azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | https://github.com/glscopeclient/scopehal-apps | Logs: https://libera.irclog.whitequark.org/scopehal
balrog has quit [*.net *.split]
Yamakaja has quit [*.net *.split]
d1b2 has quit [*.net *.split]
Yamakaja has joined #scopehal
d1b2 has joined #scopehal
Yamakaja has quit [Changing host]
Yamakaja has joined #scopehal
balrog has joined #scopehal
Degi has quit [Ping timeout: 248 seconds]
Degi has joined #scopehal
GenTooMan has quit [Ping timeout: 255 seconds]
GenTooMan has joined #scopehal
GenTooMan has quit [Excess Flood]
GenTooMan has joined #scopehal
<_whitenotifier-7> [scopehal] azonenberg labeled issue #673: Think about replacing TestWaveformSource with waveform synthesis filter blocks - https://github.com/glscopeclient/scopehal/issues/673
<_whitenotifier-7> [scopehal] azonenberg opened issue #673: Think about replacing TestWaveformSource with waveform synthesis filter blocks - https://github.com/glscopeclient/scopehal/issues/673
<azonenberg> $ git diff | wc -l
<azonenberg> 2591
<azonenberg> And we can finally compile scopehal.cpp without error
<azonenberg> the whole project? not a chance, that's a ways off
<azonenberg> but it's progress :p
GenTooMan has quit [Ping timeout: 255 seconds]
GenTooMan has joined #scopehal
<d1b2> <azonenberg> Did some failure analysis on one of the sketchy looking tip resistors. Camera had trouble focusing on the mount for a long shot
<d1b2> <azonenberg> The nickel leadwire has a "nail head" formed into the end of it, then it's soldered to what appears to be a relatively standard 0402 sized chip resistor and inserted into the fiber-reinforced plastic cup
<d1b2> <azonenberg> then potted with the overmolding compound
<d1b2> <azonenberg> but as you can see here the potting is completely missing on the upper side of the solder joint, thus providing no reinforcement for it
<d1b2> <azonenberg> a little hard to see but it's also peeling away from the wire on the bottom side, so not supporting it very much there either
<azonenberg> Anyway, I shared this data with Vishay who is doing their own investigation as well
<azonenberg> So we'll see what they come up with
<d1b2> <Darius> nice
<d1b2> <Darius> I mean not nice for you, but cool shots 🙂
<azonenberg> I'm being nice to Vishay and not tweeting these, but figured i'd share with the channel so people know status of things
<azonenberg> anyway, back to refactoring for a while
<electronic_eel> nice shots
<electronic_eel> shouldn't the nail thing also sit a little more centered? that should give a more even cavity for the molding
<azonenberg> Yes, there is also inconsistency in placement of the lead WRT the terminal of the resistor
<azonenberg> That may be a contributing factor but in these photos, the nail head is very close to the bottom side
<azonenberg> but the solder fillet is still pretty uniform
<azonenberg> so the cavity dimensions are about what they should be
<azonenberg> the concerning bit is that the top right is a big void, and at the bottom right you can see the potting is detached from the lead wire
<azonenberg> thus providing no real support on either side
<azonenberg> but at least at the bottom it's still reinforcing the solder joint
<azonenberg> at top, it's not
<electronic_eel> yeah, sure, the potting is the main problem
<azonenberg> There's a few air bubbles and such in the potting but i'm not really worried about those (although some of those bubbles are quite large)
<azonenberg> i also have seen what looks like the potting detaching from the plastic in spots. I want to probably section one of those as well, and see if the whole resistor+potting is maybe sliding out of the plastic cup a bit
<electronic_eel> i think they have to retune the process for these. maybe the machine doing them didn' t run for 5 years and needs some oiling and readjustment
<azonenberg> see there, it looks like the entire contents of the white plastic cup detached and moved right relative to the cup
<azonenberg> And yes, it seems like they have process issues. maybe the black goop is something they bought in 2017 last time lecroy or keysight ordered a thousand of them
<electronic_eel> yeah, there is like a crack
<azonenberg> and it's started to congeal or something and not flow right
<azonenberg> could be lots of things
<azonenberg> and no, it's not a crack
<azonenberg> look closely at the bottom side
<azonenberg> you can see a nice square extrusion
<azonenberg> it looks like an exact negative of the plastic cup
<azonenberg> seems like poor adhesion between the cup and the potting and they're starting to separate
<d1b2> <azonenberg> This is one of the ones I sent them for FA. the one i sectioned looks similar though
<d1b2> <azonenberg> other observed defects: excessive potting with long strings of it dangling out
<d1b2> <azonenberg> (so it's not that they are just not using enough, the volume appears inconsistent)
<d1b2> <azonenberg> Cracks, might be nothing or might be indicative of stresses or voids underneath
<electronic_eel> yeah, the overmolding process really has a problem
<d1b2> <azonenberg> This one looks like the entire overmold on the end separated
<d1b2> <azonenberg> you can see lead through the gap at ecnter
<azonenberg> Anyway, yeah this level of inconsistency is not what i would expect from a well optimized manfuacturing process
<azonenberg> but like i said, it's good that they're not telling me to get lost
<electronic_eel> oh yeah. many other mfgs wouldn't really care about this
<azonenberg> yeah quite the opposite, i'm on an email thread with my sales manager, my local vishay rep, and five other engineers/technicians from around vishay responsible for QA and production
<azonenberg> several of which appear to be based out of the factory in Mexico that actually makes these
<azonenberg> Vishay has a reputation for quality and when they get things wrong, it seems they're willing to go the extra mile to make it right
<azonenberg> The fact that these same resistors are (almost certainly) used by every other big T&M vendor for their probes probably has something to do with it
<azonenberg> I may have been the one who got unlucky and had a bad batch, but imagine how they'd have responded if keysight or lecroy contacted them about the same problem
<azonenberg> i think they're very concerned about the potential for that to happen and want to get ahead of the issue before it affects a higher volume customer
<azonenberg> I also told them we have a production order of our own that is gated on a satisfactory resolution to the process issues
<azonenberg> matt was all ready to put in a volume purchase of the HMLs for the AKL-PT5 production run when i told him i had noticed some weirdness
<azonenberg> and we put the order on hold
<azonenberg> Anyway, i expect it will be a while before we hear very much from them. the box I sent them with the components for FA was only picked up by fedex today
<electronic_eel> my guess is that scheduling a new production run is the thing that will take them longest
<azonenberg> Well yes, when we actually put in the order it will probably be a multi month lead time like the last one
massi has joined #scopehal
<azonenberg> The good news is that i've been screening them visually
<azonenberg> and i have plenty for prototyping and likely even PVT
<azonenberg> at least of the PT5. the AD4 uses twice as many of them so we might run short there for a reasonable sized PVT run
<azonenberg> and then the AV1 uses the 100 ohms, the lot i have of those appears to be better
<azonenberg> there are some sketchy looking bits of overmold but overall not nearly as much voiding and degradation like this
<azonenberg> then the PR1 will use either nothing or 100, i'm testing both configurations and still experimenting to see what will work best
<azonenberg> So i don't think the issue will be gating R&D any time soon
<azonenberg> Also, the diff for the waveform refactor is up to 3800 lines and i'm up to files beginning with M in libscopehal
<azonenberg> (scopeprotocols is still untouched)
<azonenberg> Several more days at this rate
Stephie- has quit [Quit: Fuck this shit, I'm out!]
Stephie has joined #scopehal
<miek> azonenberg: with those fancy rosenberger connectors, do they work purely on compression or does the pin still need to be soldered?
<azonenberg> The 32K243-40ML5? I generally solder the center pin and rely on compression for the ground
<azonenberg> that gives excellent performance but is still fairly easy to take off should i want to recycle the connector
<miek> thanks!
massi has quit [Remote host closed the connection]
<d1b2> <azonenberg> Comparison - i posted this some time ago, this is from March
<d1b2> <azonenberg> one of the 100 ohm HML engineering samples I was sent in January
<d1b2> <azonenberg> sectioned 90 degrees perpendicular to the one i did last night
<d1b2> <azonenberg> (this is looking into the side of the embedded 0402, while yesterday's pics were coming from the top or bottom)
<d1b2> <azonenberg> note that while there is a large void deep inside the plastic cup, the opening of the cup has a solid, uniform filling
<d1b2> <azonenberg> the potting is solidly adhered to the cup and the lead (not peeling away) and is adequately reinforcing the leads
<azonenberg> i went through the pile and i found a couple more that looked sketchy. I'm wondering if it might be instructive to try and replicate the original failure mode
<azonenberg> put it on a PT5, sonicate it, and see if the lead falls off
<azonenberg> i need to build another pt5 with the new minicircuits filter anyway
<azonenberg> So i can just document the hell out of this one and deliberately pick a weak looking resistor, knowing there is a good chance it will fail and need replacement
<d1b2> <xabean> beautiful pictures btw!
<d1b2> <xabean> I'm assuming these cross sections are all potted again in epoxy, then sanded down to the region you want to inspect?
<azonenberg> Correct. They;re put in a circular silicone mold and cast into a block of clear resin, in this case AA-Bond F113 from Atom Adhesives
<azonenberg> The back side of the mount tends to be uneven because of surface tension causing the resin to form a meniscus against the mold
<azonenberg> so i flip the mount over after demolding and rough sand with 220 grit sandpaper until it's flat
<azonenberg> just so it sits evenly on the bench or microscope stage
<azonenberg> then i sand with 220, 500, 1200 grit to reach the plane of interest in the sample
<azonenberg> (wet, to minimize dust generation)
<azonenberg> then once i'm about where i want i rough polish with 2000, 3000, 4000, 5000, 6000, 8000, and 10000 grit sandpaper
<azonenberg> and do final polish with 1um and 500nm diamond paste on a cloth polishing pad
<azonenberg> https://www.youtube.com/watch?v=lmfJ2aRNZqE goes into a lot more detail on my general sample prep procedures for cross sections
<d1b2> <johnsel> subbed
<d1b2> <johnsel> you could use a better mic though lol
<azonenberg> I've gone through several iterations. But the mic isn't the main issue
<azonenberg> the issue is that the ambient noise in the lab is quite loud, around 65 dBA, from the 24/7 HVAC plus all of the cooling fans from various equipment
<azonenberg> i have to do fairly extensive DSP to remove it from recordings
<azonenberg> and that does make voices sound less good
<azonenberg> ideally i'd get like a noise canceling aircraft mic or something
<d1b2> <johnsel> a proper lapel mic can pick up only the voice sound
<d1b2> <johnsel> they solve this exact issue
<azonenberg> yeah if anyone has suggestions on something (exact products i mean) let me know, it's one of my known weak points
<azonenberg> i just know most things i've tried still picked up the ambient noise
<azonenberg> i tried a wireless mic a while back and there was horrific background chirping noises
<azonenberg> not sure if EMI from my lab or the mic was that bad
<d1b2> <johnsel> rode is pretty good stuff, https://rode.com/en/microphones/wireless/wirelessgo and https://rode.com/en/microphones/lavalier-wearable/lavalier-ii though I have to admit to be more familiar with pro audio not so much broadcast though in the end the problems are the same
<d1b2> <xabean> now is when everyone mentions dead kittens
<d1b2> <johnsel> to stuff on the microphone xabean? :p
<d1b2> <xabean> that's what it's called!
<d1b2> <johnsel> that's supposed to lessen the effect of wind
<d1b2> <johnsel> which is different from ambient noise
<d1b2> <johnsel> but yes everyone loves kittens, dead or not
<d1b2> <johnsel> in the meantime though if you have a nvidia card they have the best AI voice filter
<d1b2> <johnsel> it's pretty great, though I would not be surprised if it's windows only
<azonenberg> i do have an nvidia card but it's probably going to be full of other things
<azonenberg> like glscopeclient :p
<azonenberg> ideally i would want something i could just record straight into audacity or my camera
<azonenberg> and yeah, the big puffballs are for reducing wind
<azonenberg> they do nothing about a lab that sounds like somebody is vacuuming in the next room, 24/7
<azonenberg> here's an unedited raw version of one take from that video (not one i ultimately used in the final cut, but that's irrelevant)
<azonenberg> just so you can hear the ambient noise level
<d1b2> <johnsel> I am actually walking back my recommendation, you might be better of with a more directional mic
<azonenberg> Directional may not be enough because the sound is all over the lab and you'll just zero in on bounce from the wall behind me
<azonenberg> i think it has to be an active a-minus-b type cancellation filter
<azonenberg> that has an omni and a directional and only records the delta
<azonenberg> or something to that effect
<azonenberg> i dont know exactly how they work but if people can use them in helicopters and understand each other, they should handle my ambient noise just fine
<d1b2> <johnsel> no, you just need to point it to your mouth
<azonenberg> Anyway, whatever it is, i need to figure out exactly what to get and ideally some way of getting the signal to a PC mic port without excessive noise or dragging cables around the lab while i move
<d1b2> <johnsel> so a cardoid lapel
<d1b2> <johnsel> though that is xlr and balanced and actively powered
<d1b2> <johnsel> not something you can put in a 3.5mm plug
<d1b2> <johnsel> but it is the type of mic that is most appropriate, they're also used to record say a single instrument in an orchestra
<azonenberg> Yeah. but like i said i'm not an a/v professional who knows about this stuff
<d1b2> <johnsel> you could go your approach but those microphones definitely don't have audio quality as a requirement
<azonenberg> if someone can suggest a reasonably priced pipeline that goes from X to a 3.5mm input
<azonenberg> i'm all for it
<azonenberg> i just dont have the time to do the research and pick out all the necessary components and adapters
<d1b2> <xabean> buy a cheap two channel mixer board off amazon that has XLR and RCA or TRS outputs?
<d1b2> <johnsel> did you use a mobo built in sound card?
<azonenberg> i've tried that and a usb sound card
<d1b2> <xabean> that's what I did, but I also had a cheap mic, and I'm not sure what if any phantom power the mixer board I had could provide
<azonenberg> no discernable difference in performance, the mic/noise were the limiting factors
<azonenberg> anyway... requirements are that the final output is 3.5mm into a PC sound card, and that there be no cables on the floor dragging behind me i can trip on
<azonenberg> you can have whatever you want on the rx side going into the pc, and i can handle some kind of portable box on my belt/pocket/whatever to transmit
<d1b2> <johnsel> sure sure but the average pc motherboard has horrible sound quality too, so you might only move the problem
<d1b2> <johnsel> or have bought something already
<azonenberg> ok fine. final output is 3.5mm or usb
<azonenberg> (and is a standard usb audio class not something proprietary that needs a windows driver etc)
<monochroma> the specific term i am familiar with for recording quality PC audio inputs is "Audio Interface"
<d1b2> <johnsel> and ofcourse you are an expert in this azonenberg, just pretend it's not hz but ghz and it's not a microphone but an antenna
<d1b2> <johnsel> half kidding, by the way
<d1b2> <johnsel> hmm the manufacturers I know either don't offer it or don't offer it in connectivity options that would be suitable. So other than what I already said about which type of mic would be suitable I can't help