One of the few videos, I watch again and again.

kicad 💜

we used wire wrap instead of bread boards, proven to be tight and last. Connections are gold plated pins but silver plated wires with tfe insulation. We had bad chips, but no bad wirewrap boards. We were making the first cellphone. And we had wire-wrap sockets for all chip types. This was before prototype labs were good... It took months and thousands to buy 4 or 5 pcb's. My stuff worked, so not a lot of debugging before we moved the chips into an LSI. My first cad based PCB was a disaster, so yeah... And yes, we designed to the available parts, getting samples just wasn't likely for most parts unless you already had production flows that made chip makers care about your business... And yes that was the 1980's.

Is there an analog version of this?

Quantum Leap is still a great TV Show that I watched when I was Highschool.

I dont understand shit. But i love it

Great video. Cheers.

ポカヨケwwwww

Cramming everything in small area because of desire to cheap out in JLCPCB for cheapest board size is not the best idea at all :) Especially when your board have strong signal generators or power MOSFET-s placed to switch around PWM with 10+A current. Also pay attention to dimensions of surrounding mechanical components like frames or heatsinks. Ignoring these may cost dearly when it will turn out that heatsink is too low and mess with some larger components on board. KiCAD definitely matured. Using 8 version now and love it.

I am doing my first pcb right now and while most of these I didn't think of, I can proudly say I've got headers for all gpio's for testing and potential expansion.

When will you guys stop saying "sodder" - ITS SOLDER

There are NO, US circuit board houses that can produce a prototype PCB for you, for a 'few dollars' (sadly). Most want anywhere from $100-$500 depending on board dimensions and some have minimum quantity orders.

I assembled my first computer in 1978 - a Netronics ELF II - after built super-minicomputer, super-microcomputers, and 1000-processor multicomputers - I'm pretty sure that world is now dead...

What EDAs do you use ?

I'm defiantly going to use some of these designee tips in my next PCB build.

These are interesting ideas to think about for new people who wish to develop circuits like me, this video was useful

Good video. I stay away from really tiny parts being a hobbyist... for complete beginners, you should mention all north/south wires go on one side of the PCB, and all east/west wires go on the other side... use vias to jump from one side to the other, when necessary... this keeps unused pathways open and available for future paths. Also provide open holes at a few places on the board, in case you need to run a jumper from one PCB side to the other. I also identify pin 1 on chips with a square pad, same for anodes of diodes and LEDs, and positive term of 'lytic caps. If you are using thru-hole components, make the drill sizes on the pads for the component leads as large as practical so it is easy to wick out all the solder, and replace a part without tearing out the thru plating on the board. To remove thru hole ICs, ( failed chips ) clip all the leads off and remove them one at a time. I always place a text somewhere that identifies the exact filename of the PCB CAD file for that board, so I can get the source info for it. Also my ham radio call sign, or email address so others can find me with any questions. If there is any spare board space, I put extra footprints there for components and ICs, so I can add them "temporary" for circuit changes, using point-to-point wiring. If wires run off the board, try to make them all enter/leave along one edge of the board, so you can just turn the board over to get to the backside for troubleshooting... like turning the pages of a book. If wires run off 2 or more edges, that can't be done... and leave adequate slack in the wires that are installed, to permit this. Unless I know exactly where and how a PCB will be installed in an enclosure, I avoid using parts that will dictate how the board is mounted... like displays, switches, controls or other things a person must operate... because they MUST be mounted on an external panel or surface. Use your imagination... put some forethought into it... don't just do the obvious stuff... Give yourself some room for design mistakes and changes down the road. Temporary thing end up being permanent, and permanent things.... aren't.

16 ... BASIC THERMAL DESIGN. How many pcbs online for sale do you see with electro caps, diodes and resistors etc almost touching heatsinks ?? Add in how does a pcb mounted heatsink breathe? If it cant find cool air then the heatsink is running hotter than design... excess heat means increasing ageing, lower lifetime, and unpredictable drift of design setpoints at best etc. Some circuits may have random logic toggling as well. I always design my pcbs with heat and passive air flow - effective "breathing" - in mind. Sometimes simply moving hot parts to a board edge is enough to allow it to find cooler air. Often ill just add 3 or 4mm holes directly under and between heatsink fins. And ill always add 1cm or more clear space around heatsinks.. I was mentoring a young fella on pcb design and i had to remind him that efective board design must always include identifying and managing heat. it took a while but he got there. Hes selling his boards on ebay now and i am happy with what he offers.. handing over to the next gen our hard won knowledge.. job done 😊 Really liked your very clear well laid out discussion well done. U just forgot this one..

I haven't scrolled through the entire list of comments, so if this issue is listed elsewhere, my apologies. But please have at least four layers in your PCB. I prefer to place the larger and active parts on the top side; under the top side components a ground plane; below the ground plane goes the power plane; the outside layer has the smaller, passive components. Decouple the power and ground planes at multiple places across the PCB. And bypass each active component with an appropriate selection of capacitors. This should be done before any signal routing. Ifthe power distribution system is not stable and adequate you're in for some headaches on high speed PCBs.

Excellent.

Great vid!

It is true that the first PCB will never be 100% correct. I used to work at a electronics company and we only once got a PCB perfect first time. We celebrated and then went to mount the only mechanical bracket on the board and found the mounting hole was in the wrong place.

Another very common error is to design your board around a microprocessor instead the other way round. With clever programming, you can reduce the number of components significantly!

Hello. Im making my first pcb. It's for a big project and will include between 3-5 AT42QT1481-AU(i still haven't figured the difference between AU and AN) and a CM4 as the microprocessor. Any recommendations on designs of the capacitive touch pads, traces and other things you might consider important? I'm using KiCad right now and i'm kind of lost. Any recommendations and tips are welcome.

Excellent advice! There is one additional quite subtle problem with configurable devices, e.g. via I2C: Make sure the circuit/board works not only for the configured, but also for the unconfigured device. Picture a designated GPIO pin which is intially a Reset pin. Bummer.

LOL, I almost fell down my chair when #1 rolled up.... nice one...

Wonderful and very useful video, thanks!

I don’t know how to design PCB’s or anything but still loved this video

I don’t know how to design PCB’s or anything but still loved this video

My eyes welled up as I watched. All components are whispering "feel the pain" to me.

That footprint for UART mix ups - in going to use that everywhere! Thanks!

I'm just getting back into designing my own PCBs again after having taken a break for a couple of years. This video's been a nice refresher and reminder that I've been on the right track since I came back to it, heh. Another tip on checking the size (especially if you've already bought the components) is to just print it out on paper and lay them on the printout. It's 100% scale so will be exactly the same size as the final PCB. Makes component size discrepancies real obvious. :)

I have always assumed that it will take three versions of a new PCB before the design will work. This can be justified as follows: The first version will likely contain the errors and misconceptions about how the design and how new components work. It sorts out foot print errors. The second version corrects the errors of the first version, it also allows design features to be added that were not envisaged in the first version. This is the result of learning about features of the ICs used and any silicon bug fixes. The third version, is the first prototype that fully works and can be tested for environmental performance, including EMC. I favour a modular approach to PCB design, so even if the whole proto type PCB is a failure, it is still possible to test various modular parts of the design. Sometimes a faulty prototype PCB is not a total loss, as some working module of it can be used for testing or even as part of a test jig. The design that ends up in production is usually the third best. The best design is perfection and is impossible to achieve. Second best takes too long to develop and cost too much, so it's third best that goes into production, with one previous, that is it fully meets the product requirement specification. Many designs fail because they are developed to a shifting specification. It is important to have a well defined requirement specification from which the technical requirement specification is developed. This spec allows the designer to determine when the design is good enough, not perfect, but good enough. As far as I am concerned there should be three levels of technical specification. The one published that the customer sees, the one the manufacture test took, and the one the product is designed too. With increasing close tolerance on each. It is the only way to ensure the customer is guaranteed to get what they purchase, the manufacturer is guaranteed to be able to produce, and the original requirement is met. One of the important design processes, particularly on a project with more than one designer, is the rationalisation of the component values used. It should be possible to justify the value an type of each component used. Some will have critical values, but others, such as pull up resistors may just be the designers preferred part, such as 10k and 0.1uf decouplers. Even here there is a hierarchy of reasons for selecting a particular value, it may be cost, preferred manufacture or foot print. In general the few different parts the better, as these take up valuable space on the component placement machine and increase inventory. Before BGA became commonplace, it was possible to populate a new prototype PCB in stages, testing each stage before installing the next. This allowed working out from the cheaper components or more critical stages to the more expensive parts. This saved the loss that resulted from fitting all the components, only to find a short in the power supply tracking, making the PCB and all its components useless. New BGA tech has made partial building impossible or expensive. This is where isolated modular design makes it possible to gain valuable information from even a flawed PCB. It allows the testing of those parts that are wired correctly. If designed with a bit of thought, an intelligent circuit, one containing a processor, DSP or FPGA, can be designed to test itself. In that once communication has been established with the intelligent part, it peripheral can be used to test each other. Fitting the processor component goes against fitting the cheapest parts first, but if the PCB is to be fully assembled because it uses BGA or similar parts, there is no loss in using in-circuit programming to expand testing out from the intelligent part, using each successfully tested section to test the next, maybe using loopback signals. The major requirement is that the power supply section and its associated tracks work and are wired correctly. This emphasises the need to check and recheck the PSU implementation and wiring.

I love the unique pyramid handlink that was used by Thames. I think the sound effects from his were more like distorted versions of the handlink Al uses. Also, the teleportation sound for Thames was a Star Wars stormtrooper blaster. I think that puck handlink is crap. The original Pilot handlink is such a better design all around. Although, considering that in the original series, they passed up a really amazing handlink prop for a simpler design, it doesn't really surprise me that they went with the lame choice in the revival as well

Regarding No 14: I always print out the layout with a normal printer and place all components on it, to check whether they fit or not. That is the easiest way to avoid that problem. Of course, it needs me to buy all components in advance. But this activity is best if one has patience to achieve bet results.Another mistake is not separating analog GND from digital GND.

The repaired wristlink video is mine , if you open it you will find a sticker with "repaired by Brent Kling" on it. I had to remove and replace a couple of the rare LEDs, replace the tether and rewire the driver chip , it was in real bad condition when I got it. The main white switch was worn out and couldn't be replaced so I repaired it.

Every MCU I use has a through hole equivalent dev board. PCB etching works fine for the most part, and I've found many bugs before a single board is purchased. I would also NEVER put component ratings or part numbers on a board which you intend to sell. China rips off everything, there's no reason they won't rip off your design too.

Just because a manufacturer CAN make a board with track and clearance X, doesn't mean you should use that. Adding just 0.001 inch to each value, will make the board easier to build, and less likely to fail during manufacture.

DEAN STOCKWELL QUANTUMLEAP & DEAN STOCKWELL ENTERPRISE

I've made the UART mistake before, but was saved because I had them on jumper resistors. A little fib wiring and it was fixed!

I have literally forgot to power a MAX232 chip one time, also did the RX TX mix up in a last minute break neck project for a trade show on the other side of the country. found out after I arrived and the booth display didn't work. Butane soldering iron, exacto knife, individual stranded wire strand, and the patience of Job fixed that. Great video, you have earned my sub!

it drives me absolutely nuts when people mix up rx and tx. you have to be exceptionally stupid to do so.

Having designed hundreds of PCBs and providing designs for over a hundred more, I can totally agree with every single item mentioned here (when possible at least). Adding test points and bringing out unused microcontroller and FPGA pins has been a life saver countless times during initial testing. The global component shortage caused me to redesign several boards to a multi board design where we would swap out the say a microcontroller block for one using a different controller easily. It also makes testing for new solutions and upgrades easier. Probably the last thing I learned to do maybe a decade or so ago was to spend more time on the silk screen, not only does it help in troubleshooting (proto and production), it is very satisfying to have a great looking board. Kudos to an excellent video.

For point 11 (I2C addresses), would it be helpful to use switches instead of resistors, something like a SPDT switch? Could you prevent the shorting concern that way? Question from a dumb software engineer, please be kind.

I have KDM-7 keypad for a TK-250 Kenwood radio They have been out of existence for decades I have a new old stock one. I need someone who can print the keypad and someone to make the PCB that goes under the plastic keypad. Do you think you can help me ? It looks like simple metal push pads and no chips or even capacitor or resistor on the PCB I just don't know who to contact to help me. It looks like you know what you are doing. Thank you

10 years too late for me...but great video for beginners :D

For #14, wrong footprint, a tip I often use is to print on paper the PCB layout and visually check sizes with my components. Saved me wrong footprints.

Agree; even Hi-Volume production will insist on testpoint on every net(if possible) 1206 components even better if possible Wrong footprint always makes me chuckle; desired part footprint not available? scramble begins.

can you give a part number for the smd test hooks? i searched mouser and digikey

Definitly great advices, but what's orders of magnitude more important is: Please explain how did you manage to get KiCad in Dark theme at 11:35 😍🙏🙏