You can also change the clearances if you think the defaults are too conservative, but there's like 3 or 4 different places where the clearance is set (board, net class, pour, pad) and a bit of a hierarchy to learn which clearance setting overrides the others.
In the case of D10 and return current, we're talking about signal integrity - ie transient operation rather than DC current. You don't need a DC connection between two components for current to flow. D10 forms a transmission line composed of two conductors - the ground plane and the trace D10. They are connected mostly through capacitance (two conductors with a dielectric between them). If you trace things far enough back though, you'll find that the whole circuit will have a shared ground (including whatever is connecting to D10), and the charge / discharge current to D10 will form a loop including ground. Return current always takes the path of least resistance (or reactance in the transient case), which is typically directly underneath the trace that is carrying a signal. I found a decent article explaining this here: resources.altium.com/p/what-return-current-path-pcb
Thanks for sharing your view of a pcb. I m always keen to learn more on how to make a good pcb. However i see on youtube people who advise to use only the bottom plane (on 2 sided pcb) as ground and route as few traces as possible on it. This gives you a better way to route the oscillator pe. Its actual return path remains long. Placing components on another spot can also reduce traces length if critical. Using some smd can be usefull for decoupling the ic. Indeed implementing a pcb board takes a lot of time if you want to do it as perfect as possible. As a hobby-ist you have to be happy with the result as some point. Some basic rules, as you pointed out, should nevertheless be followed. Looking forward to see another review of a pcb board. Kind regards Yves
Informative tutorial it is! Wondering if any custom PCBs may help for any upcoming content? Is so, would love to supply for free and reach any YT collab together if possible. (PCBWay zoey)
you are really bad at this and you pretend to be really good at it. i feel bad for anyone that takes your advice. about the only thing i agree with (i watched like 40% of this video) is the ground plane. sorry for the bad news, i'm looking out for the students here
@@electronicswithemrys ok but i'll preface this by saying that i'm by no means an expert. i've only done a handful of boards or less and am educated basically by the internet. the traces you spread out look really bad and waste a lot of space. you have to consider the signal speed and 99% it would have been perfectly fine to put them right next to each other (if you look at professional boards that's what you would see all the time, never like you have it). there is no via stitching for the ground planes, which is also super common on any professional board. for example i would have put 3 or more vias under the MCU where the trace cuts it up or between D7 and D8 for good measure. i always put 3 fiducials on my board, not sure if those are needed anymore but that's what i did. i put the layer number in copper to help with the gerber files, plus a part number, and i put a version in silk screen. the red layer with or without blue layer looks almost the same, hard to read the board. if it's a 2-layer board i would do something like red and green and where they overlap it would look yellow or some kind of color selection to make it easier to read (that's minor though). I've never heard about that you can't have traces meet up at angles like that, i'm not sure where you got that from. if you want to to be super picky about the etching process then you would use a grid instead of a solid copper plane. don't have silk screen go off the board. i don't know what RV1 is but possibly missing polarity indicator or at least a square pad, maybe it doesn't matter. the routing could be better, for example move r1 slightly to the left so almost the whole trace can be on the top layer and try to minimize ground plane interruptions by using vias and keeping most of the traces on the top. i can't find anything else and i looked hard.
@@dronefootage2778 Thank you for elaborating on your comment - I always appreciate constructive criticism. You bring up some good points. Much of what I applied to this person's board is overkill for the size of their design. As I mentioned to the person originally, their board would likely have worked fine if they didn't change a thing -- but they asked for advice and I wanted to have them get into good habits. If you design one board with poor practices, your next will likely inherit the same issues and it can cause trouble in the long run. "the traces you spread out look really bad and waste a lot of space. you have to consider the signal speed and 99% it would have been perfectly fine to put them right next to each other (if you look at professional boards that's what you would see all the time, never like you have it)." I'd say the space is only wasted if it was going to be used for something else. Spread out signal traces when you can. Another approach is to add a ground flood fill and stitching (that you mentioned) to improve isolation with the traces run much closer together. One minor note is to be careful not to confuse signal speed (for example, 1 kbps) and signal bandwidth -- many system designers forget that digital signals have very fast edges (often only nanoseconds) that can be significantly impacted by poor signal integrity practices -- even a 1 kbps digital signal has edges that push into the 300+ MHz bandwidth range. You are absolutely correct that professional boards (take a motherboard for example) will have very little free space and will have traces run very close together, but they are also very carefully designed to make sure that each trace that carries high speed signals has the appropriate impedance (and thus a good ground return path). Robert Feranec has done a lot of great videos on signal integrity -- I couldn't find the exact one I was thinking of on this subject, but here's one that discusses problems with running traces too close together: ua-cam.com/video/EF7SxgcDfCo/v-deo.html "i always put 3 fiducials on my board, not sure if those are needed anymore but that's what i did." Fiducials are only necessary if you're getting a board assembled by a contract manufacturer. I use them on all of my designs for my work, because I (almost) always have to work with a CM, but for home/hobbyist use, there's no reason to include them -- unless you just want to as a good practice. I can't disagree with doing things right, even when you don't have to :). "i put the layer number in copper to help with the gerber files, plus a part number, and i put a version in silk screen. the red layer with or without blue layer looks almost the same, hard to read the board. if it's a 2-layer board i would do something like red and green and where they overlap it would look yellow or some kind of color selection to make it easier to read (that's minor though)." These all seem like more of personal preferences than real requirements -- not a bad approach to use in my opinion, but I've never labelled layers besides naming gerbers, and I've never had a board house get one wrong. I know plenty of people who like to change the default color schemes - generally I don't, but to each his own. "I've never heard about that you can't have traces meet up at angles like that, i'm not sure where you got that from." I had a number of grey-haired engineers burn this into my mind at a relatively young age and it's stuck with me. My company's default rule set in Altium also identifies any angles less than 90 degrees, so it pops out at me every time I see it. Just briefly searching I found many articles on the topic -- here's one from a CM: gesrepair.com/pcb-acid-trap-causes/ "if you want to to be super picky about the etching process then you would use a grid instead of a solid copper plane." I haven't heard anything about a grid vs plane changing the etching process at all. I'm getting outside of my area of expertise here, but I understand that grids are usually done for flexibility or concerns with uneven heating / warping of boards. I've never used a grid or seen one used, but that may be related to my particular specialty. "don't have silk screen go off the board. i don't know what RV1 is but possibly missing polarity indicator or at least a square pad, maybe it doesn't matter. the routing could be better, for example move r1 slightly to the left so almost the whole trace can be on the top layer and try to minimize ground plane interruptions by using vias and keeping most of the traces on the top. i can't find anything else and i looked hard." I definitely focused my response to this board on the digital signals and almost completely ignored the power section and the silkscreen - I also didn't look over the schematic, so technically the board could be totally wrong and I wouldn't know - I assumed a lot. I'm sure you could find other ways to improve the board if you keep looking -- the person that shared the board was very receptive to feedback and others put in their $0.02 on reddit, where I originally saw this. You can see more of what I and others said here: www.reddit.com/r/AskElectronics/comments/10syxyw/pcb_design_review_please/
@@electronicswithemrys "Thank you for elaborating on your comment - I always appreciate constructive criticism." - you're welcome and sorry for my poor attitude, i got issues in the way i talk to people probably. i appreciate that you appreciate feedback. most people never tell me when i could do something better even if they know, makes it harder for me to improve myself. thanks for all your comments. one more thing I wanted to say is that you're right that because of the edge of the digital signal that part has a high frequency, however the signal is given plenty of time to stabilize before it is interpreted. i guess one might wonder if that induces something in the neighboring traces but i only think this would apply to analog inputs because the digital inputs are either pulled low or high and it takes more current to overcome that. i guess if you don't need to space for anything else then i don't see a problem with spreading them out. in my case i would be working with a 4-layer board and like some ground plane on top too so that's what i was thinking in my mind what i would use that space for, but with the 2-layer board i don't know that i see any point in having 2 ground planes on top of one another.
I really don't know how I missed the release day of this video, but it certainly did put a smile on my face
Awesome video, would love to see your review on more complex board. Great stuff.
Yes, kicad will ensure your nets and components are all separated / connected as appropriate
You can also change the clearances if you think the defaults are too conservative, but there's like 3 or 4 different places where the clearance is set (board, net class, pour, pad) and a bit of a hierarchy to learn which clearance setting overrides the others.
very useful practical tips. thank you.
Very useful & straight forward Tips !! Looking forward for in depth tips & techniques. Thanks !!
Awesome content. Would love to see more vids like this
Just subscribed. Looking forward to more videos like this.
the ground plane runs under the substrate and is not connected to D10 right. So how does it provide a return path for the current?
In the case of D10 and return current, we're talking about signal integrity - ie transient operation rather than DC current. You don't need a DC connection between two components for current to flow. D10 forms a transmission line composed of two conductors - the ground plane and the trace D10. They are connected mostly through capacitance (two conductors with a dielectric between them).
If you trace things far enough back though, you'll find that the whole circuit will have a shared ground (including whatever is connecting to D10), and the charge / discharge current to D10 will form a loop including ground. Return current always takes the path of least resistance (or reactance in the transient case), which is typically directly underneath the trace that is carrying a signal.
I found a decent article explaining this here: resources.altium.com/p/what-return-current-path-pcb
Perfect
Thanks for sharing ❤
Thanks for sharing your view of a pcb. I m always keen to learn more on how to make a good pcb. However i see on youtube people who advise to use only the bottom plane (on 2 sided pcb) as ground and route as few traces as possible on it. This gives you a better way to route the oscillator pe. Its actual return path remains long. Placing components on another spot can also reduce traces length if critical. Using some smd can be usefull for decoupling the ic. Indeed implementing a pcb board takes a lot of time if you want to do it as perfect as possible. As a hobby-ist you have to be happy with the result as some point. Some basic rules, as you pointed out, should nevertheless be followed.
Looking forward to see another review of a pcb board.
Kind regards
Yves
Good job :) subscribed
Informative tutorial it is! Wondering if any custom PCBs may help for any upcoming content? Is so, would love to supply for free and reach any YT collab together if possible. (PCBWay zoey)
you are really bad at this and you pretend to be really good at it. i feel bad for anyone that takes your advice. about the only thing i agree with (i watched like 40% of this video) is the ground plane. sorry for the bad news, i'm looking out for the students here
Thanks for the feedback. I'd love to hear your take on what I'm getting wrong.
@@electronicswithemrys ok but i'll preface this by saying that i'm by no means an expert. i've only done a handful of boards or less and am educated basically by the internet. the traces you spread out look really bad and waste a lot of space. you have to consider the signal speed and 99% it would have been perfectly fine to put them right next to each other (if you look at professional boards that's what you would see all the time, never like you have it). there is no via stitching for the ground planes, which is also super common on any professional board. for example i would have put 3 or more vias under the MCU where the trace cuts it up or between D7 and D8 for good measure. i always put 3 fiducials on my board, not sure if those are needed anymore but that's what i did. i put the layer number in copper to help with the gerber files, plus a part number, and i put a version in silk screen. the red layer with or without blue layer looks almost the same, hard to read the board. if it's a 2-layer board i would do something like red and green and where they overlap it would look yellow or some kind of color selection to make it easier to read (that's minor though). I've never heard about that you can't have traces meet up at angles like that, i'm not sure where you got that from. if you want to to be super picky about the etching process then you would use a grid instead of a solid copper plane. don't have silk screen go off the board. i don't know what RV1 is but possibly missing polarity indicator or at least a square pad, maybe it doesn't matter. the routing could be better, for example move r1 slightly to the left so almost the whole trace can be on the top layer and try to minimize ground plane interruptions by using vias and keeping most of the traces on the top. i can't find anything else and i looked hard.
@@dronefootage2778 Thank you for elaborating on your comment - I always appreciate constructive criticism.
You bring up some good points. Much of what I applied to this person's board is overkill for the size of their design. As I mentioned to the person originally, their board would likely have worked fine if they didn't change a thing -- but they asked for advice and I wanted to have them get into good habits. If you design one board with poor practices, your next will likely inherit the same issues and it can cause trouble in the long run.
"the traces you spread out look really bad and waste a lot of space. you have to consider the signal speed and 99% it would have been perfectly fine to put them right next to each other (if you look at professional boards that's what you would see all the time, never like you have it)."
I'd say the space is only wasted if it was going to be used for something else. Spread out signal traces when you can. Another approach is to add a ground flood fill and stitching (that you mentioned) to improve isolation with the traces run much closer together.
One minor note is to be careful not to confuse signal speed (for example, 1 kbps) and signal bandwidth -- many system designers forget that digital signals have very fast edges (often only nanoseconds) that can be significantly impacted by poor signal integrity practices -- even a 1 kbps digital signal has edges that push into the 300+ MHz bandwidth range.
You are absolutely correct that professional boards (take a motherboard for example) will have very little free space and will have traces run very close together, but they are also very carefully designed to make sure that each trace that carries high speed signals has the appropriate impedance (and thus a good ground return path). Robert Feranec has done a lot of great videos on signal integrity -- I couldn't find the exact one I was thinking of on this subject, but here's one that discusses problems with running traces too close together: ua-cam.com/video/EF7SxgcDfCo/v-deo.html
"i always put 3 fiducials on my board, not sure if those are needed anymore but that's what i did."
Fiducials are only necessary if you're getting a board assembled by a contract manufacturer. I use them on all of my designs for my work, because I (almost) always have to work with a CM, but for home/hobbyist use, there's no reason to include them -- unless you just want to as a good practice. I can't disagree with doing things right, even when you don't have to :).
"i put the layer number in copper to help with the gerber files, plus a part number, and i put a version in silk screen. the red layer with or without blue layer looks almost the same, hard to read the board. if it's a 2-layer board i would do something like red and green and where they overlap it would look yellow or some kind of color selection to make it easier to read (that's minor though)."
These all seem like more of personal preferences than real requirements -- not a bad approach to use in my opinion, but I've never labelled layers besides naming gerbers, and I've never had a board house get one wrong. I know plenty of people who like to change the default color schemes - generally I don't, but to each his own.
"I've never heard about that you can't have traces meet up at angles like that, i'm not sure where you got that from."
I had a number of grey-haired engineers burn this into my mind at a relatively young age and it's stuck with me. My company's default rule set in Altium also identifies any angles less than 90 degrees, so it pops out at me every time I see it. Just briefly searching I found many articles on the topic -- here's one from a CM: gesrepair.com/pcb-acid-trap-causes/
"if you want to to be super picky about the etching process then you would use a grid instead of a solid copper plane."
I haven't heard anything about a grid vs plane changing the etching process at all. I'm getting outside of my area of expertise here, but I understand that grids are usually done for flexibility or concerns with uneven heating / warping of boards. I've never used a grid or seen one used, but that may be related to my particular specialty.
"don't have silk screen go off the board. i don't know what RV1 is but possibly missing polarity indicator or at least a square pad, maybe it doesn't matter. the routing could be better, for example move r1 slightly to the left so almost the whole trace can be on the top layer and try to minimize ground plane interruptions by using vias and keeping most of the traces on the top. i can't find anything else and i looked hard."
I definitely focused my response to this board on the digital signals and almost completely ignored the power section and the silkscreen - I also didn't look over the schematic, so technically the board could be totally wrong and I wouldn't know - I assumed a lot. I'm sure you could find other ways to improve the board if you keep looking -- the person that shared the board was very receptive to feedback and others put in their $0.02 on reddit, where I originally saw this. You can see more of what I and others said here: www.reddit.com/r/AskElectronics/comments/10syxyw/pcb_design_review_please/
@@electronicswithemrys "Thank you for elaborating on your comment - I always appreciate constructive criticism." - you're welcome and sorry for my poor attitude, i got issues in the way i talk to people probably. i appreciate that you appreciate feedback. most people never tell me when i could do something better even if they know, makes it harder for me to improve myself. thanks for all your comments. one more thing I wanted to say is that you're right that because of the edge of the digital signal that part has a high frequency, however the signal is given plenty of time to stabilize before it is interpreted. i guess one might wonder if that induces something in the neighboring traces but i only think this would apply to analog inputs because the digital inputs are either pulled low or high and it takes more current to overcome that. i guess if you don't need to space for anything else then i don't see a problem with spreading them out. in my case i would be working with a 4-layer board and like some ground plane on top too so that's what i was thinking in my mind what i would use that space for, but with the 2-layer board i don't know that i see any point in having 2 ground planes on top of one another.