Home Fab with BGA's?

[AaronD]

I've always shied away from BGA packages because "they're hard to hand-solder".  But with so many interesting parts (to me at least) being primarily or only available in BGA packages, I'm wondering how well-founded my fear really is.

I can't visually inspect under the chip (there are X-ray machines for that, but good luck getting one as a hobbyist!), so I pretty much have to trust that the joints are good, and find that out for sure when the project works or not, in addition to debugging the design itself.  (same problem with the thermal pad of a QFN, which I *have* done before)  But, given my success rate so far with paste and a dirt cheap toaster oven with my own reflow controller on it, I'm not too concerned about that...provided that the rest of the process is okay.

BGA's come "pre-balled", meaning that the exact right amount of solder is already on each pad of the chip, and no additional solder is required on the PCB.  Just align it on its footprint, within 1/4 pitch or so, and bake it.  Slather it with flux, maybe, as a sort of "tacky glue" as I carry it from the bench to the oven?

I've never had to replace a chip, except for when I put too much solder paste on before the reflow and the touch-up went south as well, so I'm not too concerned about that.  But if I do have to remove a BGA (with hot air), how hard is it for a hobbyist to "re-ball" it?  Wick it good, put a dot of paste on each pad, and put it back down again?  Or just forget it and buy spares?


And of course, we also have the board layout.  Any tips and tricks to get ALL of the pads out to the world?

I'm guessing to do the power first, with a via right next to each power pad and a bypass cap right next to that via on the back of the board, but beyond that, then what?  Just be creative?  Or is there more to it?
I imagine that the pin mapping that a lot of programmable chips have, is really going to shine here!  And multi-layer boards are good too.

[AaronD]

I might be wrong, but I thought BGAs do get a small amount of solder pasted added on the pads below the ball grid array.

It's only solder with no flux.  Solder paste is tiny particles of solder ("solder dust", if you will) suspended in flux.  But yes, BGA's come "pre-balled" from the factory, with the exact right amount of solder already on them.

[asmi]

I've soldered hundreds of BGAs at home (both in oven and with a hot air gun), and I'm yet to see a single soldering failure. As long as they have reasonable ball pitch (0.65 mm and up), they are the preferred package for me. And, like you said, for some parts there is simply no choice. See a project in my signature for example of the kind of designs I'm working with - it's very much on the simple end of the spectrum, but still should give an idea.

[AaronD]

I've soldered hundreds of BGAs at home (both in oven and with a hot air gun), and I'm yet to see a single soldering failure. As long as they have reasonable ball pitch (0.65 mm and up), they are the preferred package for me. And, like you said, for some parts there is simply no choice. See a project in my signature for example of the kind of designs I'm working with - it's very much on the simple end of the spectrum, but still should give an idea.

That's good to hear!  And from reading that thread, it looks like I've already been more difficult than that, with the Thermal-pad QFN.

I did get a solder failure on that - too much paste on the pad, bridged to another pin - and it was a pain to fix.  Hot-air'ed it off, cleaned it good, mixed up some extra-thin paste by adding more flux (a bit like mixing epoxy, except it doesn't harden before you're ready) and just painted it on without the stencil that no longer fit between the other parts.  Since I had soldermask between the pins, and I was using extra-thin paste, I just drew a continuous line across all the pins, and put a dot on the pad.  Hot-air'ed it back on, and it worked.  Took a few tries to get to that point though, and to figure out the trick.

[pdp11]

I have soldered in my oven some bga, without any failure. I have encountered a lot of failures on DFN/QFN.
I use a good qaulity stencil, with non rohs solder paste.
I inspect the solder paste with microscope, and if ok I manually place the bga at microscope.

[Psi]

You can do BGA at home, it's just a higher level of skill required compared to QFN/TQFP etc..
Totally doable after some practice.

The more balls the BGA has the harder it gets and the more your equipment needs to be up to the job.
And you might need a preheater for bigger ones etc..

There is a very big difference between soldering a ~30 pin BGA chip vs soldering a CPU or GPU with many 100's of balls.
A small BGA you can do with some skill any semi-decent hot air station.

BGA is a lot less forgiving of mistakes, you ideally need to get it right first time. If you have to pull the chip off and try again your
chances of damaging the chip or the pads goes up. And you have to spend time reballing it for each attempt.

Some BGA chip types have little metal plugs rather than pads and trying to clean them up for another attempt without a plug falling out can be challenging.
EDIT: Actually I think those metal plugged ones are kinda more considered VFQFN rather than BGA

[SMTech]

I've soldered hundreds of BGAs at home (both in oven and with a hot air gun), and I'm yet to see a single soldering failure. As long as they have reasonable ball pitch (0.65 mm and up), they are the preferred package for me. And, like you said, for some parts there is simply no choice. See a project in my signature for example of the kind of designs I'm working with - it's very much on the simple end of the spectrum, but still should give an idea.

That's good to hear!  And from reading that thread, it looks like I've already been more difficult than that, with the Thermal-pad QFN.

I did get a solder failure on that - too much paste on the pad, bridged to another pin - and it was a pain to fix.  Hot-air'ed it off, cleaned it good, mixed up some extra-thin paste by adding more flux (a bit like mixing epoxy, except it doesn't harden before you're ready) and just painted it on without the stencil that no longer fit between the other parts.  Since I had soldermask between the pins, and I was using extra-thin paste, I just drew a continuous line across all the pins, and put a dot on the pad.  Hot-air'ed it back on, and it worked.  Took a few tries to get to that point though, and to figure out the trick.

First things first - QFN/DFN really shouldn't be giving you issues.

In practice QFN is more forgiving of iffy placement than QFP, and BGA is/should be more forgiving again as each one offers less opportunity for bridging. As you say it's the inspection of the finished solder joint that presents the issue, BGA also is somewhat more prone to certain solder defects than other device types. This becomes more true with low volume poorly stored parts or poorly looked after solder paste & PCB finishes. An example of this would be "Head in Pillow", a fault you would only catch with X-ray. You can use cameras to try and peer underneath BGAs from the side, I have seen commercial offerings for doing so but I'd imagine in 2023, you could lash something up similar if you thought it would help.

On a production line BGA can be soldered with paste, or using a dip fluxer, where the machine presses the device into a film of flux on route to placement. I think dip fluxing is more commonly used for mounting PoP memory chips on top of processors where the alternative would be fitting your machine with height measurement and paste dispensing options (assuming they exist for your platform).

If you were very lucky, you might stumble on an affordable BGA rework station with alignment, that would let you tackle finer ball pitch or annoying BGAs where the balls are way underneath where you can't see them and even some reflow control- I'd imagine eMMC would not be much fun without some kind of assist.

[AaronD]

I've seen several people mention paste with BGA's.  Paste adds more solder.  Are the balls not enough?  I thought they were exactly sufficient, so that adding more risks bridging.

Of course, you need a way to keep it from sliding around between placement and reflow, and paste is slightly tacky...but so is plain flux.  (at least the stuff that I use is)  So I was thinking to just flux it good and stick it on.  Does that not work?

[asmi]

I've seen several people mention paste with BGA's.  Paste adds more solder.  Are the balls not enough?  I thought they were exactly sufficient, so that adding more risks bridging.

Of course, you need a way to keep it from sliding around between placement and reflow, and paste is slightly tacky...but so is plain flux.  (at least the stuff that I use is)  So I was thinking to just flux it good and stick it on.  Does that not work?

You can do it without paste (just a good amount of flux and clean pads), I personally only use paste when I'm assembling the whole board in an oven and so I'm pasting entire board via stencil. The key to soldering BGA is to see it align itself and drop down - that means all solder balls have been melted, all while trying to not locally overheat the chip - meaning you've got to constantly move your hot air gun across the device such that no portion of it would be significantly hotter than other areas - as this can crack the package and/or the chip. BGA chip will align itself if the placement error is less than half of it's pitch, so the smaller the pitch, the more tricky it is to place properly. To aid that, make sure your footprints for BGA parts have some sort of alighment marks - like an outline which is just outside of the package so that when you look from the top, you can see it "hugging" device. Also when designing a board for assembly with a hot air gun, make sure to leave some space around BGAs to help with soldering such that you wouldn't disturb nearby parts - it's easy (and tempting) to place BGAs very close to each other, but that will make soldering them with a hot air gun more risky and you will inevitably heat up corners of surrounding devices which again can lead to thermal cracks.

[SMTech]

I've seen several people mention paste with BGA's.  Paste adds more solder.  Are the balls not enough?  I thought they were exactly sufficient, so that adding more risks bridging.

Of course, you need a way to keep it from sliding around between placement and reflow, and paste is slightly tacky...but so is plain flux.  (at least the stuff that I use is)  So I was thinking to just flux it good and stick it on.  Does that not work?

It should work fine, and is probably the best approach when assembling by hand, just be sure yours is either inert post reflow or thoroughly cleaned off afterwards.
>>>>>>>>>>>

 BGA paste apertures are quite small - you can see the suggested dimensions in several datasheets - about 1/2 the ball diameter, It is not adding enough to create a bridge, just enough solder and flux to hold it in place and activate during reflow. Every (to all intents and purposes) assembly house has a printer & some form of paste inspection (at least in the printer), a dip fluxer is a dedicated extra piece of equipment to own and maintain on a production line so pasting is more common in a production line until its no longer the best solution. My example previously being one, fine pitch being another, accurately printing paste in 0.2mm circles or less is asking for issues to present themselves, needs an expensive (multi)stepped stencil and a type V paste.

[AaronD]

You can do it without paste (just a good amount of flux and clean pads), I personally only use paste when I'm assembling the whole board in an oven and so I'm pasting entire board via stencil.

It should work fine, and is probably the best approach when assembling by hand, just be sure yours is either inert post reflow or thoroughly cleaned off afterwards.

So if my stencil ignores BGA's, I should be okay?  With the trouble I've had with them not sitting flat on the board, and getting paste under them between pads (it might also be that my old paste doesn't flow well, so I have to work it excessively), it seems attractive to me to not bother with parts that don't really *need* it.

My flux seems to be okay to not clean.  It leaves a residue, but it doesn't interfere with operation.  If I were selling them, I'd want to get that off, but for just me, to put in a corner somewhere and do its job, it seems to be fine.

...It is not adding enough to create a bridge, just enough solder and flux to hold it in place and activate during reflow...
...pasting is more common in a production line until its no longer the best solution...

Sounds to me like a case of economic expediency becoming the de-facto solution that people take for granted, even when problems are shown with it.  The problems are studied and mitigated, still without really addressing the root cause that is taken for granted as "the" way to do it.

For a fab house making that decision on purpose, that's fine, but for a hobbyist that takes what they do as gospel and blindly copies it, not so much.  Different constraints, different solution.

Also when designing a board for assembly with a hot air gun, make sure to leave some space around BGAs to help with soldering such that you wouldn't disturb nearby parts - it's easy (and tempting) to place BGAs very close to each other, but that will make soldering them with a hot air gun more risky and you will inevitably heat up corners of surrounding devices which again can lead to thermal cracks.

Yes!  I do tend to pack my boards tightly, partly to keep the cost to a minimum, charged per area (even omitting mounting holes, which sometimes creates a challenge later), and partly because I lose my sense of scale and then get paranoid about signal integrity over what starts to feel like a long distance.  Absolute-minimizing the trace lengths gives me a sense that whatever mistakes I might have made don't matter.

But yes, such tight packing causes other problems, like not having space to label things on the silkscreen, and the impossibility to rework without destroying the whole thing.

[NorthGuy]

BGA are actually the easiest to deal with both in routing and in soldering. The only drawback is that you cannot probe inner balls.

If somene builds a board on the machine, it's certainly easier to put paste to the pads because the mechanism is already there. If you use a stencil manually, it is also easier. But if you want to add/replace BGA on the existing board, it's easier to use flux than to fit in a stencil.

When replacing I was able to do it without re-balling - just with solder applied to both the board and the chip. It worked well for me with 0.8 and 1.0 mm pitch (but I did it only few times). May be the re-balled solution would last longer, but my prototypes don't need to last long.

[asmi]

With the trouble I've had with them not sitting flat on the board, and getting paste under them between pads (it might also be that my old paste doesn't flow well, so I have to work it excessively), it seems attractive to me to not bother with parts that don't really *need* it.

Ever since I started using room temperature stable paste, I can't remember having any issues with a paste, as long as it's reasonably fresh, or "refreshed" by adding some fresh tacky flux.

My flux seems to be okay to not clean.  It leaves a residue, but it doesn't interfere with operation.  If I were selling them, I'd want to get that off, but for just me, to put in a corner somewhere and do its job, it seems to be fine.

I always clean flux residue if anything for aesthetics. I use a tacky water-soluble flux and wash assembled boards in the ultrasonic cleaner just because I like the look for clean boards, this is especially important when I use matte solder masks (which I like using because it looks awesome!). And for those boards which I build for a customers, they obviously need to be clean and look presentable.

Yes!  I do tend to pack my boards tightly, partly to keep the cost to a minimum, charged per area (even omitting mounting holes, which sometimes creates a challenge later), and partly because I lose my sense of scale and then get paranoid about signal integrity over what starts to feel like a long distance.  Absolute-minimizing the trace lengths gives me a sense that whatever mistakes I might have made don't matter.

Mounting holes is the first thing I ever place on a PCB design, because they are critical for my workflow - 1) most of my PCBs have components on both sides, so I have to solder it in two passes and holes are necessary so that I can make standoffs while doing the second pass, 2) my manual stencil printer requires mounting holes to hold PCB in place - again, because of double-sided assembly.
So except for a very simple prototype subcircuit PCBs, which I can assemble without stencil at all, or with a simple frameless stencil (my stencil printer requires framed stencils), I always have mounting holes. Though I assembled some of my very small PCBs a whole panel at once, and so mounting holes were on a tooling strips, which were then broken off because the area of individual PCBs was comparable to an area which would be taken up by mounting holes

[asmi]

BGA are actually the easiest to deal with both in routing and in soldering. The only drawback is that you cannot probe inner balls.

You can if you design in such capability.

If somene builds a board on the machine, it's certainly easier to put paste to the pads because the mechanism is already there. If you use a stencil manually, it is also easier. But if you want to add/replace BGA on the existing board, it's easier to use flux than to fit in a stencil.

Agreed - I don't bother removing paste masks from BGA footprints, but if I need to solder just a single BGA, I just use a crap ton of tacky flux - Louis Rossmann's style.

When replacing I was able to do it without re-balling - just with solder applied to both the board and the chip. It worked well for me with 0.8 and 1.0 mm pitch (but I did it only few times). May be the re-balled solution would last longer, but my prototypes don't need to last long.

First off, using solder paste risks having different amount of paste on different pads, and so you can have some opens when there is not enough solder on some pads compared to others.
Also, in my experience prototypes tend to be in use for much longer than initially anticipated, so I try not to skimp on them.

[AaronD]

Ever since I started using room temperature stable paste, I can't remember having any issues with a paste, as long as it's reasonably fresh, or "refreshed" by adding some fresh tacky flux.

...I use a tacky water-soluble flux and wash assembled boards in the ultrasonic cleaner...

Mounting holes is the first thing I ever place on a PCB design, because they are critical for my workflow - 1) most of my PCBs have components on both sides, so I have to solder it in two passes and holes are necessary so that I can make standoffs while doing the second pass, 2) my manual stencil printer requires mounting holes to hold PCB in place - again, because of double-sided assembly.
So except for a very simple prototype subcircuit PCBs, which I can assemble without stencil at all, or with a simple frameless stencil (my stencil printer requires framed stencils), I always have mounting holes. Though I assembled some of my very small PCBs a whole panel at once, and so mounting holes were on a tooling strips, which were then broken off because the area of individual PCBs was comparable to an area which would be taken up by mounting holes

I don't bother removing paste masks from BGA footprints, but if I need to solder just a single BGA, I just use a crap ton of tacky flux - Louis Rossmann's style.

Sounds like you've developed a good process.  Care to share the details?

I've done double-sided boards too, by doing the "lighter" component side first, and then just setting it upside down on the bench and then the sandwich tray that came with the toaster oven.  Sounds like you have a better set of tools than that.

[asmi]

Sounds like you've developed a good process.  Care to share the details?

Sure.

I use ZB3040H manual stencil printer for pasting PCBs. It uses 40x30 cm framed stencils, which is more than big enough for my PCBs and panels I work with. I typically order them at JLCPCB.

For reflow I have ZB2520HL oven. It seems to work pretty well with boards as large as 20x20 cm, It's specs says it can work with PCBs up to 25x20 cm, but 20x20 cm is the largest panel I've actually reflowed. You've got to make sure you DO NOT place PCBs right on a drawer because it will wick away a lot of heat off PCB and will lead to solder failures. That's another reason I need mounting holes, so that I can use small M3 screws for standoffs to lift a PCB off the drawer and reduce heat transfer to a drawer and provide enough clearance for components on the bottom side during reflow of the second side components.

I also have ATTEN ST-862D hot air gun (I think Louis Rossmann recommended it at some point, which is why I purchased it), and it's FANTASIC! It's got high enough power to quickly heat up parts for rework, This additional power is very helpful for 6+ layer boards with several ground/power planes as those planes can wick away quite a bit of heat, and so your hot air gun needs to be able to pump heat faster than it's wicking away so that solder around component(s) you are working on can actually reach melting point.

I also developed a small C# program which takes gerbers of a PCB and a BOM file and generates an assembly document, which shows which parts go where (see attached pdf as example of what it looks like). As my boards tend to contain a lot of 0402 parts which do not have any markings on packages, making sure I don't place wrong parts is extremely important, which is why I invested my time in that program. Its' sources are available here: https://github.com/asmi84/GerberTool but it's very ad-hoc program which was designed around my process, so if someone else wants to use it, he/she will probably have to adapt it to their process. I also use that file (in printed form) as a checklist for gathering parts before assembly, as well as marking parts already placed. Before I begin assembly, I always gather all components required into one box, and have another box for leftover components after placement. This also helps to keep tabs on what's placed and what isn't. This slows down assembly a bit, but it helps to ensure correct parts are placed in correct places.

I've done double-sided boards too, by doing the "lighter" component side first, and then just setting it upside down on the bench and then the sandwich tray that came with the toaster oven.  Sounds like you have a better set of tools than that.

Now I see why are you complaining that solder paste gets under the stencil - no wonder because boards with components on the underside is not going to lie flat
There is no need to "sandwitch" PCB for a second reflow, as surface tension of a molten solder will ensure parts stay in place. In my experience even large-ish parts like 2917 size tantalum caps stayed in place no problem.

[AaronD]

I use ZB3040H manual stencil printer for pasting PCBs. It uses 40x30 cm framed stencils, which is more than big enough for my PCBs and panels I work with. I typically order them at JLCPCB.

I might have to do some thinking on that.  Googling that one specifically, gets me about $430.  Surely the concept can be done for cheaper than that.  Maybe not as nice, but...

For reflow I have ZB2520HL oven. It seems to work pretty well with boards as large as 20x20 cm, it's specs says it can work with PCBs up to 25x20 cm, but 20x20 cm is the largest panel I've actually reflowed. You've got to make sure you DO NOT place PCBs right on a drawer because it will wick away a lot of heat off PCB and will lead to solder failures. That's another reason I need mounting holes, so that I can use small M3 screws for standoffs just lift a PCB off the drawer and reduce heat transfer to a drawer and provide enough clearance for components on the bottom side during reflow of the second side components.

Oh, you have a REAL reflow oven!  I got the cheapest toaster oven that I could find, on purpose: thin sheet steel all around, no insulation at all, the thermostat is a bi-metal switch outside the chamber, etc.  I got the cheap one on purpose, so it would have all-mechanical controls that I could easily set out of the way.  No smarts that can't be removed easily and would get horribly confused.

I set the thermostat all the way up and the timer to never shut off, and plugged it into a software-controlled triac outlet.  That software takes feedback from a K-type thermocouple inside the chamber, to run a configurable profile made from ramps and holds, and then opens the door with an R/C hobby servo to cool faster.  (my own design)

I've done double-sided boards too, by doing the "lighter" component side first, and then just setting it upside down on the bench and then the sandwich tray that came with the toaster oven.  Sounds like you have a better set of tools than that.

Now I see why are you complaining that solder paste gets under the stencil - no wonder because boards with components on the underside is not going to lie flat
There is no need to "sandwitch" PCB for a second reflow, as surface tension of a molten solder will ensure parts stay in place. In my experience even large-ish parts like 2917 size tantalum caps stayed in place no problem.

The sandwich tray is a slightly corrugated metal thing that came with the toaster oven.  You're supposed to put literal sandwiches on it.  It also came with a wire rack that is supposed to be for other foods, or maybe in combination with the tray to catch drips or something like that.  Anyway, it's dirt cheap, designed for food, and I've replaced the controls to use it for PCB's.

For the second side, I just put the first side face-down on the bench to place the parts, and then on the tray again to reflow.  If it doesn't sit flat enough, I'll put a ring of blank PCB's under the edges.  Haven't had a failure yet, except for a few tombstones and the QFN that got too much paste on the thermal pad.  I often feel the need to clean up after the stencil though, so I probably do need something better than just taping it down to the bench.

I also have ATTEN SS-862D hot air gun (I think Louis Rossmann recommended it at some point, which is why I purchased it), and it's FANTASIC! It's got high enough power to quickly heat up parts for rework, This additional power is very helpful for 6+ layer boards with several ground/power planes as those planes can wick away quite a bit of heat, and so your hot air gun needs to be able to pump heat faster than it's wicking away so that solder around component(s) you are working on can actually reach melting point.

I have a cheap combination iron and hot air station, with a thermostat for each.  Seems to be bulletproof though, and definitely well used.
(For more than just solder too: I left the hot air on a low setting for several hours trying to straighten a small rubber hose.  Small nozzle stuck into the hose, and the handle supported with string.  It...sorta worked.  The air gun was fine, but the hose barely responded.)

I also developed a small C# program which takes gerbers of a PCB and a BOM file and generates an assembly document, which shows which parts go where (see attached pdf as example of what it looks like). As my boards tend to contain a lot of 0402 parts which do not have any markings on packages, making sure I don't place wrong parts is extremely important, which is why I invested my time in that program. Its' sources are available here: https://github.com/asmi84/GerberTool but it's very ad-hoc program which was designed around my process, so if someone else wants to use it, he/she will probably have to adapt it to their process. I also use that file (in printed form) as a checklist for gathering parts before assembly, as well as marking parts already placed. Before I begin assembly, I always gather all components required into one box, and have another box for leftover components after placement. This also helps to keep tabs on what's placed and what isn't. This slows down assembly a bit, but it helps to ensure correct parts are placed in correct places.

My bench has an LCD monitor on the back wall, driven by a Raspberry Pi.  Its USB and GPIO are available on the bench, via a long-tailed hub and a 40-pin ribbon, and WiFi to my main network so I can put the KiCAD project on it while I place the parts.  Seems to work well.  It's also one of several machines that can run my USB oscilloscope.  I specifically got a monitor that had a DC power input, so I could tap off of that to run a DC-DC converter to the Pi.  One plug-in AC switch covers both...and I'm back to the "It's now safe to turn off your computer" mentality, for those that remember that.

It also helps to use as much of the same part as possible.  (fewer lines in the BOM, with correspondingly larger Qty for each)  Analog designs need a zoo of different values, but even then there's room to consolidate.  Two "standard value" resistors in parallel, for example, instead of a single "half value", so I don't have to keep straight that "that one" is different.  Digital designs, in my experience, are even more forgiving.  If power is a non-issue, then a pull-up resistor can have several orders of magnitude tolerance.

[AaronD]

Ever since I started using room temperature stable paste, I can't remember having any issues with a paste, as long as it's reasonably fresh, or "refreshed" by adding some fresh tacky flux.

...I use a tacky water-soluble flux and wash assembled boards in the ultrasonic cleaner...

What paste and flux do you use?  That might be the easiest thing to change.  What I have now, lives in a dedicated mini-fridge and only comes out when I build a board.  The flux seems to work well enough, but the paste is still pretty dry.

It probably doesn't help that the smallest size I could buy at the time has easily lasted me several years, so it's nowhere near fresh at all.  And still over half left.

[asmi]

I have a cheap combination iron and hot air station, with a thermostat for each.  Seems to be bulletproof though, and definitely well used.
(For more than just solder too: I left the hot air on a low setting for several hours trying to straighten a small rubber hose.  Small nozzle stuck into the hose, and the handle supported with string.  It...sorta worked.  The air gun was fine, but the hose barely responded.)

That's what I used in the past as well, but once I upgraded to ADS200 soldering station and ST-862D hot air gun, I will not touch that cheap crap ever again! The difference is a day and night. Cheap stuff is great for beginning (as it reduces your upfront investment before you commit more significant amount of money), but once things get even somewhat serious, investing into quality tools is totally worth it.

My bench has an LCD monitor on the back wall, driven by a Raspberry Pi.  Its USB and GPIO are available on the bench, via a long-tailed hub and a 40-pin ribbon, and WiFi to my main network so I can put the KiCAD project on it while I place the parts.  Seems to work well.  It's also one of several machines that can run my USB oscilloscope.  I specifically got a monitor that had a DC power input, so I could tap off of that to run a DC-DC converter to the Pi.  One plug-in AC switch covers both...and I'm back to the "It's now safe to turn off your computer" mentality, for those that remember that.

I tried using a tablet instead of paper, but I still prefer paper. I still do use tablet for larger/more complex boards because the image on that page is just too small to see details.
Also I do have a real oscilloscope (none of that USB crap), and a real stereo microscope (again, none of that cheap garbage). Like I said above, once you get seriously into assembly, investing into quality tools will pay off.

It also helps to use as much of the same part as possible.  (fewer lines in the BOM, with correspondingly larger Qty for each)  Analog designs need a zoo of different values, but even then there's room to consolidate.  Two "standard value" resistors in parallel, for example, instead of a single "half value", so I don't have to keep straight that "that one" is different.  Digital designs, in my experience, are even more forgiving.  If power is a non-issue, then a pull-up resistor can have several orders of magnitude tolerance.

It's not always possible due to other requirements (like for RF - more parts - more noise, or for high-speed lines, when each part adds impedance discontinuities), but yeah I do try my best to reduce number of BOM lines as much as practical.

[asmi]

What paste and flux do you use?  That might be the easiest thing to change.  What I have now, lives in a dedicated mini-fridge and only comes out when I build a board.  The flux seems to work well enough, but the paste is still pretty dry.

It probably doesn't help that the smallest size I could buy at the time has easily lasted me several years, so it's nowhere near fresh at all.  And still over half left.

This is the paste I use: https://www.chipquik.com/store/product_info.php?products_id=470005 It's a SAC paste, but they also have a leaded version if you don't mind that.
The flux I use is this one: https://www.chipquik.com/store/product_info.php?products_id=310004 Works great, and it's quite easy to clean. I do have a rosin-type flux for those cases when water-soluble flux is not up to the task (like working with really old joints with lots of oxidation), but these are very rare occasions. I wash my boards in an ultrasonic cleaner, which I got on Amazon for I think a couple hundred bucks, it also works great for washing jewelry, old kitchenware and other similar stuff which is hard to properly clean by conventional means.

[AaronD]

...The difference is a day and night. Cheap stuff is great for beginning (as it reduces your upfront investment before you commit more significant amount of money), but once things get even somewhat serious, investing into quality tools is totally worth it.

Hey, it's a lot better than what I started with!  A no-switch, no-thermostat pencil with a non-locking fold-down wire 'M' stand (you can guess how that went!), and the tip immediately tarnished...

Also I do have a real oscilloscope (none of that USB crap), and a real stereo microscope (again, none of that cheap garbage). Like I said above, once you get seriously into assembly, investing into quality tools will pay off.

The USB 'scope was a recent thing.  As a freshman in college, I bought an Iwatsu 4-channel pure-analog 'scope (not even digitally controlled) from eBay, and that was my daily driver for many years.  I still use it in addition to the USB one.  So I know what an oscilloscope really does and how to adjust it manually, and I don't rely on the "auto-set" button.  (I think I found it once in the USB 'scope app, but I've forgotten where it is now; I don't use that function at all.)

It's *very* handy to take a laptop and a USB "hockey puck" to the garage or wherever, to see how something is misbehaving, and take a screenshot to post on the relevant forum, instead of the "big iron" analog 'scope and a camera (which I've also done), and keep track of what the settings were in my report.  Yes, a dedicated digital 'scope would do that too, but I already had the laptop...

It's not always possible due to other requirements (like for RF - more parts - more noise, or for high-speed lines, when each part adds impedance discontinuities), but yeah I do try my best to reduce number of BOM lines as much as practical.

I don't do much RF...yet?  So far, all of my high-speed stuff (relatively speaking) has been simple enough to just follow the rules and watch it work.

Once I get a custom audio DSP going, I want to start another project to process HDMI video.  I started the audio project with a Pi Pico that felt infinite at first, coming from the 8-bit world at 20MHz or less, but it barely kept up with a single signal generator at 48kHz.  So I got a Teensy 4.1 instead, and eventually want to spin my own board with that chip on it.  I'm thinking to do the video one on an FPGA of some kind, having no experience at all with those.  (*that's* going to be fun!)

[AaronD]

This is the paste I use: https://www.chipquik.com/store/product_info.php?products_id=470005 It's a SAC paste, but they also have a leaded version if you don't mind that.
The flux I use is this one: https://www.chipquik.com/store/product_info.php?products_id=310004 Works great, and it's quite easy to clean. I do have a rosin-type flux for those cases when water-soluble flux is not up to the task (like working with really old joints with lots of oxidation), but these are very rare occasions.

Great!  Thank you!  Looks plenty cheap enough to just give it a shot.

I wash my boards in an ultrasonic cleaner, which I got on Amazon for I think a couple hundred bucks, it also works great for washing jewelry, old kitchenware and other similar stuff which is hard to properly clean by conventional means.

I've always been leery about combining food and solder.  I might not ever put food in that fridge again, even though the syringe of solder paste is in a sealed bag.  I think I'd feel the same way about an ultrasonic cleaner.

[asmi]

Hey, it's a lot better than what I started with!  A no-switch, no-thermostat pencil with a non-locking fold-down wire 'M' stand (you can guess how that went!), and the tip immediately tarnished...

Yeah, I started with crap tools as well, but that was a while ago. So I guess with time you'll end up with tooling kind of like what I have. I didn't buy it all at one moment, but rather was slowly upgrading equipment one thing at a time.

The USB 'scope was a recent thing.  As a freshman in college, I bought an Iwatsu 4-channel pure-analog 'scope (not even digitally controlled) from eBay, and that was my daily driver for many years.  I still use it in addition to the USB one.  So I know what an oscilloscope really does and how to adjust it manually, and I don't rely on the "auto-set" button.  (I think I found it once in the USB 'scope app, but I've forgotten where it is now; I don't use that function at all.)

It's *very* handy to take a laptop and a USB "hockey puck" to the garage or wherever, to see how something is misbehaving, and take a screenshot to post on the relevant forum, instead of the "big iron" analog 'scope and a camera (which I've also done), and keep track of what the settings were in my report.  Yes, a dedicated digital 'scope would do that too, but I already had the laptop...

The reason I prefer "real" scopes is to do with muscle memory, once you use your scope enough you will be able to control it just by feel, without looking at controls, which sometimes is super-helpful. This is why I kind of have a problem with most new scopes coming out with touch interfaces, as it's only a matter of time until non-touch controls are deleted entirely.

I don't do much RF...yet?  So far, all of my high-speed stuff (relatively speaking) has been simple enough to just follow the rules and watch it work.

Things get more complicated as frequencies rise and rise time of signals fall. For example, for serial termination two 10 Ohm resistors will perform significantly worse than a single 20 Ohm one.

Once I get a custom audio DSP going, I want to start another project to process HDMI video.  I started the audio project with a Pi Pico that felt infinite at first, coming from the 8-bit world at 20MHz or less, but it barely kept up with a single signal generator at 48kHz.  So I got a Teensy 4.1 instead, and eventually want to spin my own board with that chip on it.  I'm thinking to do the video one on an FPGA of some kind, having no experience at all with those.  (*that's* going to be fun!)

FPGAs is where I'm currently at, it's the Holy Grail of digital design, as close as you can get to designing your own ICs without spending absurd money on it.

[asmi]

Great!  Thank you!  Looks plenty cheap enough to just give it a shot.

I think they have it also in syringes if you want smaller quantity. I prefer a jar because you can put excess paste back in it after you take it off stencil - try doing that with a syringe!

I've always been leery about combining food and solder.  I might not ever put food in that fridge again, even though the syringe of solder paste is in a sealed bag.  I think I'd feel the same way about an ultrasonic cleaner.

Fine - I allow you to use your (future) ultrasonic cleaner only for non-food items I don't really care - there's solder all over my home, so as long as one doesn't use solder paste instead of butter for morning toasts, he/she going to be just fine.

[AaronD]

...I guess with time you'll end up with tooling kind of like what I have. I didn't buy it all at one moment, but rather was slowly upgrading equipment one thing at a time.

Yeah, probably so.

The reason I prefer "real" scopes is to do with muscle memory, once you use your scope enough you will be able to control it just by feel, without looking at controls, which sometimes is super-helpful. This is why I kind of have a problem with most new scopes coming out with touch interfaces, as it's only a matter of time until non-touch controls are deleted entirely.

Same problem with live media control: touch screens technically work, and I use one weekly (I even wrote the app that runs on it, in C++ with wxWidgets on a Raspberry Pi, to control some cameras and a digital audio console), but physical faders, buttons, and knobs are still much easier to use, and to find quickly and know where they're set without looking.

I wonder if someone will invent a "tactile screen" that follows a similar concept as a visual one?  There are braille "displays" already, but all I've seen of those were low-resolution and not designed to be visual at all.  Might have been comparable to a 1x4 character display, or even a single-character display.  Think of the possibilities (and confusion for those that didn't grow up with it, and get to use the first few generations of commercial products) of having a controllable localized texture on top of a visual display that also accepts input!
I've seen it approximated with the vibrator in a phone, but it wasn't convincing.  When it can create a grabbable knob, *then* I'll be impressed!

Things get more complicated as frequencies rise and rise time of signals fall. For example, for serial termination two 10 Ohm resistors will perform significantly worse than a single 20 Ohm one.

I'll probably get there, and wonder why it doesn't work, like I did when I used similar resistor values for a VGA processor as I was used to from 15-0-15V (pro line level) audio!  (Why is the picture smeared to the right???  Answer: those high resistor values combined with parasitic capacitance...)  Dropped them by an order of magnitude (lots of hot-air rework), and it was good!

FPGAs is where I'm currently at, it's the Holy Grail of digital desing, as close as you can get to designing your own ICs without spending absurd money on it.

I kinda got that feeling as well.  Different mindset though: more combinational logic with an *optional* clock, instead of sequential with a required clock.  Of course, you can *make* a sequential thing in there, and you can even make a CPU if you like, that does exactly what you want (I did that once, in Ladder Logic on an industrial PLC, and then wrote an assembler for it in C#), but it doesn't HAVE to be that.

When I got the USB 'scope, they were running a special on it.  $200 off if you also buy something on <this list>.  One of the things on that list was a Cora Z7 dev board for $100.  (so they effectively paid me $100 to take that board, as part of the special)  Figured that was the thing I was most likely to use soon, despite not being ready for any of it at the time.  Still not ready yet - still a long way to go on the audio project that I want to finish first - but while we're here, would you happen to have some advice to get started with that and eventually get to an inline HDMI processor?  Probably need to upgrade parts along the way too...

I think they have it also in syringes if you want smaller quantity. I prefer a jar because you can put excess paste back in it after you take it off stencil - try doing that with a syringe!

Actually I do.  It's dry enough and flows so poorly that I just take the needle off and dig it out.    And I do put the excess back in before putting the needle back on like a cap.
Maybe what I called "extra thin" because I mixed some more flux into it, is more like what it should have been to start with...

...as long as one doesn't use solder paste instead of butter for morning toasts...