Top Fifteen Mistakes People Make When Designing Prototype PCBs

after assembling and helping test countless prototypes for my day job as well as designing and shipping several of my own PCB designs I thought I'd share some observations about how people end up making prototyping and testing harder for themselves in their race to get to production here are my 15 mistakes people make designing prototype pcbs number one stale breadboards breadboards are a great way to learn Electronics try out designs test modules and get circuits up and running in an afternoon however breadboards are somewhat fragile and every wire is just friction fit and will eventually become loose the more complex your breadboard gets and the longer you keep it around the more chances things can fail breadboards are great to test specific Concepts and parts of your design but trying to design an entire project on a breadboard especially one developed over many weeks or months will become an exercise and frustration of chasing loose wire connections the breadboard might even delaminate and fall apart true story as as you can see in this picture number two making hardboards DIY or etching if breadboards are so temporary you might be tempted to solder a hardboard or a perf board but this has some of the same problems as breadboards the more wires you have the more chances there's a hidden break or short that goes unnoticed tracing problems becomes more difficult with each wire you add the concept of prototyping pcbs at home using a CNC mill laser or acid etching is also tempting but they to be messy require volatile chemicals and have limits in how fine a pitch they can support secondly the entire electronics Industry is against you everything is going towards mobile chips are smaller chips are multi-functional footprints are getting ridiculously small very few chips developed in the last 25 years are even available in a dip or S soic package any modern design you work on is going to call for at least one leadless qfn vfn or BGA footprint while manufacturers are happy to produce a few components for us hobbyists we're buying in quantities of 100 at a time their main customers are ordering by the millions every inch of production line space has to be justified by how much product it can sell and we lose out for these reasons I suggest going to a PCB as fast as possible if you can wait 7 days you can have a PCB in your hands for a few dollars number three designing for production when designing a PCB for a new project we always have this optimism that the first PCB will work and we'll only have to make a few minor changes to go forward to production after seeing this not play out too many times on too many occasions I now design my first PCB for any project expecting it to fail and with the mindset and goal of prototyping and solving problems I don't try to make the PCB a certain size or shape I don't try to fit it into enclosure all I care about is testing all the functionality sensors serial communication battery recharging power routing Etc and because PCB manufacturers like jlc PCB and PCB way will sell you a PCB that's 100 mm Square 4 in for a dollar or two I take advantage of that and I make my pcbs bigger and worry about shrinking things down later on number four no test points one of the biggest mistakes I see on new prototype pcbs is not adding test points on a brand new design you will make mistakes to find and fix those mistakes you need to test and probe and if you find a mistake reroute if you didn't place any test pads then you're going to have to hook up wires or clip test leads onto a leg of a very small IC try to probe small components cut traces run bodge wires or you may be totally stuck and have to order an entire new PCB because of one mistake by adding test pads especially for the more common things like 3.3 5V 12vt battery ground i2c uart you make it less like likely that a single mistake will totally block your progress you can even use test hooks which are surface mount copper Loops that you can hook test probes onto number five no power or diagnostic LEDs I'm always surprised at prototype pcbs that don't have any diagnostic lights something as simple as an indicator light showing 3.3 volts 5 volts battery charging or a ready LED when your microcontroller switches on at first run can save you hours of banging your head against a wall on simple mistakes number six overcrowding components the Temptation when designing a PCB is to use as little space as possible and pack everything close together with the anticipation of the enclosure it will be in however at the prototyping stage you're just making more work for yourself while the minimum distance between components is determined by the courtyard it's best to provide additional space around your components so that they can easily be added or removed or changed out as you test and find problems only decoupling caps need to be super close and I like to keep my passives relatively large with 805 caps and resistors so that they're easy to remove at this stage once you've solved every problem and have a working design then you can worry about part size and smaller pcbs number seven not taking advantage of the silk screen when assembling your new board there's nothing more frustrating than having to constantly check the data sheet to figure out which way each component goes or your design files to figure out which part goes where your PCB design should have a clear silk screen marking for each component with a DOT or line indicating pin one on each IC and additional markings to suggest part and Chip orientation as well as clear labeling of any test pads and headers Etc it should be extremely obvious which direction each part goes even something as simple as LEDs can go Ary if you haven't standardized on clear markings like cathode pin one silk screen text should ideally be at least 2 mm tall and should be designed so so that when the board is populated the text isn't covered up only on the smallest most space limited boards should silk screen text be less than 1 mm anything smaller will not be readable on pcbs from the most popular Fabricators as they use a direct to print technology by making good use of the silk screen and maybe adding some graphics for fun you'll thank yourself later for an enjoyable build experience and faster testing on simple boards with a short parts list you may prefer to put part rate things like 100k 150 resistor 100 NF mcp23017 rather than reference designators like u12 r98 c37 ETC number eight not using isolation jumpers on a new design there's a high probability of a mistake that renders the entire board non-functional by adding zero ohm resistors or cutable Jumpers on the power lines to each major IC you can easily disconnect part of your PCB and just test part of it without having to remove expensive or rare components the best testing is methodical bring ups and jumpers make that a lot easier number nine not breaking out unused gpios most projects won't use every single general purpose input output or gpio on a microcontroller if you can you should break out one or two additional gpios to test pads on your PCB why if you find a mistake in the finished circuit boards you can use these addition iOS to rewire a component or even add on an external module with extra gpios you can cut a trace and reroute the new connection and be back in business it takes just a few seconds to add these and you can shave off a week's time not having to wait for a new PCB revision to arrive number 10 uart mixups we are living in a Serial world and making sure that transmit and receive pins of all the chips in your designs are correctly paired is important if if you are using uart components in your project look very closely at the data sheets as this can be very confusing you may also wish to use a jumper configuration such as this one posted by topi 44 so that if you make a mistake you can just cut a trace and solder across the finished board to switch the two if you're really concerned you can use this design which requires you to put zero ohm resistors either north south or east west to handle all situations number 11 locking yourself into i squar c addresses no matter how well you plan if you're using more than a few iqu C serial devices in your design it is easy to get tripped up by two devices sharing the same address this can also happen when certain data sheets report it Square C addresses pre-multiplied with the read WR bin looking at you nxp anyway if you've got a few different iqu C devices on your PCB and they have changeable address pins go ahead and give yourself options place two resistors off of that address pin one to power one to ground and you'll be able to change the address with a little solder instead of having to order a whole new PCV just make sure you don't assemble both resistors or you'll have a dead short number 12 in PCB designs of sufficient size and complexity what happens if your power solution doesn't work now you have to populate a whole second PCB even though there was nothing wrong with the microcontroller part of the first one instead you should break up your design into two boards or more it can be wise to have a main or microcontroller area and a separate power PCB that specifically handles power buck boost generating reference voltages of 1.8 3.3 5 volts Etc some PCB design software makes it easy to split a board into two if yours doesn't you can place a row of mouse bites or small drill holes in your PCB design such that with a sharp X-Acto Knife and a crisp snapping motion the board can be cleanly broken into two then if your power design doesn't work out you can revise and order a new PCB and you don't have to throw away all the parts you put on the main board and in the meantime you can test using a benchtop power supply number 13 since 2013 some component manufacturers have stopped marking values on their surface mount resistors this makes visual inspection and methodical testing of prototypes much more difficult if you have a choice especially in the Prototype stage you should really choose to order labeled surface mount resistors number 14 many a PCB has been ordered only for the designer to find out the they've used the wrong footprint on the PCB compared to the parts they have on hand or that they have purchased there are numerous variations within each footprint that vary from manufacturer to manufacturer don't assume the s soic tssop qfn or other footprint in your PCB software from one manufacturer will match your part which might be from another always compare the dimensions on the data sheet to the dimensions on the footprint using the measuring tools in your Eda software just a little extra verification on each major component in your product can save you time and money and number 15 designing and laying out without checking Parts the global part shortage seems to finally be settling out after 3 years but it is still prudent to consider what parts you need versus which parts are actually available prior to designing your circuit and certainly before choosing Footprints and laying out your PCB it might be prudent to speculatively order some parts before you even order your PCB even connectors are becoming difficult to sort plan ahead otherwise this could be you special thanks to topy 44 for their contributions and reactions to this video and thanks to kycad PCB design software which is just a few months away from releasing its eighth major version and third major release in the last 3 years kycad 8 is expected to have a ton of quality of life improvements user interface improvements simulation and already has full support for importing Altium pcbs and schematics give it a try it's really mature in the last 3 years tell me what you think do you agree with these mistakes what other mistakes do you see people making I've already got another video in mind with more about decoupling capacitors Trace widths and mistakes people make when preparing for automated assembly thanks for [Music] watching

Channel: Cosplay Light and Sound

Views: 134,833

Keywords:

Id: hkSad4n76Lc

Length: 12min 26sec (746 seconds)

Published: Wed Jan 10 2024