Dual brushed ESC based on AtTiny1616

[Electro Dodo]

This is my second try to design a small dual brushed ESC(electronic speed controller).

The ESC takes 2 "drive" signals and 1 auxiliary signal for configuration and to drive inverted.

The PCB is a 20x20mm 4layer PCB.
For motor drivers I use two DRV8231. The expected current draw is around 500mA per motor and around 2.5A stall current.(two N20 motors per side)
The DRV8231 is rated for 3.7A and should be plenty.

The onboard LDO makes 5V for the AtTiny1616 and to give a 5V 100mA output to power a RC receiver.
I'm a bit unsure abut the thermal management on the PCB, as the LDO alone produces a lot of heat before the motor drivers even do anything.
For higher voltages I might need to use an external LDO for the receiver.

For programming I use the UDPI pin on the AtTiny1616 and 1mm pitch pogo pin clamps.
I'm not sure how to design  them properly, so I just used 0.8x0.8mm pads for now.
The 1616 might be a bit overkill for he application, but I'm programming with the Arduino IDE and the extra flash is quite cheap.

The top and bottom layer are signal, the second Layer is GND and the third layer is VCC.

Since people complained about me putting everything separate using net labels I tried to make a "complete" version of the schematic without using a lot of net labels. (Not sure if it look better tho)
I'm happy about all feedback and tips.

[Salitronic]

You are using a 500 mA LDO to drive 2×500 mA motors with 2.5 A stall. Even if you assume only one motor at 500mA, with the 24V max input (as stated on your silkscreen) you are burning ~10 W in the LDO to drive a 2.5W motor. That is not a thermal problem, it is simply the wrong topology.

You also have a minimal LDO footprint, which means you are essentially at the datasheet RθJA 100 °C/W ...you do the math. On a 20×20 mm PCB you cannot possibly get useful heatsinking for this.

You need to reconsider the power strategy.

[langwadt]

You are using a 500 mA LDO to drive 2×500 mA motors with 2.5 A stall.

Vm is the power input to the driver so the motors are driven from the Vcc not the 5V

[Salitronic]

You are using a 500 mA LDO to drive 2×500 mA motors with 2.5 A stall.

Vm is the power input to the driver so the motors are driven from the Vcc not the 5V

You're absolutely right, my bad, I misinterpreted it somehow.

However even assuming only the 100 mA receiver load and maybe another 10–20 mA from the board, the LDO topology used here is fundamentally wrong. If the intended input range truly goes up to 24 V, there is no way this can work. In a best-case scenario at 12 V, the regulator would be essentially operating at maximum junction temperature, with case temperature above 100 °C. In a more realistic scenario, the effective junction to ambient thermal resistance on such a small PCB will be worse than the datasheet condition, making the situation even worse. The DPAK footprint here is absolutely bare minimum not even thermal vias to the other side.

In practice, this design cannot realistically be run above 7 V input, and even that would already be hotter than what is generally considered good practice.

[Electro Dodo]

I used the LDO before on a similar sized PCB.
And it was able to output 100mA at 25V input, so I'm hoping it can supply 130mA on this PCB design

But I don't have a good way of knowing except trying it out, so that's the plan.
(I have no Idea how to calculate thermal performances of PCBs)

If the LDO overheats I just use a separate BEC for the receiver.
The LDO should have no problem handling just the MCU

[xvr]

I have no Idea how to calculate thermal performances of PCBs

The thermal performance of this PCB isn't very good. First, try adding thermal vias to the voltage regulator tab—you have a good copper substrate underneath the regulator, but bare FR4 has poor thermal conductivity, so you'll need vias to connect them.

[gspeed]

To begin with, why not use proper connectors, instead of bare solder pads?

[Electro Dodo]

To begin with, why not use proper connectors, instead of bare solder pads?

Proper connectors take up a lot of space and most are not particularly shock resistant.
Space/weight is the most important consideration for the ESC and connectors aren't good for that
I would have made the whole PCB bigger too if space wasn't such a priority

[langwadt]

To begin with, why not use proper connectors, instead of bare solder pads?

Proper connectors take up a lot of space and most are not particularly shock resistant.
Space/weight is the most important consideration for the ESC and connectors aren't good for that
I would have made the whole PCB bigger too if space wasn't such a priority

soldered wire suck with vibration, there's a reason crimps are used

[Electro Dodo]

To begin with, why not use proper connectors, instead of bare solder pads?

Proper connectors take up a lot of space and most are not particularly shock resistant.
Space/weight is the most important consideration for the ESC and connectors aren't good for that
I would have made the whole PCB bigger too if space wasn't such a priority

soldered wire suck with vibration, there's a reason crimps are used

That's what hot glue is for 

[Salitronic]

I used the LDO before on a similar sized PCB.
And it was able to output 100mA at 25V input, so I'm hoping it can supply 130mA on this PCB design

If the board was really similar in size, there’s no way you could dissipate 2W in that LDO and have everything keep working normally. I’m not saying you’re lying, don’t get me wrong, but either that 100mA wasn’t truly a constant 100mA (maybe only 100mA peaks with a much lower average) or, given the small enough thermal mass, the LDO could have been going in and out of thermal shutdown fast enough that it wasn’t noticeable functionally. I’m not sure what ventilation conditions you’re using, if you have good enough forced ventilation that changes the scenario completely, but in still air it would definitely be entering thermal shutdown.

But I don't have a good way of knowing except trying it out, so that's the plan.
(I have no Idea how to calculate thermal performances of PCBs)

I actually developed a tool that you might find useful, it is available (free, online) here: https://salitronic.com/gerber_analyzer/
I started this out a few years ago in my free time as a Gerber viewer but have since been adding analysis features. You can load your Gerber files (package Gerber and NC-drill files in a .zip) and then you can use the 'Thermal' mode, set the parameters and place a heat source anywhere on the board. The tool will then simulate the thermal performance until it reaches energy equilibrium. Please note that this is still a work-in-progress so you should take the actual temperature readings with a pinch (or maybe a bucket) of salt but in general it has been fairly accurate when compared to actual thermographic measurements and definitely if you want to do A/B comparison between different designs it should be useful. Anyway, if you find it useful do let me know, I'd appreciate feedback. Note that the Gerber files remain on your PC, it is fully client side processing - I don't get any data.

[Electro Dodo]

The PCB was definitely on the limit while dissipating that much power, but I expect to use an external BEC(battery elimination circuit) at higher voltages.
The wires soldered to the PCB might have helped with the thermals

The thermal simulation is really nice.
Definitely gonna use that in the future.
It would be quite interesting to be able to select a pad as a heat source, since the current heat source is quite small and the temperature gradient in the pad is quite big.(Not that I know how the real thing would behave)