For now I have still some issues, like the 50ohm trace for antenna should be 1-2mm wide (only .5mm here resulting an approximate 75ohm impedance -> .2 reflection coeff), also the fuel gauge eats a lot of power since it has a coulomb counter (might choose another ic just to measure voltage and with lower consumption) and adding silkscreen infos
#BLE Temperature & Humidity Sensor
hallow yacht
static peak
can you upload schematic with better quality? PDF maybe?
static peak
CH340K (U3) operates at 5V. Can STM32 tolerate such logic levels on UART?
hallow yacht
Ohhh i think you're right. I checked that 3.3V was ok for the ch340 but not the other way around! Thx for pointing that out
Might add a level shifter
or just resistor divider might be cheaper and good enough for this application
static peak
For U6, why is VINA / EN connected to VIN?
hallow yacht
Well by reading datasheet 1 is for enable
so I think they might have the wrong schematic bc it doesn't make sense
Maybe i'll just add 0R just in case
static peak
there is internal connection, you should not connect it on your board
hallow yacht
Yeah i think you're right, it might just act as a LP filter to delay the EN/PS pins
static peak
what is the purpose of D2 (Schottky diode) ?
hallow yacht
It's an artefact from previous design, totally useless here
It's the part of the circuit i am the least sure about but since it's only an edge case (connecting the battery while usb is connected to disable tp4054) i din't put too much thoughts into it as it may be totally useless
static peak
I would suggest to remove this diode. Transistor gate can be in high impedance state, sensitive to noise, and it might enable the transistor by accident.
and now the layout. GND connection for STM is very bad. Look at pin 32. It has very poor connection to boards GND trough the center pad. You need to put GND via glose to pin 32.
Pin 22 also needs GND via
hallow yacht
Do you think it's a good idea to fill the gnd thermal relief pad with vias ?
static peak
do you have capped vias? Soldering is the main concern
hallow yacht
I thought that .2mm vias might be small enough to not let solder flow through it
I should read smthg on that
static peak
But I think you can put some vias there. Just expect potential solder and flux contamination on opposite side. But it should be fine.
But personnaly I would try to also put GND vias outside the package, close to pins.
hallow yacht
And if it does flow it'll just increase the thermal couple between gnd plane and the stm32
yeah it looked nice but I should move thoses caps
static peak
maybe slightly, but worth it in my opinion
hallow yacht
The only thing where i'm concerned about this layout is the 3.3v under the rf line
static peak
yes this also would be nice to avoid.
hallow yacht
Even though it's perpendicular and might not catch so much field I should try to route it differently
static peak
just go on blue layer, under the IC
hallow yacht
You're helping me so much :D thx
static peak
and if you do not have capped vias, also remove vias from U4 pads, not worth the risk or soldering issues
and what is your via pad and hole size? it looks small
hallow yacht
.35/.2 (diameter/size)
basically near the minimum size of jlcpcb capabilities
which is .25/.15 I think
static peak
ok, sounds quite hardcore for me, but if they say they can do it then it is ok
hallow yacht
which size are you usually using ?
(I just have a pair of .6/.4 for the Vbus to handle 500mA)
static peak
0.2mm/0.5mm, or even 0.3mm/0.7mm. But it all depends on your manufacturer and expected quality 😄
can't you also rotate this filter?
hallow yacht
Yes but I think i'll redo all the RF bc these traces are way to thin
I'm calculating about 75ohms with this circuit, for 2 layers i should have 1-2mm
static peak
and in my opnion you should not worry too much about 50 ohm char. impedance. Your connection is very short.
Anyway, when you make the calculation, make sure to include the co-planar effect of GND fill.
hallow yacht
Yeah but still i'm 2-3x too thin
But on the other hand I only get 20% reflection which seems acceptable for this kinf of project
It's the first time i'm using the curve tool, i should have better curves at the pi junction also
static peak
for 2.4Ghz this is not critical. Just avoid 90 degrees bends. That's why I suggested to rotate this filter 45 degrees
how do you calculate your characteristic impedance?
static peak
so with 0.5mm you will get ~57 ohm, in my opinion (guessing) this is good enough. Your trace is very short
just put more GND vias closer to trace
static peak
it is probably OK, but I would still add more vias to exposed pad of STM32 (e.g. 4x4 pattern). And personally I would remove vias from pads. But this is jus to be safe
cold pelican
theres a plugin for stitching vias on kicad if you want a shortcut
hallow yacht
I use one from an RF pluging I think
I just add after some of them manually
For now i'm trying to setup a workflow from https://github.com/nguyen-v/KDT_Hierarchical_KiBot to get better docs/schematic
