ioc.exchange is one of the many independent Mastodon servers you can use to participate in the fediverse.
INDICATORS OF COMPROMISE (IOC) InfoSec Community within the Fediverse. Newbies, experts, gurus - Everyone is Welcome! Instance is supposed to be fast and secure.

Administered by:

Server stats:

1.4K
active users

Andrew Zonenberg

Back side of the line card is pasted! Used my new... 8 8 inch i think? stainless steel drywall knife with great results.

Also running low on solder paste, I only have what's in the syringe on the bench now. Deep storage in the fridge is empty.

Minor print defect in the very corner, probably insufficient paste at the edge of the array.

Tried to tin the pads with an iron but the board had too much thermal mass. Couldn't use preheat or hot air because board was already pasted.

I did the best I could, it'll probably bridge but I can fix it post reflow.

Time for a stretch break. This is gonna take a while.

At some point I'm definitely going to have to start thinking about PnP-ing some of these bigger boards. My designs are already beyond the point most people would have given up on hand assembly.

This one has a touch over 600 components on it. Given the time I'm probably going to try and finish the back side but not do the front tonight.

Coming along nicely.

All the 0402 0.47 uF (must be several hundred) are done, plus the 32 4.7 uF 0603s.

Still have to do one ferrite bead, a handful of assorted pullups/downs and voltage regulator feedback dividers, and the port status LED series resistors.

Done and out of the oven.

Two bridges on adjacent parallel caps, but the shorted pins were the same net so I'm not going to bother fixing.

Just gonna count out some of the parts for the front side then time for bed, I'll stuff tomorrow.

That looks like most of the non MSL stuff.

@azonenberg
My point of view is all those hand assembly are an invaluable waste of time (which could be spent with family...).
I do all PCB+Assembly since years with Fab doing PCB+Assembly.
About PnP it seems also not interesting to spend time configuring PnP/Stocking reel of components when Fab can do all with better quality and for less money/time.

@bvernoux There's no family awake. I'm a night owl and they go to bed early, so it's not like I'd be hanging out with them this time of night anyway.

My experience is that commercial assembly is a massive cost increase especially for one-off prototypes. Using parts I have on the shelf dramatically reduces both the cost and turn time for prototyping, small BOM changes, etc.

And provides a hedge against supply chain issues - if I design a board around parts I have on the shelf I *know* I can build it.

@bvernoux Also maybe it's my neurodivergent nature but I find stuffing a ton of 0402 or something to be almost meditative.

@azonenberg If that help you to relax and have fun why not
Anyway the process is interesting to see step by step especially the end result.

@bvernoux Which fab do you use? My experience is that it's a real hassle to get a quote, and then to have to provide parts that are hard for the vendor to get with a bigger hassle of dealing with customs coming/going with parts that may have encryption on them.

@azonenberg I want a PnP machine, mainly so I can stare at it while it works

@jpm I'm never going to have one here, no space.

I have friends with a shop across town that have, I think, a LumenPNP. It's a WIP, they dont have it working reliably yet. May bring some of my designs there for hybrid placement (e.g. robot the 0402 passives and do the rest by hand since it's less work than setting up feeders for a single part)

@azonenberg @jpm I think the feeder setup time is a big issue for one-off prototypes. If you have some odd component like a 9k1 resistor on the board once, placing it by hand is faster than setting up a feeder. At the cost of even the Lumen’s feeders, having a couple decades of E24 already on feeders is very expensive.

@azonenberg @jpm
I think for a prototyping setup, automatically placing standard passives and hand placing the rest is probably fastest. JLC’s cheap assembly process works similarly. They have a line of machines with E24 resistors, standard caps and a few jellybean parts already on feeders, and you only pay per feeder for everything else. I have used their service to just place passives before, on a board that I was planning to hand solder anyway.

@jaseg @jpm Yep I'm honestly thinking of just machine placing 0.47 uF, 4.7 uF, 33R, 470R, 1K, 4.7K, 10K. By volume that's probably 80% or more of the parts on one of my typical designs.

@azonenberg @jaseg yep this would be my approach if I had one. I’m considering building one of those small open source ones, mostly to watch it move around.

@jpm @jaseg My friend has her lumen working OK-ish on paper tape but is apparently having problems getting plastic tape to feed reliably without yeeting parts all over the place.

She's been too busy to actively work on it for a while though. Might be a simple fix to tension somewhere etc.

@azonenberg at some point the solder paste dries out by the time you place the last parts

@azonenberg do you want to try some gc10? I bought a huge tube a month ago and there's no way I'm going to use all of it before it expires.

@kevswims Maybe another time. I have my process pretty well dialed in using TS391 for now.

@azonenberg I want to figure out an A/B test between some of these. The first board I did with the GC10 just worked but I'm not sure if that was due to luck with my paste application or if it was actually that much better than the ChipQuick stuff that I have been using.

@kevswims Yeah. I definitely want to optimize my process more, but I don't want to change too many variables at once. I'm already playing with electropolished vs standard stencils, and using the larger putty knife to eliminate multi-pass applications on larger boards (this was i think my single biggest issue with bad prints... the stencil was lifting up slightly due to board flex then going back down again and smearing the already-printed deposits when i did the adjacent pass)

@azonenberg I'll have to find some old rev boards and run a few tests with better isolated variables.

I have 4 or 5 different designs that I'm going to be building in the next month or two so I should get some better data about process reliability out of those.