azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing |,, | Logs:
maartenBE has quit [Ping timeout: 258 seconds]
maartenBE has joined #scopehal
Degi has quit [Ping timeout: 240 seconds]
Degi has joined #scopehal
electronic_eel has quit [Ping timeout: 240 seconds]
electronic_eel has joined #scopehal
m4ssi has joined #scopehal
sorear has quit [Ping timeout: 264 seconds]
sorear has joined #scopehal
<azonenberg> lain: ping
<azonenberg> So i'm working on fixing up the jitter spectrum filter
<azonenberg> The current problem with it is that it just FFTs the TIE values (adding filler if there's not a toggle in the waveform)
<azonenberg> But it assumes a uniform unit interval
<azonenberg> If you have a spread spectrum clock, your Pj is gonna be all wrong
<azonenberg> So i need to resample the jitter values onto a common timebase
<azonenberg> so now here's the question... what should that timebase be?
m4ssi has quit [Remote host closed the connection]
<azonenberg> My input waveform is sampled at 200-to-201 ps intervals, in this example
<azonenberg> i feel like resampling it to a 200ps period is potentially problematic
<azonenberg> but if i resample to say 100 or 50, what do i do with the gaps? linear interpolation that far would be iffy
maartenBE has quit [Ping timeout: 272 seconds]
maartenBE has joined #scopehal
m4ssi has joined #scopehal
smkz has quit [Quit: smkz]
smkz has joined #scopehal
m4ssi has quit [Remote host closed the connection]
<azonenberg> First attempt at silicone molding completed. It's definitely on the soft side (I went with the lowest viscosity it could and it only cures to a Shore A8 hardness) but that might actually be better for probing
<azonenberg> because i feel like the soft surface gives it a more "grippy" feel
<azonenberg> less likely to slide around
<azonenberg> This was just a solid puck cast in a plastic cup. The actual 3d printed mold prototype is out for delivery now
<azonenberg> I have some clay for sealing around the connector coming tomorrow. Boards are at fab now but expected to ship any day
<azonenberg> Mold came in. The SMA doesn't fit
<azonenberg> So i need to respin the mold
<azonenberg> Doing a test cast of silicone around an old-rev PCB to see how well that works
<azonenberg> (unpopulated)
<_whitenotifier> [scopehal-apps] rob0tsunny forked the repository -
<d1b2> <DuncanHaldane> I've done quite a bit of silicone molding. Give me a shout if you hit a snag and want help.
<azonenberg> DuncanHaldane: Right now i have a first prototype curing. I ran into two snags
<azonenberg> First is that I mismeasured the SMA connector and/or screwed up the CAD work. So it doesn't actually fit in the mold :p
<azonenberg> i'm casting the bare PCB now as a proof of concept but need to 3d print a new mold
<azonenberg> (after i figure out exactly what dimensions on the sma are wrong)
<azonenberg> Second is that I'm literally injection molding the silicone
<azonenberg> as in, i draw the mixed silicone (low viscosity, around 2400 cP) up into a syringe
<azonenberg> and squirt into the mold with a needle
<azonenberg> through the sprues
<azonenberg> There's an air leak somewhere around the luer lock fitting, causing bubbles to get into my freshly vacuum degassed silicone
<azonenberg> it's aerated as i draw it up
<d1b2> <DuncanHaldane> Yeah that'll get some bubbles in there. What silicone product are you using? Also does it need to be silicone, or would a polyurethane also work?
<azonenberg> Right now I'm using Magikmold P-508
<azonenberg> Thinking of switching to P-525 which is quite a bit stiffer when cured and only slightly more viscous in the liquid state
<azonenberg> My first attempt used 3D printed SLS TPU
<azonenberg> printed as a standalone shell from a powder bed, then slid over the PCB
<azonenberg> Surface finish wasn't as smooth as i would have liked, and it was hard to slid the board in. plus the SLS process required fairly thick walls so it was bulky
<azonenberg> I'm now experimenting with casting the rubber around the PCB directly rather than fabricating separately and sliding it in
<d1b2> <DuncanHaldane> If you have compressed air, a pressure pot can help a lot with the bubbles. Like this one:
<azonenberg> This is the previous version
<d1b2> <DuncanHaldane> The white bit is the molded component?
<azonenberg> I do not currently have compressed air. I have an oil-free diaphragm vacuum pump and a fairly small (2 quart) stainless steel chamber. Eyeing a 3-gallon one as an upgade in the nearish future
<azonenberg> The white bit was the 3D printed version
<azonenberg> I'm aiming for a similar shape but much less bulky around the pcb
<azonenberg> I'll have pics of the latest prototype in ~3 hours when it's done curing
<d1b2> <DuncanHaldane> Nice. Looks pretty slick!
<azonenberg> That version also just has two contacts at the tip you have to solder wires from to your test points
<azonenberg> The new version goes back to what i had in previous prototypes (removing them was a bad idea) and has castellations
<azonenberg> so you can solder the probe directly to signal/ground contacts at 1mm spacing
<azonenberg> or add extensions if needed for other pitches
<azonenberg> That board is at fab now and is a few days out
<azonenberg> (v0.4, the photo is v0.3)
<azonenberg> right now i have a scrap v0.3 board in my mold, no components, just to test the silicone casting
<d1b2> <DuncanHaldane> That is some challenging geometry to mold... What do you sprues look like?
<azonenberg> the four corner pins are alignment pegs for registering the two mold halves
<azonenberg> the cylindrical pins in the cavity pinch the PCB to hold it centered in the cavity
<azonenberg> the strain relief notches are slightly less deep
<azonenberg> and this is the top half showing larger holes (sprues) at each end plus smaller vent holes in the middle
<azonenberg> and again the four corner holes are alignment pins, not sprues
<d1b2> <DuncanHaldane> Gotcha - two sprues one at either end.
<azonenberg> There's a reason i went with ultra low viscosity silicone
<d1b2> <DuncanHaldane> Yep 🙂
<d1b2> <DuncanHaldane> The relatively short working time of that RTV silicone limits some options
<azonenberg> Anyway, this is very much still in the R&D phase so i'm open to alternatives/ways to improve
<azonenberg> I'm also considering switching the AKL-PT1 (handheld probe) from a 3D printed slip-on enclosure to a cast epoxy unibody
<azonenberg> this would make it a lot less bulky
<d1b2> <DuncanHaldane> Let's see... If I was back in my robot lab with all the goodies, I would try doing a vacuum transfer molding with an RTV polyurethane with a long working time.
<azonenberg> But the current AKL-PT1 uses an air dielectric coplanar waveguide, so the big engineering challenge there would be how to maintain the air cavity going through the entire axis of the mold
<azonenberg> while potting the surroundings
<d1b2> <DuncanHaldane> So mix (carefully without adding bubbles) and degas the polymer. Pull a vacuum on one end of the mold (which was modified to be air-tight), connect the other end of the mold to my tub of polymer through a tube. Use vacuum pressure to fill the mold with polymer. Throw the whole thing in the pressure pot while it cures, making sure there's a reservoir of polymer to fill voids).
<azonenberg> I think i'm going to explore the current injection molding strategy a bit more
<azonenberg> it seems promising
<azonenberg> i just need to not get air into the syringe, or remove it before injection if that's not possible
<azonenberg> a vacuum chamber large enough to fit the entire syringe would easily let me degas it
<d1b2> <DuncanHaldane> Yeah, that degassing would help. Curious to see how the current molding turns out!
<azonenberg> Yeah i expect some bubbles. and i need to seal around the resistors
<azonenberg> the "pinch" seal around them was not tight enough and that cavity filled with silicone, which it wasn't supposed to
<azonenberg> so i think i'm just gonna cover them with sulfur-free clay
<azonenberg> then peel the clay off once the silicone cures
<d1b2> <DuncanHaldane> Yeah, that's a good plan. Getting good seals is hard without making design tradeoffs
<d1b2> <DuncanHaldane> Can I ask what pressure your vacuum pump can draw down to?
<azonenberg> It's a two stage oil free diaphragm pump. If my uncalibrated gauge is to be believed, somewhere just past -29 inHg
<d1b2> <DuncanHaldane> OK, great.
<azonenberg> although that takes a while, 28ish is more rapidly achievable
<azonenberg> the last bit is super slow
<d1b2> <DuncanHaldane> In the past, I've tried to minimize the working volume of the chamber to save the pump some work.
<azonenberg> Yeah
<azonenberg> So if i get a bigger chamber i'll keep the small one for when i dont need a big area
<d1b2> <DuncanHaldane> Yep! You can also just fill up the big one with chunks of stuff to displace air volume if you need to fit something long and skinny for example
<azonenberg> yeah that would work too, as long as it wasn't something that would outgas significantly
<azonenberg> although i doubt that's a major problem at low vac like this
<azonenberg> compared to turbopumped UHV chambers i'm used to working with
<d1b2> <DuncanHaldane> UHV for something like a sputter-coater? Not as bad as that, but getting close does help with bubbles
<d1b2> <DuncanHaldane> I just had big blocks of aluminum and bronze sitting around
<azonenberg> I've done a fair bit of work with SEMs and FIBs
<azonenberg> pre-covid i spent a while hanging out in the lab watching our sem tech baking out the columns to get off any residual gas molecules before firing up the ion pumps
<d1b2> <DuncanHaldane> Oh, very cool.