Electronics > Projects, Designs, and Technical Stuff
SMD prototyping techniques -- looking for advice
techman-001:
--- Quote from: mark03 on July 29, 2019, 06:25:36 pm ---
--- Quote from: techman-001 on July 28, 2019, 02:06:42 am ---This is a deadbug technique I'm experimenting with. I have more pictures and details on my doc page at:
https://mecrisp-stellaris-folkdoc.sourceforge.io/prototyping.html#prototyping
--- End quote ---
Tempted to draw an analogy between this and programming in Forth, but I won't >:D
Seriously, that is impressive. You have way more patience than I.
On the main topic, I may be lucky, but I almost always prototype with a custom PCB and some common-sense measures like extra pads, etc. I can only think of one "coaster" I've made out of a couple dozen boards. It's the rare case that can't be reworked with some bodge wires and creativity.
--- End quote ---
Hahahah, some real Forth humor at last :)
It does require a 20 - 40 x microscope and steady hands ... but not that steady as I'm 65 years old.
This is a experiment leading into much more interesting possibilities. For instance, BGA MCU's can be very cheap indeed compared to the leaded versions, in fact I paid $0.45 each for the STM32L053 chips in that picture. They have the memory below plus all the M0+ low power peripherals :
64 KB Flash memory with ECC
8KB RAM
2 KB of data EEPROM with ECC
20-byte backup register
Just because a tiny BGA chip has 64 pins doesn't mean one has to use or even connect to them all ;-)
Imagine if one only needed 4 I/O's for a miniature project ?
nigelwright7557:
I design usb scopes as a hobby for the development experience.
My first design was 460K samples per second and all through hole.
Once I got up to 2MHz samples I found I needed to use SMD TQFP 0.5mm pitch PIC microcontrollers but I kept the rest through hole for ease of building.
My first attempt wasn't good as I bent a couple PIC pins being to rough and had to bin that attempt.
On the next attempt I was more careful and got the PIC lined up as spot on as I could then applied some solder.
I then put plenty of solder on the other 3 sides, way too much really.
I then found with plenty of liquid flux and a copper braid I could remove most of the solder. I found the trick was t ohold braid on the pins and move soldering along the braid so it didn't bend pins.
I put the PIC on its own small PCB that had through hole connectors to the motherboard.
Through the connector pads I could buzz the pins out for adjacent shorts.
In fact in some cases a short didn't matter as both pins were inputs and not used.
I have built 3 pcb's now and they work great.
ludzinc:
Oh god that board is atrocious!
ebastler:
@ludzinc: Your comments comes across as judgmental and derogatory. It is also quite unhelpful since you don't say what is wrong with nigelwright's layout.
Personally, I would route the tracks symmetrically and at 45° and 90° angles, but really that's mostly to satisfy my OCD. I am pretty sure the electrons will happily flow in arbitrary directions. ;) And I would probably use SMD caps and quartz and place them on the bottom of the PCB, which avoids interference with the routing on the top and gets them closer to the processor chip. But since the adapter is meant to go onto a breadboard with further (longish?) external wires, we can assume that it is meant to work at clock rates where the exact placement of the caps is not critical.
Are there other aspects of the layout one shoud be concerned about?
KL27x:
That wouldn't work very well on a breadboard. Gotta do SIP or DIP for breadboard. Quad pack, no good. this board is obviously for:
--- Quote ---I put the PIC on its own small PCB that had through hole connectors to the motherboard.
--- End quote ---
The only obvious things with this board, to me, are lack of ground plane and the trace width and routing. Trace width probably a mil or 2 smaller than necessary. And the arbitrary angles do not help with the PCB etch. The size of the gaps in the copper should not get too large compared to the trace size. I try to keep the gaps around a small trace to be roughly same thickness as the trace to maybe only 50% thicker. An easy way to fix this is to just use a top pour and set the gap/clearance as desired.
Of course if you are only ordering a small amount, it doesn't matter. You pay based on a formula that doesn't include this stuff. The fab will obviously make extras and throw away the excess, including the ones that fail. At some point of high volume order, your fab might actually make suggestions to increase the yield. Tiny traces with large empty space around them tend to break when you etch the board.
This might seem totally counterintuitive. You might think adding the top pour would near double the areas that could bridge and that the larger the clearances the better. But it's just not so. The areas right next to traces/pads is etched away, first, due to turbulence from the resist. Large empty spaces take the longest time to etch away, and during this time while waiting for them to clear, the traces are undercutting themselves.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version