What SMD equipment do I need?

As a wearer of spectacles since I was 5 years old (many decades ago) I note that opticians not only sell expensive cleaning fluid but also expensive spectacles and lenses that seem not to differ from cheaper ones except for the price charged

Me cynical ?
Yes !

3 Likes

Some of the special coatings and high-index materials are pretty cool, but I'm not sure they are worth the added cost.

I've heard that one feels especially dumb when you lose an expensive pair in the surf on day one of a remote vacation.

Luckily for me I am never likely to lose anything in the surf on day one of a remote vacation :grinning:

2 Likes

Or a set of orthodontic retainers.

Don't ask.

2 Likes

Indeed.  I was surprised at the amount of solder remaining when viewing through a 20Xlens.

I already had some but two pair of tweezers were included with the hot air gun (858D) mentioned by @jremington.

Well, here's what I achieved with my minimalist approach.
 
It took a bit of fussing but I was finally able to get a driver mounted – I had ordered five to allow for losses and two died in the effort. I believe it was a Celsius vs. Fahrenheit thing.  The protoboard didn’t work out because the edges of the solderable holes were too far apart for the contact pads on the chip.  I ended up using a piece of strip board, and that was barely small enough.  I also mounted a number of terminal blocks to allow different configurations as shown in the datasheet.   For reference, the holes are on .1" centers.

1 Like

Is that a 4.7uf capacitor to the left with a resistor and mosfet on the right?

I use low melt solder-paste, because I was soldering SMD LED's that melt easy.

I also try to set and keep the heat just long enough to melt the paste.

You'll get better with practice, trust me. :wink:

Yes, 4.7uF, as shown in the datasheet example circuits. The other object is a flyback diode for the motor.

I have all four hobbyist affordable soldering methods. I have a weller about as old as I am as an iron, i have a crapola chinese reflow oven (T962 that, despite mfg claims otherwise, can't do SAC305 (though I found a good solder that is lead free, tests lead free, and works extremely well. Only one company is making it, at present. Is SnBiCu, but VERY heavy on the tin - miles from the eutectic, and low enough bismuth that it doesn't form that godawful 95C melting phase if lead gets in I have three hot air guns, one of which is fit for soldering, and I have a $70 chinese hotplate (946C)

You really do need a breakout board to get good results imo (I might be able to help with that - https://www.tindie.com/products/drazzy/qfndfn-breakouts-from-6-to-32-pins-4-12pcs/ ).
Ideally you have a stencil and stencil the paste on heat it on hotplate or reflow oven depending on solder and components (LED beads on aluminum core PCB can only be done effectively on hotplate because the heat required to heat up the board to melt the solderhas to come from below - if it comes from above, it will melt the plastic body of the LEDs), while many other things only work well with reflow. iron is for rework and one-offs, whilw the hotplate gets used to strip boards and assemble boards with SAC305 solder,

I have found hot air to be not the most useful on it's own, but exceptionally useful with a hotplate run slightly cool to avoid discoloring solder mask, or to give an iron some extra kick

Oh and CML supply sells little bottles of kester soldering flux, they will last you years cost like 8 bucks, and the two with rosin in them (186 - a bit more aggressive, but stinkier) and 1544 (nearly as aggressive, doesn't reek). Blows every other flux I've used out of the water, and is way easier to clean off than gel flux.

As for solder paste I'd just use tin lead from china - Relife63:37, 2uul189 (its actually 182C silverbearing leaded) either of KEK's leaded solders... Avoid conduction, PPD, and Mechanic brands of paste. Mechanic's products are never above average and often below average, conductions solder paste is the worst I ever used, and PPD's only 183C solder is "S600 lead free", has shitty properties - but it still lights up the lead test swab bright red!

1 Like

Thanks very much for the update!  When the new boards go up on your site will drawings also be available?  I'm confident I can solder one on - given my success at doing so with my hacked up stripboard (which has an intermittent loose connection) - but it'd be nice to start with a board that will fit the chip.

The extra holes will come in quite handy since a .47µF cap. is required.

1 Like

For manual paste deposition, a tool similar to this DIY version makes the job easy:

1 Like

@dougp which ones you waiting on? I thought the only one that people wanted that I was out of was the 3226 boards (and there are strictly better 3227 boards - even if developing for the 3226, the 3227 is better, because you can put those 4 extra pins on debug LEDs during development), I will never understand why the 3227 doesn't blow past the 3226 in sales. The 3227 is strictly better and binary compatible (ie, code compiled for a 3226 will run on a 3227 without modification. The reverse is also true, as long as you don't use those pins that the 3226 doesn't have (but this isn't catastrophic - hardware just ignores the writes to those pins). They are also code compatible as long as pins are referenced with PIN_Pxn notation (eg, PIN_PA2, PIN_PB3, etc) instead of the deprecated numeric pin numbers (the numbers are used internally, but I want to get people away from what I see as a dangerous paradigm of knowing pins by arbitrary numbers that can be mapped at the discretion of the core author (I am very vigilantly guarding the core against the sin of alternate pin mappings, yasee - because I saw what happened to ATTC)

begin rant
Look at ATTinyCore - The x7, x8, x61, x41, 1634, and x4 ALL HAVE AT LEAST TWO PIN MAPPINGS. The only difference is the numbers are assigned to the pins differently and some I/O API functions are bulker and slower or smaller and faster vis-a-vis other pinmaps (there are smart and stupid ways to number pins. An example of a smart way is numbering a Tiny84 clockwise, starting with PA0, that way digitalPinToAnalogChannel() trivial. An example of a stupid way is numbering darts from 0-14 (last pin is always reset over there), and throwing them at a package diagram taped to a bulletin board. That's the most charitable explanation I can come up with the the original 167 pin map. Only time I've had to do digitalPinToAnalogChannel with a lookup table (other explanation include demonic possession or a techno sadistic motivation to make users miserable, Then Digispark came up with their numbering, which had even more dumb whackjob stuff about it.....

All the parts in 2.0.0 dev on ATTC have at least one modern Spence Konde pinout (If you look at the pinout diagram, a "correct" pinout usually jumps right out (usually PA0-7, PB0-7, etc), which tries to maximize efficiency and have a simple logic to it. That's the pinmap I use. I don't know why bad pinmaps are so popular.
End rant
Start whining
Today was the final day of a hellish moveout. So I haven't been able to focus so intently on the store the last week or few. Second undesired move in 2 years first time the landlord turn from Dr. Jeckle to Mr. Hyde kicked us out and threatened to sue us. Second time I got sold down the river by my roomates (I didn't want to move.... but they didn't tell me they did until it was too late. And when I finally did have people it was rented.
End whining
Start rant 2
It costs more than a month of rent to move - easily. $1500 for movers, $300 for storage room that doesn't even fit half my stuff plus $200 per month for that, my childhood bedroom stacked literally to the ceiling, $100 in boxes, >$500 in stuff that I wanted but couldn't get into a car in one of the limited number of trips I had, and if I valued my time at $10/hr (ridiculous, as I charge upwards of 60/hr freelance possibly less if it will be public additions to the cores and isn't some odd-ball), another $1000-1500 in time. Moving is one of the most miserable experiences imaginable. And I have to leave >$500 of stuff behind simply because we were out of car trips and the car was full. (note that I don't own a car (worked great at my last place, pre-pandemic), and the roof over my head since the bed was removed by the wallet-cleaners on saturday is.... a ~1 hour drive away. I can't even borrow a car because the xxxxxx (mod edit rude word removed) in my state don't send out license renewals and mine expired.
End rant 2

So yeah I have not really been paying attention to stock.

I dunno. Whichever ones you meant by post #34.

Just checked the store site and can't see anything 'new' or that looks like my itty-bitty chip would fit on. Doesn't mean it's not there, just that I can't find it.

No worries, this isn't a production environment nor do I have any specific project in mind.

Aaaaha! Okay, yeah, what package did you want to use (or better link the datasheet - because a lot of these new packages are not standard (ie, thee might be two SON-8 packages that were totally different. Just like QFN's (even sticking ti QFN's with a single exposed pad and quadrilateral symmetry, "QFN12" can refer to 4 different packages, the smallest half the size in both dimensions. QFN28 is another 4-variant package, plus at least two rectangular QFN28 packages with different parameters)? A few new breakout boards were posted so it might already be available - but a few haven't been designed - and as it happens I'm trying to fill some space on a poor man's panel right now, and breakout boards like these are perfect for that.

So if you let me know the SMD packages you're looking for I either have them or will have them with my next order, which I am hoping to place very soon because my Rev. - of my UV curing light (which I need for curing solder mask and UV-curing resins) was bungled - and it is a constant inconvenience to not have a decent one.

That is coming alongside an Azduino Nano DA (which will go on sale shortly after the Azduino Nano DB and DB+ ) There are also ATtinys (412, 1616, 3217, 1624, 3224, 1626, 3226, 3227 and AVR32DD-20 boards coming out in my ultramini line which are sized and shaped to fit perfectly in a DIP socket - these are likely the first thing to list, just need to finish docs for those. Same for the Nano DB's but they need more doc work, and the assembled DB and EA48 boards also need docs work. Plus, if you name another package that I don't have a breakout for, I have a place on a poor-man's panel for it

Oh, and Revision -----ing H of my light controller board - yup, I've bunged 9 versions of it. Rev - was just thrown together, Rev. A mounted a pro mini on the back, but that had a wide variety of issues, then I put a 328 or 1284p onto the board instead of sticking another board down - but I then realized I'd forgotten an extra pin I needed, so I made some 328pb boards. Burned all of those and all of the earlier ones out trying to debug an intermittent whereby several terminals in a connector simutaneously could fail to make contact and applied 19v to the 5v rail. Rev D was no good, it used a avr128db32, but one of the power and ground pins was reversed. And I was still tight on pins, so I went to an AVR128DB48, (it was with this that i finally figured out why they were failing, I think after losing one. When I finally made the key connection, I was replacing a controller, it was working with the new controller (which I'd added some entirely insufficient countermeasures to just in case (but guessed wrong what was going on) and then the animation stopped and I felt the chip getting hot, and all I'd done was move it. That meant only one thing - an intermittant in connectors between the sections of the system. Now knowing that, I drew out the diagram and then erased combinations of connections between segments at connectors until I found one that put 19v across the 5v rail. Now we were up to Rev F when I added an OVP IC (which cost 50 cents. Discontinued. The next cheapest option costs almost 10x the price. I bought 100 of the (mod edit rude word removed) things), and a quartet of dual schottky protection diodes and current limiting resistors on the lines that go to the lights. Rev G switched to a smaller chip for that, and added even more OVP countermeasures. Finally Rev. H switched to a latched connector to hold the wires that go to the lights, eliminated the dupont header for programming via bootloader and instead brought it out to a molex picoblade connector which should be much more convenient, cause it can go through the side of the case, and it moves the I2C pins from east (mod edit rude word removed) to my expansion header and sneaks in a debugging/developers button. (If I ever finish this, that'll be the "L. E. D. Zepplin" wifi interface, which will connect to my (unwritten so far) "Cement Cloud" backend. (a "Lead zepplin" (or Lead balloon) as in the metal, and "cement cloud" are used in english ideoms like "That's gonna go over like a cement cloud", for "Nobody's going to accept that" - that's where the band LED Zepplin got their name. Was originally spelled Lead, but people pronounced it as leader, which as not the intent at all.

And now, the very first thing I do when I get a string of those 281x-alikes?

Cut those JST SM connectors off, and put them in the pile of stuff to put out in front of someone else's house, because it's embarrassing to have wound up with them in my posession, and I'd be the laughing stock of garbage collectors, can hobos, and even the rats that get into the trash in that shithole. At least I'm moved out of that awful place. Nothing worked there. None of the neighbors were sane - the most normal neighbors was the weekly fiesta next door, and the boozers downstairs. I've seen one adult son yelling at cars, and instead of reining him in, his elderly dad helps out by banging on the hood of the car with his cane. there's someone who looks like they've been a heavy meth user for their entire adult life across the street, or maybe they just age really poorly, and another person in the same building gets periodic wellness checks, and we've seen her get dropped off by an ambulance and immediately chainsmoke three cigarettes. Next door there's a bed in the lawn, a refrigerator on the porch, and the upstairs sunroom over there looks like it could fall at any minute, an elderly lady described in graphic detail what she'd do if the landlord made another pass at her daughter, to my elderly mother while we were loading the car.. There's no nearby supermarket, but there's a bakery that looks like a mob front, and a convenience store that - I wouldn't trust any of their food to have been made after the turn of the millenium, but they sure sell a lot of scratch tickets. Then there was the other neighbor who died, and the person who got taken off for a fent overdose, and at least two nearby domestic disputes, and the constant smell of natural gas outside that the gas company comes to regularly and claims to find nothing. In short that place sucked, but even on that lunatic fringe neighborhood, it would be embarassing to use what is likely the worst of the common connectors. Yeah, that place SUCKED but it didn't suck as much as moving did.

I have now repinned those cables three times (twice with JST-SM, the second time because of a systematic error that resulted in every female terminal being dead upon crimp. Then after discovering JST-SM is an utterly unfit connector purpose for anything requiring an electrical connection, I switched to Molex MicroFit 3.0, and have added numerous countermeasures. 5 terminals per end, 1 of which is really nasty and the other of which was nasty (wire was too thick. I used wire 4 AWG more than the maximum from the datasheet. Since chinese wire is average 4 AWG undersized, I thought I'd be good, but what a time to get wire that was only 2 AWG undersized. Crimping was hell. And yeah there are at least a dozen sections.... 360 terminals,480 crimp actions (the last part was done with P-707, because after all that the last thing I was going to do is use a chinese ratchet crimp.

Here's a pdf link: https://ww1.microchip.com/downloads/en/DeviceDoc/MIC5019.pdf

Image of current mounting - showing small package size. What SMD equipment do I need? - #68 by dougp

Eeeeew....

That package is not gonna be fun to solder, you realize that right? No matter how to do it. QFNs with just a few pins and no exposed pads are the hardest parts Ive soldered (I haven't done BGAs - but 0.4mm pitch QFN20 I consider to be "not hard generally, 0.5mm pitch 48-64 pin TQFP slightly worse - both get near 100% yeild, just a matter of how much farting around with it required, and QFN's get easier the bigger they are (and hence the larger the exposed pad is - that exposed pad is your best friend, cause it gives you self-centering on any non-iron soldering method.

I'll make the breakouts - I think there's another similar package with 4 pins and just different dimensions for the back side. But they.... they are not going to be fun to solder I'm afraid.

I'll have the stencil file posted as usual by then, so you can laser-cut/beg-someone-with-a-laser-cutter/have OSHstencil make one That would make either hotplate or reflow +touchup viable. I honestly might first resort hotplate if using stencil. Though the way I'll shape the pads, it will definitely be ironable, and possibly even more easily iron-soldered than with stencil (for small quantities at least), since you'll be able to do the standard tack-a-corner-at-a-time (as you can see it's almost like a compacted QFN8, with the 2 leads in each corner combined into one - A single pad under both would I think be the way to go.

I'll extend the pads out from the corners so you can get heat under it that way, which should help. I mean, no matter what it'll still be a cake walk compared to MiniQFN 16

Yeah, I know.  It took a couple of hours experimenting, testing, thinking, destroying and grumbling to get the kludged version working.

Anyhow, I figure a board more oriented toward this part has got to be easier to work with.

Too bad about your location travails.  That sounds pretty dicouraging.

1 Like

I dunno what it is with some companies and these whackjob packages... Our friends at Microchip list on their website no fewer than 30 "4 terminal" packages (that means 4 terminal, as well as 4 terminals plus an EP). Some of those are duplicates (or effective duplicates - that is, they differ only in height/thickness/presence of wettable flanks - the sort of stuff that matters in industry, but less to individuals, and a couple have leads - looks like... they're 4 te), but that still leaves like a dozen 4-terminal no lead packages from Microchip alone Obviously, there's good deal of overlap between companies Very - but there are also a lot of semiconductor manufacturers. Even if each company originally makes only one new package, as they buy eachother out, and then keep supporting existing customers, the number of whacko packages will concentrate in the largest chipmakers. Similar conceptually how some pollutants concentrate in the apex predators.

Packages as large as 5x7mm, or as small as 0.85mm x 0.85mm are in microchip's 4-terminal no lead catalog (and with the exception of uncommon high power parts, that pretty much runs the gamut - much below 1mm square (which translates to extremely tight placement tolerance) is hard to place, and the largest common SMD part seen in 5mm x 7mm is a crystal - everything else has more pads if it's that gigantic.

Probably the hardest part of making these breakout boards is determining which packages are being sold as the latest and greatest, and which are near EoL, and thus, which packages to make breakout boards for, and which to shun.