I have been designing my own PCB. First attempt, but going OK.
I have a couple of questions however.
Ground plane. Is this a necessity? My board is pretty compact with traces, and it doesn't make any difference to the cost of the PCB being made.
When I flood the board with a plane, I find it much harder to check for errors or make alterations (Although the flood would be the last step).
My +5v, +3.3v and ground traces are larger than the signal lines anyway. It's not a noisy circuit.
Suppression. My 3.3v regulator has a 100uf on the 5v input and 10uf + 100nf on the output (as the datasheet specified).
As a default, I have been placing an SMD 100nf cap across the supply pins of all my IC's as close as possible.
I have a couple of PIC IC;s, an SD controller, couple of 24LC256's, couple of logic IC's etc.
This means I have somewhere in the region of 10/12 100nf's across the 3.3v supply rail across the board. Is that OK? Can you have 'too much' suppression?
The 100uf is on the input before the 3.3v. 10uf is on the output (I can always change these).
I have a 100nf on both the input and output of the 3.3v reg (as close as I can get them).
Current... not calculated that yet exactly, but certainly no more than 450mA on the 3.3v rail is expected. My 3.3v supply is a LM1117MP-3.3 which is rated at 800mA.
I may upgrade that slightly
The question regarding the ground plane was really if it was critical. I am using Diptrace, and if I get the nets wrong, the ground plane will connect pins that should not be connected (I don't think I have... but it's possible).
phoneystark2020:
My 3.3v supply is a LM1117MP-3.3 which is rated at 800mA.
Only if it has a heatsink. Does it?
phoneystark2020:
The question regarding the ground plane was really if it was critical. I am using Diptrace, and if I get the nets wrong, the ground plane will connect pins that should not be connected (I don't think I have... but it's possible).
Only if you get the nets wrong, in which case you have major trouble for sure.
I have to say, the nets drive me mad.
I have found that importing some components from the library has the pads incorrectly linked, which messes up your nets.
I have been ungrouping these components and then re-grouping (but not as a component) to get over this issue.
Trouble is, that no longer shows the pads/item as a component in the components listing.
No... no heatsink. How do you heatsink an SMD SOT223 device?
I have upped the spec of the 3.3v to a 1.2A rated device. Quick calc of my possible 3.3v loading is actually 310mA.
I have never used Diptrace so I don't know anything about it. I have been an Altium man for decades but recently downloaded KiCAd for my Linux machine and for a freebee it is fantastic and reasonably easy to use. Placing your GND planes last it will automatically create it, not covering the wrong tracks.
I have checked all my pcb tracks once, but there have been changes, so I need to do it again.
Yes, ground pour will be last, once I am 100% sure that all my traces are correct.
Diptrace. I do like it. BUT.... This really annoying net thing means that when you place a component, and then decide to maybe re-route a trace to a different pin, it still leaves the pin you disconnected associated with the original net (if that makes sense).
If you 'delete net' on that pin, or 'disconnect net'..... it deletes all the damn net connections, which can be quite a lot of track.
I am sure it's operator error, but my work around is to ungroup the component, delete that associated net pad and then replace it with a fresh 'un-linked' pad. Then re-group the component (but not grouped as a new component - as that links all the pads!!!).
I think maybe it's heavily designed around using their schematic to PCB conversion. I am not using that, as I don't have my project as a schematic in Diptrace.
Anyway.... fills the evenings.
Plus, £8.49 for 5 pcbs.... not the end of the world if they are wrong. Just need to test all the pins on the new pcbs first before committing to components.
The tab is 3.3v out on this device. Linked to the centre pin.
How is having a very slightly higher rated device silly? I know my current draw, the device is the same footprint, and I have used the groundplane as a sink. Surely not much more I can do. I am not pushing the limits.
The supply is from a MicroUSB (from a laptop). All works fine on the breadboard demo layout.
phoneystark2020:
How is having a very slightly higher rated device silly? I know my current draw, the device is the same footprint, and I have used the groundplane as a sink. Surely not much more I can do. I am not pushing the limits.
If you're responding to reply #9, I think the explanation there is that it is futile, that it won't gain you any margin vs. a lesser rated part. Using it, suggests that you might believe otherwise.
No. I have gone back to my original part. That is what I have used on my prototype board (mounted on a breakout DIP board), and it doesn't even get warm.
I think all will be fine.
The tab of an LM1117MP-3.3 is indeed Vout.
It should have a small flood-filled area of the board connected to it.
(5volt - 3.3volt) * 0.31A is ~0.5watt, and that will get an not cooled smd regulator hot.
Leo..
I have gone through and redesigned a lot of components, and changed them to more efficient versions.
Got the current down to 60mA I should be fine.
BUT.... I have a question regarding this nightmare I am having with 'NETS' in Diptrace.
If I run verify, it throws up errors saying that my net 0 (My ground rail) has 5 errors in it (as in, no continuity).
But, I have gone over the PCB and I cannot find these errors anywhere. They all seem to be connected as they should.
The PCB when exported as Gerber looks correct. The 'net' has caused me a lot of problems when it doesn't disconnect from a part when you change your routing.
The PCB will be manufactured from the Gerber footprint I assume (the traces layer), not the net information. Think I will just have the boards made and then test them.
I have cleared the errors, even though they were not technically errors.
I removed the ground plane for now, and I am considering not putting it back. The board is pretty compact and with a LOT of traces. I am not sure I really need to flood the board with a ground plane.
My Ground and +3.3v rails are considerably larger than the signal traces.
I will see what I think after another check through the circuit
I should be fine.