Beginner pcb layout review

[electriclawnmower]

Please review my layout for the bq25703a buck-boost charge controller. And what would be the best way to improve heat dissipation? 

[Warhawk]

I must say the layout looks reasonable.

[Warhawk]

Do you have pull-up resistors for the i2c interface somewhere?
Also, the filter around the current sensing resistor is bit weird. I checked the DS and you have it as per instructions. Still, I would add at least 0-ohm resistors between the cap and the shunt. Look at figure 56 of the lm5170-q1
https://www.ti.com/lit/ds/symlink/lm5170-q1.pdf
This comes from the old National guys. They know they craft.
The idea is that you compensate the parasitic inductance of the shunt.
Attached is a picture of the low-side current-sense in a flyback circuit.
R1, R2, CH1 are different configurations. If my notes are right the last one is 15nF X7R 50V and 1R5 resistor.
the initial peak is ~1.5A which exactly corresponds to the driver strength of the IC. I used a high-speed, low-capacitance active probe for the measurements.



"Connect0.1-μF ceramic capacitor from SRP to SRN to provide differential mode filtering"
"Connect SRP pin with optional 0.1-uF ceramic capacitor to GND for common-mode filtering."

I have a  feeling that this is what the TI engineer wrote when he/she felt that a cap would help but have no clue why. Throwing 100nF caps at the problem is quite common these days.

Hmm, I take this back. Now I see that this does not sense the current through the inductor but rather the DC current to the battery.

[electriclawnmower]

Thanks, i will add those resistors. Anything that can save me from cutting traces later is nice . Also i noticed now that i forgot to add the required capacitors parallel with the battery so that's one head scratch less. The i2c pull up resistors are located in a different part of the board. This board has also a bluetooth receiver and some analog audio circuitry so hopefully this charger won't cause any interference with them.

[javascripter]

would agree that what I can see of the layout looks decent

for thermal dissipation you could use heatsink but at least digikey doesn't have any that small; if you really needed it I guess you would have to make your own (could also put on opposite side since you've
 done the via farming fine, that way you don't worry so much about the size of the component, but it is of course less effective); but I probably wouldn't bother. Another thing is to increase the area of the copper fill; some manufacturers will specify an area that the datasheet thermal resistance is provided form, but it seems TI doesn't. It looks like they use JEDED standards, but I don't have access to them, so I've got no idea what area they would expect. Also dropping the switching frequency would decrease heat generation, but also requires larger components or increased ripple current, and it looks like you only have two options.

for the intereference issues I wouldn't expect it to affect bluetooth at all given bluetooth is very many harmonics into the switching frequency. For the analog stuff it depends a lot on the analog stuff, but if you are doing things at typical hobbyist level and have a decent layout I wouldn't expect any problems.

Not actually sure how reliable this would be, but if you are really worried maybe temporarily set it up so sampling frequency is for sure not a multiple or divisor of the switching frequency and see how much the result changes between samples. It will probably change regardless of the dcdc unless you put it in a pretty nice metal box, but if there is too much interference/noise I guess you could hack on linear regulator (and disable the switching regulator temporarily) too check if the problem is the regulator or insufficient reduction of intereference from other things.

Another related thing is the possible series I2C resistors (very small ones) to limit inrush currents; but if you add them be aware even fairly small values can stop the i2c from working and I probably just wouldn't use them here unless you have really sensetive analog circuitry (and then you have lots of other things to consider).

[electriclawnmower]

Thanks, that's useful to know Somehow we managed to get first sounds out from this thing and charge all batteries with desired current.