Do you know what the difference between a bypass cap and a decoupling cap is, and do I need both?
My take is that is two different words for essentially the same thing. Donāt worry about that!
What do you mean by stiffer ground connection? Currently, I have a 0.75mm clearance, I didnāt want to make the clearance too small.
I just had a look at the board I am working on. My clearance is 0.5 mm and that is quite large=conservative for low voltages (voltages under 24V for example). I do not know much about your design, but the large clearance means that the rows of pins for a DIP cause a slit in the ground. In a very critical design this would reduce the effectiveness of your ground plane.
I like big clearance, but even 0.35 mm should not be a problem to work with for low voltages.
I cannot be certain, but I suspect that because your board is mostly through hole (not SMT) that the ICs are older and none of the signals or switching is VERY fast. FWIW any time a forum member is doing a new design and buying the components, I strongly recommend going with surface mount. It is not difficult (and is easier in some ways) so long as you are not using tiny lead pitch devices such as 0.5 mm or smaller. Of course if you have the through hole parts already, you might as well use them. I do that also.
Anyway I think it is unlikely that those slits I observe and mention above are likely to cause problems. That would seem like a more critical design than most of what I think is discussed on this forum.
Iām using 5V and 15V in my board. And Iām using THT because Iāve never hand soldered SMT before, I also have all the components already. Iāll probably reduce the clearance to 0.4 or 0.5mm
See this one:
#11/36
Older ICs but manufactured recently can be much faster than maximum times specified in their datasheets simply because competition forces cheaper production (die shrinking).
We ran into this problem 15 years ago when some our device designed in 90s refused to work when made with recently bought (the same type and manufacturer) serial EEPROM.
Since many years I have to use 0.2mm for whole PCB. As there are some places at PCB with 0.2mm clearance than for zones I use little bigger - 0.25mm. In designs I made this year I had to use 0.18mm whole PCB clearance. It is because of 0.4mm raster QFN element. Datasheets specifies 0.22m pads so distance between them is 0.18mm.
Around 1990 I switched from THT to THT at top and 1206 R and C at bottom and few years later to all SMD with 0805 as typical size and shortly to 0603. 25 years later 0603 is all the time the basic size for me. There is not problem with soldering them.
I have no argument with that for professional boards. Mainly I am discussing boards that I assemble by hand in my home lab. Before starting with KiCad I did many boards with ExpressPCB, and many of those had no solder mask.
I was sort of a chief engineer at a power supply manufacturer during 1981-1998. I was responsible for the first boards that used SMT, and yes I had us start with 1206ās and SOT23ās. The board assembly replaced a previous design which used 1/8 Watt leaded resistors all standing on end. But these days I am completely comfortable soldering 0603s; I use a somewhat small 0805 footprint for those 0603s.
0805_Top_Small.kicad_mod (1.5 KB)
Great effort!
Iāve designed an Arduino 328 for a project. The only issue I had that really had me scratching my head was the reset circuit. I ended up using the exact design Arduino use and no more troubles after that.
You donāt look like you are using the UART for uploading codeā¦which is fine as Iām guessing youāll preprogram the Atmega.
I would stress too much about the decoupling caps. I use 100nF for them if I think itās going to be an issue.
If a ground current has to flow from somwhere left of your controller to somewhere right of it, it now must flow around it. If there ground fill was connected between the pins it could flow straight under it. That would give smaller loop area and thus lower EMI.
On a double sided board, I would always use a copper fill on the front. In most cases a copper fill for +5V on the bottom. Vias donāt cost extra, so you can stich a plane back together when it got separated.
For your first board, the routing and parts placement look very good!
Nick
I appreciate everyoneās feedback so far! This is the LED array I plan on using. It consists of 57 WS2812E-V5 LEDs. There is a ground pour on the back and I am using vias to connect the ground of each LED to the ground layer. If I could get some feedback thatād be great.
LED_v5 Schematic.pdf (193.0 KB)
I have never designed anything like that so Iām not sure about what I am saying.
As it is digital transmission between LEDs shouldnāt there be some capacitors?
I donāt know the current. May be it would be better to have 5V fill zone at top to reduce voltage drop at tracks.
I know that for a lot of these types of LEDs, capacitors are typically used. The datasheet for this LED is in a different language, but thereās a schematic and it doesnāt show capacitors.
WS2812E-V5.pdf (779.7 KB)
I found a datasheet in English for WS2812B-V5 which seems to be similar to what Iām using (WS2812E-V5), and the datasheet says
āThe peripheral circuit donāt need to add filter capacitorā
WS2812B-V5_V1.0_EN.pdf (640.2 KB)
I could be wrong, but it seems that for these LEDs, you donāt need capacitors. They may already be built in, Iām not too sure.
The max current draw is going to be about 2A. Having a 5V fill on top wouldnāt be a bad idea, I may add that.
Yes, sounds like a good idea and could also help with heat dissipation. Or at least some more direct connections like between D37 and D38 or between D7 and D18 would be probably good. The current goes a really roundabout way otherwise. But adding a fill is probably the simplest solution.
I would move the crystal and associated caps closer to the CPU. You might also consider a guard ring around the crystal and caps. Here is one reference.
Regarding your original design, I wouldnāt put electrolytic capacitors on the underside of the board. I donāt know how you intent to store or mount the boards, but it looks really awkward. For simple decoupling, Iād probably use small ceramic smd capacitors instead, which also have a lower ESR. If you really need large capacitance, thereās enough free space at the top side of the board.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.