
<azonenberg>
the Vishay HML series is so obscure no distributor stocks them
<azonenberg>
digikey doesnt even list them as a non-stock part
<azonenberg>
it's literally not in the catalog
<azonenberg>
the reason i did dev with the 100 ohm is that i was able to get samples of the 100
<azonenberg>
because that is a value they produce every so often since a lot of differential scope probes use them
<azonenberg>
the 450, on the other hand, it seems nobody uses
<azonenberg>
it exists as a valid part number for ordering
<azonenberg>
but i am probably the first person to have ever requested they be made
<azonenberg>
i couldn't get samples because none existed :p
<azonenberg>
and they weren't about to spin up a production line and send me a bunch of parts free
<azonenberg>
so the only way i will ever know if the 450 is suitable for this application is to buy a hundred (MOQ) and if they don't work, too bad i wasted my money :p
<azonenberg>
We should find out in a month or so
<azonenberg>
Anyway, then the final probe design I have in progress is a bog-standard 10x R-C divider probe, except in solder in form factor
<josuah>
when going ahead of my modest electronics experiment, I was impressed by the part of purchasing, planning gears and stocks ahead to get anything done
<azonenberg>
it's basically a PT5 with the 350 ohm resistor swapped out for a 9M and a capacitor in parallel with that
<azonenberg>
then a compensation box at the other end
<azonenberg>
working part name is AKL-AR1
<azonenberg>
sorry AKL-PR1
<azonenberg>
passive, r-c divider
<josuah>
it is getting to a whole other level with this kind of interction with providers
<azonenberg>
i have a housing for the compensation box coming... today i think
<azonenberg>
the board to go in it is at fab, probably another week
<azonenberg>
the special lossy cable is on order but i have no idea on ETA
<azonenberg>
And then i still have to design the tip PCB but that's trivial, like i said it's basically a PT5 clone with some parts swapped
<josuah>
is the RC division here to compensate artifacts or loss at certrain frequencies introduced by the feed line?
<azonenberg>
this is to compensate for the 1M impedance of the scope not being purely resistive
<azonenberg>
there's substantial C too
<azonenberg>
on the order of 15-20 pF
<josuah>
at the scope level, thank you I was curious
<azonenberg>
So if you just put a R in series with that you'd get horrible frequency response, it would turn into a lowpass filter with a very low cutoff freq
<azonenberg>
adding the C at the tip gives you a comparatively flat divider from DC to a couple hundred MHz
<azonenberg>
at least, until you add some L
<azonenberg>
aka the ground wire on a typical probe :p
<azonenberg>
The PR1 is unique in that it's a solder-in high density form factor like the PT5. which again i have never seen on a r-c probe
<josuah>
radiating all over! high-frequencies sees all sort of non-wire components built out of wires
<josuah>
when what one need is nowhere for sale, last solution is to build it
<azonenberg>
Yep
<azonenberg>
anyway, i hope that by mid 2023 most of these designs will be fairly well developed, complete enough that i can recommend people build them if they want
<azonenberg>
and well on their way to being buyable as assembled units
<Stephie>
i wonder what the difference is between the sensepeek 100 and 200mhz probes are when the key factor is grounding lol
<azonenberg>
Good question
<azonenberg>
i can't comment as i have only personally evaluated the 200
<Stephie>
yeah
<azonenberg>
the SP10 is just a direct wire connection
<azonenberg>
there's no attenuator, it's meant for multimeter usage mostly
<azonenberg>
or power supply
<Stephie>
yep
<Stephie>
I should make a test PCB i guess
<Stephie>
i have the TDR, i could buy a SP100 and compare
<azonenberg>
Anyway, so yeah - that's quite the fleet of probes in progress lol
<Stephie>
yep...
<Stephie>
the lecroy test pcb looks like just a transmission line with no soldermask
<Stephie>
is that right?
<azonenberg>
It is
<Stephie>
ok
<azonenberg>
i just use it because it's convenient, other people can get them so results are reproducible
<azonenberg>
and it has a clip for solder-in leads
<Stephie>
didn't you make something like that for testing SMA connectors?
<azonenberg>
yes. i was going to respin it actually, i had some new connector footprints and tweaks i wanted to evaluate
<Stephie>
neat
<azonenberg>
that's been on my todo for six months
<azonenberg>
and i keep finding other things to do instead :p
<Stephie>
same, same
<azonenberg>
anyway, in particular the lecroy fixture is not a great choice for >4 GHz ish probes
<azonenberg>
as losses in the fixture itself become nontrivial
<azonenberg>
i want to look into VNAing it and de-embedding that from my measurements
<azonenberg>
In general, i would love someone to put some time into the network analysis features in libscopehal
<azonenberg>
my two highest priority wants there are
<azonenberg>
1) a filter that, given the s-params of a 2x thru structure, splits it and outputs left and right port de-embedding networks
<azonenberg>
2) a filter that, given a set of s-params measured for a DUT and s-params for a fixture at one side, de-embeds said fixture
<d1b2>
<DanielG> Does glscopeclient currently support USB 2.x decode? I noticed that the manual only had headers for the USB section.
<GyrosGeier>
at some point it might make sense to assimilate the sigrok project
<d1b2>
<DanielG> Does that mean that USB 2.0 decode is currently not natively supported in glscopeclient?
<azonenberg>
danielg: we currently have usb 2.x decoding
<azonenberg>
there are dozens of filters that have skeleton documentation but no writeup
<azonenberg>
but they're still functional and usable
<azonenberg>
gyrosgier: they're GPL last i checked, and architecturally quite different to us
<azonenberg>
and we're so far ahead of them at this point there's not a ton we could gain other than support for a bunch of low end logic analyzers
<azonenberg>
danielg: (on that note, documentation for all of the incomplete headings is very much a todo - just not enough hours in the day for me to write all that. PRs are welcome!)