very small STM32F405 board

Any other STM32 based boards
Ollie
Posts: 193
Joined: Thu Feb 25, 2016 7:27 pm

Re: very small STM32F405 board

Post by Ollie » Thu Jul 27, 2017 10:04 pm

The text on U4 chip is DE=A1D, which is also known as RT9193. In my applications, I use an external voltage regulator for all external sensors and actuators. The power of this local regulator has been enough in all my cases.

User avatar
Pito
Posts: 1592
Joined: Sat Mar 26, 2016 3:26 pm
Location: Rapa Nui

Re: very small STM32F405 board

Post by Pito » Thu Jul 27, 2017 10:30 pm

We discussed that regulator in an another thread, I did some review of it based on the datasheet. It is the same reg as the one on the BluePill, so be careful. I would not go over 100mA with it - that could be an issue provided the stm32F4 can take 80mA and an Sdcard easily 150mA. Based on the datasheet page 8-9 the 235mA is the absolute max current rating at 5V input (400mW), where the RT9193 chip will have 125degC at ambient temperature (it means you are cooling the pcb beneath it to 25degC).

http://www.stm32duino.com/viewtopic.php ... 390#p31985

That regulator may work up to 200-250mA with 3.6-3.8V input.
For example:
P = (3.8V -3.3V ) * 0.25A = 125mW, the chip temp at ambient temp will be:

T = 0.125W * 250degC/W + 25degC = 31.25degC + 25degC = 56degC (hmm, you will burn your finger)

The problem of the chip used is its package - the thermal resistance (250degC/W) is so high, that that chip
catches fire. The larger 1117 reg package on other boards gets something like 60degC/W so the chip would be 33degC (with those 125mW) at ambient temp.
Pukao Hats Cleaning Services Ltd.

ag123
Posts: 767
Joined: Thu Jul 21, 2016 4:24 pm

Re: very small STM32F405 board

Post by ag123 » Fri Jul 28, 2017 2:59 pm

looking at the rt9193 datasheet
http://www.richtek.com/assets/product_f ... 193-16.pdf
it would seem to me that its attractiveness is the 220mv dropout voltage, this would make it feasible to feed a 4.2v LiPO battery into the 5v lines, and there is sufficient headroom 4.2 - 0.22 ~ 3.9v to keep running off a single charge till around 2v. a lm1117 or ams1117 has a 1v dropout which would be ok for 5v but with 4.2v that leaves 3.2v at the stm32 even when fully charged

but i'd guess f4 being more power hungry vs f103, i'm not sure if we could or if it make sense to swap the on board LDO for a ams1117 or lm1117 when running off 5v

in addition i'm not sure if there is after all a feasible way to cool a sot23 ldo with a heatsink etc
:?

victor_pv
Posts: 1681
Joined: Mon Apr 27, 2015 12:12 pm

Re: very small STM32F405 board

Post by victor_pv » Fri Jul 28, 2017 3:41 pm

Pito wrote:
Thu Jul 27, 2017 10:30 pm

The problem of the chip used is its package - the thermal resistance (250degC/W) is so high, that that chip
catches fire. The larger 1117 reg package on other boards gets something like 60degC/W so the chip would be 33degC (with those 125mW) at ambient temp.
Thanks for the warning, I'll feed it with 3.3V from another regulator.

User avatar
Pito
Posts: 1592
Joined: Sat Mar 26, 2016 3:26 pm
Location: Rapa Nui

Re: very small STM32F405 board

Post by Pito » Fri Jul 28, 2017 4:09 pm

Below the tables to visualize the issue. Mind the Temperature of the chip inside the regulator's package at Ambient temp (25C) means you are keeping the surrounding temp at 25C - that is usually not the case, thus the T_chip must be increased accordingly. The 1117's SOT-223 has the thermal resistance 66-120 based on the pcb copper layout.
Note: below calculations are simplified calculations. The precise modeling usually requires to consider about 6 thermal resistances (serial/parallel) placed along the thermal path from the chip up to heatsink's fins, material properties, airflow and geometry.
Voltage Regulators Imax and Temp.JPG
Voltage Regulators Imax and Temp.JPG (101.22 KiB) Viewed 198 times
Voltage Regulators Imax and Temp 800mA.JPG
Voltage Regulators Imax and Temp 800mA.JPG (61.59 KiB) Viewed 196 times
Voltage Regulators Imax and Temp 42.JPG
Voltage Regulators Imax and Temp 42.JPG (94.71 KiB) Viewed 196 times
Last edited by Pito on Fri Jul 28, 2017 5:08 pm, edited 6 times in total.
Pukao Hats Cleaning Services Ltd.

ag123
Posts: 767
Joined: Thu Jul 21, 2016 4:24 pm

Re: very small STM32F405 board

Post by ag123 » Fri Jul 28, 2017 4:14 pm

i'm postulating about doing some experiments, e.g. get some of these thermistors
https://www.ebay.com/sch/i.html?_from=R ... r&_sacat=0
stick that with some thermal compound to the sot23 LDOs, then perhaps run some benchmarks or some heavy gpio processing on stm32f4 (it may be even more fun if it is overclocked)
that'd give a feel of just how much heat gets generated (at least the termperature), but then the thermistor leads would likely conduct away some heat as well
:lol:

User avatar
Pito
Posts: 1592
Joined: Sat Mar 26, 2016 3:26 pm
Location: Rapa Nui

Re: very small STM32F405 board

Post by Pito » Fri Jul 28, 2017 4:21 pm

Voltage Regulators Imax and Temp 36.JPG
Voltage Regulators Imax and Temp 36.JPG (51.4 KiB) Viewed 193 times
Pukao Hats Cleaning Services Ltd.

ag123
Posts: 767
Joined: Thu Jul 21, 2016 4:24 pm

Re: very small STM32F405 board

Post by ag123 » Fri Jul 28, 2017 4:40 pm

i'm taking a look at wikipedia
https://en.wikipedia.org/wiki/Thermal_resistance
From Fourier's Law for heat conduction, the following equation can be derived, and is valid as long as all of the parameters (x and k) are constant throughout the sample.

R = x / ( A * k )

where:
R is the absolute thermal resistance (across the length of the material) (K/W)
x is the length of the material (measured on a path parallel to the heat flow) (m)
k is the thermal conductivity of the material (W/(K·m))
A is the cross-sectional area (perpendicular to the path of heat flow) (m2)
this would seem to imply that the rather high thermal resistance is partly due to the small size, hence small A, assuming that there isn't any way to influence x and k.

this may somehow mean that if there is some ways to 'heatsink' that sot23 LDO, we may literally reduce the thermal resistance (and possibly significantly), this could for instance be possible by using some heatsink thermal compounds, but it would also mean attaching a heatsink so that all that heat is conducted away and dissipated over a (much) larger area.

the obvious problem is that it'd seem quite difficult to heatsink that sot23 LDO, as it would seem otherwise the LDO may possibly be able to handle higher currents than specs suggests. but on top of thermal problems, the MOSFET transistors used in the LDO may have actual electrical limits e.g. the minimum on resistance at the current voltages, that'd probably mean that beyond 300mA, the voltage dropout would be limited by the MOSFET on resistance and loses its ability to supply the desired voltages e.g.< 3.3v

User avatar
Pito
Posts: 1592
Joined: Sat Mar 26, 2016 3:26 pm
Location: Rapa Nui

Re: very small STM32F405 board

Post by Pito » Fri Jul 28, 2017 5:23 pm

The thermal shutdown of the RT9193 is at 165C, with 30C hysteresis. That may happen by 200-250mA (5V input) as the pcb and copper heats up as well (and quickly), up to say 60C, thus the chip will be over 165C and it shuts down. When it cools down by 30C it goes on again. So it may even oscillate..
Pukao Hats Cleaning Services Ltd.

User avatar
Pito
Posts: 1592
Joined: Sat Mar 26, 2016 3:26 pm
Location: Rapa Nui

Re: very small STM32F405 board

Post by Pito » Fri Jul 28, 2017 5:42 pm

Majenko at chipkit started to experiment with this switcher:
http://www.ti.com/lit/ds/symlink/tps82130.pdf

Up to 17V input, 3A output, 2MHz, 90% eff at 1A, with integrated inductor, 3.0x2.8mm package.
Maybe it could be a good replacement.. :)
Pukao Hats Cleaning Services Ltd.

Post Reply