@superslot, part of the reason why Particle went from TI to Broadcom was the former’s very slow or non-existing response to requests for change or bug fixes. You have to admit that your use case is not typical IoT in that you are essentially streaming data with half-second pauses with the WiFi on running at near 100% duty I suspect. As for a disclaimer, I disagree in that there are LOTS of Cores out there running without these issues and primarily because of HOW they are being used. I understand your frustration nonetheless.
It would be great to see the same application running on the Photon to see how it does.
sure. not a proper schematic but I think it is clear enough … it’s trivial after all.
note1: all the power comes from the pololu regulator,
note2: on the led bar there is never more than one led “on”
note3: all display are on minimum brightness, I do check the total current and I am below 250mA, where 180mA are only for wifi module
I basically count the number of interrupts from the ir sensor and every 0.5s I send a MQTT message to the server and update displays (total count and “speed” of counter)
@superslot, after discussing with @BDub, here are some ideas to try out:
Put a 10uF to 47uF electrolytic capacitor on the output of the Pololu regulator. There could be a weird oscillating thing going with the power.
Fire up a Core with NOTHING attached powered via USB and toggle D7 high with Tinker. Soak test, see if it’s hot. If that’s ok, same test just powered from Pololu SMPS. Then add other parts. Then run the Display/NeoPixel app.
mmm … I already did the test where I power up the unit by USB (removing the Pololu)
same issue there.
I power up everything (including display) and I shut off the wifi unit only (still counting interrupts),
everything is cold as ice.
@peekay123 … the issue is the wifi unit, it is going to overheat if you keep it “on” all the time (well … 46C)
I even tried a “barebone” setup with ONLY the Particle.Core sending “random” MQTT messages every 0.5s and it does overheat… (no display attached at all)
Coming from the chip-design world, I don’t really feel like 46C is “hot”. That translates to 114F for those of us raised on F. That is slightly warmer than body temperature but cooler than the outside of a tea or coffee mug with a hot beverage inside. Your measured current of 250mA presumably at 5V is 1.25 W which is very consistent with the temperature you seeing.
The TI part can tolerate up to 70C which is generally too hot to keep your hand on and about the temperature of the hottest hot water at the kitchen sink in most houses. The TI WiFi chip is certainly not overheating.
I think you have also heard the message the transmit power is controlled completely inside the TI part with no way for users of the chip to change it. There really isn’t anything you can control there except moving the device close the WiFi access point so it can automatically lower its output power.
So what can you do? Well you do have a few things that are in your control.
You are using the on-board regulator on the Core to convert 5V down to 3.3V. That is adding heat to your project and you could reduce overall heat by redesigning your project to supply 3.3V to your little box instead. Maybe this is practical for you, maybe not: I can’t say. Perhaps a dual output (5V and 3.3V) wall supply could be used.
You have several peripherals, but the only concerning one is the strip of WS2812 Neopixels, which can also dissipate some power when on. I know you said that only one is used at time, but these are up to around 60mA per WS2812, so 300mW just for that one at full brightness.
Keeping any high dissipation devices (like the 9V to 5V regulator) outside of the little sealed box with help with removing heat.
Adding airflow to the box so that the heat can naturally chimney out will help too. A sealed box is hot box. A slot or holes at the lower back and holes on the top of the box will allow the heat to naturally rise out of the box due to convection. Have a look at any consumer electronic product enclosure for other ideas how air flow slots can be hidden into molded plastic enclosures.
Alternatively, you could make the entire rear or bottom of the enclosure out of aluminum to act as a larger heat sink. This will spread the heat over a larger area reducing the apparent heating. A lot of consumer items like AppleTV and Roku boxes have thick aluminum bottom panels which act as heatsinks.
ok guys, this thread is long&clear enough to be closed.
looks like I’m the only one feeling the Particle.Core is not really the best for IoT product that do involve sealed enclosures (btw: my prj is exposed to rain and sun, it’s an outdoor box) or surfaces that can be touched by people (and especially kids).
it does not matter what we (engineers) decide is “hot” or “overheating”, if the end users tell you “that feels too much latency” you have to speed up things, if users tell you “ouch, that is super hot!!” you need to be worry about it.
I understand your position bko, I simply think that this “heat” thing should have been made more clear on the product specifications (and I’m one of the kickstarters… both core and photons …). A disclaimer like your: “add a generous heat sink for continuous wifi connections”.
