Unless you are building a really low budget board, you should have a non-routed ground plane which is well connected by non-thermal vias to the surface ground, poured or not. Four layer boards are pretty inexpensive today, and for many applications, the additional planes and the ability to have impedance controlled traces justifies the extra few dollars to get the boards built.
During design I route critical grounds directly even if I know they will be swallowed up by the pour at the end, you can say with certainty that the ground path is good. I think that etching your own boards is a dying art nowadays (sort of sad ?) and double sided boards are the same price so you can stick a full ground pour on the bottom and use 'Vias' but even then care must be taken.I loved your last example as that really is a "Trap for young players" and you had to be careful when just having one side of copper to play with and no DRC ! I enjoy any stuff others have to say on this subject and I hope you will do more sometime in the future as It's always a 'hot' subject...cheers.
Thank you so much for sharing this. I literally ordered boards the night before I saw this video they had the very problem that you talked about. I wish I had seen this before I ordered those boards. It's just on a decoupling capacitor that I have before a rotary encoder so I can jump over a wire from the ground on encoder to the ground on the capacitor and not have a problem but dang it I hate that I missed it. It's the first PCB I've ever created. Anyway love you videos thank you so much please keep doing this I have learned so much from you!
Kicad does too but the point is the last example where up close examination the route is just nuts but electrically sound, this is useful for beginners.
Thanks for the video. I just watched a video presentation from Altium Design featuring an engineer named Rick Hartley on the importance of grounding. You guys raise the same issues. On a completely different topic, as I continue down the rabbit learning RF and everything that goes with it, I’ve been getting into Software Defined Radio. Just last night , I ran into a few scholarly papers on Software Defined Antennae. I could get my head around SDR but antennae configured by software is a bit bizarre. It seems like it’s not brand new but I can’t find much outside Academia on the topic. If you have any info you can share or a direction I can be pointed, I’d appreciate it. Thanks for your time- Jason Burchell
One more detail: If You have several stages on one board it may be a good idea to create separate groundplanes. Otherwise it is so easy to get noise from the digital part right into the analog input stage for example. Or you can easy end up with positive feedback and oszillation if you have an loudspeaker power amp and a input stage sharing the same ground plane for example 🙂
Most of the time this would be bad advise. What you really don't want to happen is having a large the loop area. Use a single ground plane, but place the different circuits far enough apart and don't let currents flow where they shouldn't (both DC and AC). But don't just break the ground plane in order to separate circuits. And most importantly, don't route signal accross different ground planes, except you want to create an antenna.
Given the modern possibility of chinese board manufacturing, where 2-layer and now even 4-layer boards are super cheap, (heck even western PCB prototyping isn't that bad if you don't mind) although a lot of the advice in here is valid, I am puzzled that you never gave the context of the availability of avoiding 75% of the problems presented here with a bottom-layer ground pour. Indeed I'd thought that the most appropriate advice for beginners laying out a board nowadays would be to start off with a 2-layer, bottom-layer ground pour design and route from there. Maybe this stems from the days when a single-layer home-manufactured PCB prototype was the most common, but nowadays ordering them is far easier for me than getting a hold of ferric chloride. I don't mean to be negative just would love to see that possibility mentioned if it makes sense for someone learning from these. Still love the series.
I agree. A good default strategy is a solid ground plane on the second layer, without any traces. That way you don't have to think about the return path (as long as you're routing on an adjacent layer). I don't know how to approach routing if a solid ground pour on layer 2 is not an option: probably first route ground? Given that the (ground) return path should be close and contiguous to every (fast edge/high speed) signal, it seems that manually routing the (ground) return path at the same time would be quite a lot of effort.
Super easy to get muriatic ("pool") acid and hydrogen peroxide these days. I suspect part of the reason for not jumping to the bottom layer for ground by default is because it's not a universal solution. It can be useful for an RF amplifier with plenty of vias. One must omit bottom copper for oscillators and any other circuit where the capacitance between the layers will disrupt the circuit. I look forward to seeing how this series unfolds!
I really like the series, in the research labs I was the one who made the PCB and everything that was said here is true. It's also true that it's easier to order a board these days (I make small boards myself and when I'm super lazy I build it on a prototype board), but in order to order it, I have to be able to design it. The layout of the paths and the placement of the components is not so simple, we must always remember that the circuits affect each other and a wrongly designed PCB can have functional problems even if it is connected correctly. Nice day 🙂Tom
but ground is ground! as long as it's connected to the chassis it's fine lol In school I built a lot of guitar pedals and amplifiers of various sizes, shapes and forms (nothing of big power really). For the most part, the ground schemes worked fine (just by chance), but in things with low signal (high amplification with low noise), very high/very low impedance, or in circuits carrying digital and analog components, the correct ground routing it's an art. I only learned many years later, after several headaches xD
I still don't understand ground in circuits, ive watched many videos. If I take a circuit diagram and try to build it, its very hard to understand where the grounds are supposed to go. They just end abruptly with the little symbol, how are you supposed to know where that connects to in reality?
yes, it is not obvious. you need to mentally think about all the current flows. it's like knowing where something is located on the map but finding the best way to get there the map doesn't tell you.
No. It will show a non-route trace, but that is very easy to not see and miss. and it does not show up when you do a error check. So, it is up to you. I like to click on a ground trace and then the ~/` key. this will highlight all grounds
Unless you are building a really low budget board, you should have a non-routed ground plane which is well connected by non-thermal vias to the surface ground, poured or not. Four layer boards are pretty inexpensive today, and for many applications, the additional planes and the ability to have impedance controlled traces justifies the extra few dollars to get the boards built.
During design I route critical grounds directly even if I know they will be swallowed up by the pour at the end, you can say with certainty that the ground path is good. I think that etching your own boards is a dying art nowadays (sort of sad ?) and double sided boards are the same price so you can stick a full ground pour on the bottom and use 'Vias' but even then care must be taken.I loved your last example as that really is a "Trap for young players" and you had to be careful when just having one side of copper to play with and no DRC ! I enjoy any stuff others have to say on this subject and I hope you will do more sometime in the future as It's always a 'hot' subject...cheers.
I really enjoy these PCB design videos! Thanks!
Thank you so much for sharing this. I literally ordered boards the night before I saw this video they had the very problem that you talked about. I wish I had seen this before I ordered those boards. It's just on a decoupling capacitor that I have before a rotary encoder so I can jump over a wire from the ground on encoder to the ground on the capacitor and not have a problem but dang it I hate that I missed it. It's the first PCB I've ever created. Anyway love you videos thank you so much please keep doing this I have learned so much from you!
I didn’t think about those thermal reliefs when I designed my lpf boards. Thanks for the info! …back to KiCad…
5:54 a DRC check would tell that there are errors. At least Eagle does.
Kicad does too but the point is the last example where up close examination the route is just nuts but electrically sound, this is useful for beginners.
Thanks for the video. I just watched a video presentation from Altium Design featuring an engineer named Rick Hartley on the importance of grounding. You guys raise the same issues.
On a completely different topic, as I continue down the rabbit learning RF and everything that goes with it, I’ve been getting into Software Defined Radio. Just last night , I ran into a few scholarly papers on Software Defined Antennae. I could get my head around SDR but antennae configured by software is a bit bizarre. It seems like it’s not brand new but I can’t find much outside Academia on the topic. If you have any info you can share or a direction I can be pointed, I’d appreciate it. Thanks for your time- Jason Burchell
One more detail: If You have several stages on one board it may be a good idea to create separate groundplanes. Otherwise it is so easy to get noise from the digital part right into the analog input stage for example. Or you can easy end up with positive feedback and oszillation if you have an loudspeaker power amp and a input stage sharing the same ground plane for example 🙂
Most of the time this would be bad advise. What you really don't want to happen is having a large the loop area. Use a single ground plane, but place the different circuits far enough apart and don't let currents flow where they shouldn't (both DC and AC). But don't just break the ground plane in order to separate circuits. And most importantly, don't route signal accross different ground planes, except you want to create an antenna.
Thank you @1:40 for thermal vias. 💪
Given the modern possibility of chinese board manufacturing, where 2-layer and now even 4-layer boards are super cheap, (heck even western PCB prototyping isn't that bad if you don't mind) although a lot of the advice in here is valid, I am puzzled that you never gave the context of the availability of avoiding 75% of the problems presented here with a bottom-layer ground pour.
Indeed I'd thought that the most appropriate advice for beginners laying out a board nowadays would be to start off with a 2-layer, bottom-layer ground pour design and route from there.
Maybe this stems from the days when a single-layer home-manufactured PCB prototype was the most common, but nowadays ordering them is far easier for me than getting a hold of ferric chloride.
I don't mean to be negative just would love to see that possibility mentioned if it makes sense for someone learning from these. Still love the series.
I agree. A good default strategy is a solid ground plane on the second layer, without any traces. That way you don't have to think about the return path (as long as you're routing on an adjacent layer).
I don't know how to approach routing if a solid ground pour on layer 2 is not an option: probably first route ground? Given that the (ground) return path should be close and contiguous to every (fast edge/high speed) signal, it seems that manually routing the (ground) return path at the same time would be quite a lot of effort.
Super easy to get muriatic ("pool") acid and hydrogen peroxide these days. I suspect part of the reason for not jumping to the bottom layer for ground by default is because it's not a universal solution. It can be useful for an RF amplifier with plenty of vias. One must omit bottom copper for oscillators and any other circuit where the capacitance between the layers will disrupt the circuit. I look forward to seeing how this series unfolds!
I really like the series, in the research labs I was the one who made the PCB and everything that was said here is true. It's also true that it's easier to order a board these days (I make small boards myself and when I'm super lazy I build it on a prototype board), but in order to order it, I have to be able to design it. The layout of the paths and the placement of the components is not so simple, we must always remember that the circuits affect each other and a wrongly designed PCB can have functional problems even if it is connected correctly.
Nice day 🙂Tom
but ground is ground! as long as it's connected to the chassis it's fine lol
In school I built a lot of guitar pedals and amplifiers of various sizes, shapes and forms (nothing of big power really). For the most part, the ground schemes worked fine (just by chance), but in things with low signal (high amplification with low noise), very high/very low impedance, or in circuits carrying digital and analog components, the correct ground routing it's an art. I only learned many years later, after several headaches xD
@0:14 Your grounded 😊 - ok let’s see how this plays out…
I still don't understand ground in circuits, ive watched many videos. If I take a circuit diagram and try to build it, its very hard to understand where the grounds are supposed to go. They just end abruptly with the little symbol, how are you supposed to know where that connects to in reality?
yes, it is not obvious. you need to mentally think about all the current flows. it's like knowing where something is located on the map but finding the best way to get there the map doesn't tell you.
Are you gonna save these to a PCB play list?
no, not that many, just search 'PCB -' on my channel
Does KiCad warn about floating ground planes?
No. It will show a non-route trace, but that is very easy to not see and miss. and it does not show up when you do a error check. So, it is up to you. I like to click on a ground trace and then the ~/` key. this will highlight all grounds
@@IMSAIGuy ~Thanks ~ didnt work for me but I had to select and right click a trace and theres net inspection there where it has highlight net option.