Hi there, has anyone run the Arduino R4 Minima from 24V? Noticed that the regulator chip gets really hot as the voltage is increased and worried that it would overheat if powering a serial shield and an Ethernet shield too. Any thoughts?
Thanks,
Brian
24V is the maximum allowable voltage. I think the recommended voltage is lower. Let's see what the other users say.
It will run cooler if the voltage is lower. Also the more load you put on the Arduino the hotter the regulator will get. I have seen some desolder themselves. I would not run it above 12V.
Thanks both. Turns out that I have seen the regulator desolder itself too now - four times!
This post has the same experience: Power supply hot - #2 by liaifat85
I don't really understand this, saying it's not a power supply, yet there's no alternate for powering shields. The other thing its the regulator is supposed to shut down if too hot, so what went wrong that it got so hot to desolder itself but didn't shut down?
I have put a 12V power supply in now and they are running much cooler, so hopefully that's the last it. Pity none of this is mentioned in the docs - something like 24V with no shields/extra load, 12V with
Brian
You need a DC-DC step down converter.
Indeed, it isn't. You've found out why.
Of course there is; many suitable external power supplies are available that can connect directly to a board's 5V line, or you could use a step-down converter to go from e.g. 24V to 7.5V or so and then feed that into the Vin of your board.
Thermal shutdown is not a 100% reliable mechanism. It's certainly not something one should rely on in normal operation; it's a last line of defense to protect about high-impact failure such as explosive self-disassembly.
It's not mentioned there most likely because it's impossible to tell what kind of 'shields' people will connect and specifically what the power requirements of those shields are. You might be able to connect a couple sensor shields and still run from 24V fine-ish.
It's a bit like expecting the manual of your car to state that yes, it'll go 200mph, but only on a straight road with no bumps. In reality, you only drive as fast as local regulations and road + traffic conditions allow it. Or, perhaps, more aptly: if the car's specs say it'll go 200mph, evidently it may not quite reach that speed while hauling a trailer. It's the same with electronics; you design a system for the specific set of conditions it'll be used in.
1 Like
Please elaborate on this
I mentioned above that I am using an ethernet shield and a serial shield. The only 5V input to these that I can see is on the pins - maybe I missed something. There is certainly nothing in the manuals for these about external power supplies
True, but I expected better than 0%
Exactly why it should be mentioned. Easy enough to say 24V with maximum 50mA (or whatever) additional load
Brian
While you CAN power the board from 24V it doesnt mean tha is a great idea.
Lets look at the voltage regulator
ISL854102 Datasheet (renesas.com)
The converter efficiency is lower for 24V suppy, around 85%.
Note the thermal resistance - 42C per watt; so if your circuit is using power you can expect that regulator to get HOT.
So use the pins.
Easy enough to work that out if you know that 24-5 = 19V * 0.050 = 0.95W which is way too much for a tiny SMD component with no heatsink.
Look, I see where you're getting at, but don't confuse an Arduino or any electronics kit/board with a microwave oven or smartphone. It's not a consumer product in a similar sense. It's a component/subsystem and it's up to you to integrate it into an application. Part of that exercise is to be aware (and thus, to learn) of some of the basics of electrical engineering.
If you expect everything you're going to run into to "be in the manual", you're going to have a disappointing experience. If you are willing to learn and handle every challenge by learning about what you run into, you'll (1) have a very educational experience and (2) you'll recognize at some point that the expectation of a complete manual was kind of silly.
Also, if you have an board/module/etc that runs very hot, something's probably not right. So unplug it and investigate. It doesn't make much sense to keep operating it that way because "there was nothing in the manual about this".
1 Like
They are inaccessible unless I start hacking things off, or bending pins out
SMPS, not linear
The problem is that the basics are missing - there is no spec for the 5V pins on the Arduino board, but you need this to power the shields. I'll see if I can get that added
They were thousands of miles away at the time, so hard to tell why they were failing. Duplicated the setup here and found the problem
Brian
Yes, thanks - I'm getting that feeling too
I'm not even sure that would supply the board on its own without overheating, never mind a couple of shields - even serial and ethernet that I am using
Running on 12V as opposed to 24V, the power drops by 240mW, and all of that was going into the SMPS chip - probably enough to tip it over the edge. That power is averaged, so a varying load (which it is) will make it worse due to I2R losses
Brian
1 Like
Looking at the schematic I see no reason why you cant power it from the 5V pin.
the pin is marked as "OUT"
However by choice I'd feed it from a regulated PSU of 6-12V ; and use a seperate converter for any additional devices.
You made me go back to the schematic of the R4 Minima and it sure has a DC-DC buck converter. Still, that doesn't mean it's perfect:
This is the efficiency of the buck converter that's supposed to be on the genuine Arduino UNO R4 Minima. Note that the efficiency isn't too great when fed with 24V, 5V output and at very low current draw.
240mW @ 12V difference suggests 20mA RMS, implying you're all the way to the left of the diagram, as expected from your project description. Running the board by itself from 24V is likely a marginal affair; trying to power anything other than maybe a single LED at very low current will push it over the edge.
As to the thermal shutdown: note that it's supposed to engage at a die temperature of 150C. This means it'll shut off at that point, but what this doesn't guarantee is that the peak temperature of the device as such will be limited to this temperature. 150C isn't too far off from the point where you can expect solder to start melting. There's some capacitance at both input and output and that'll discharge somewhere on power down - apparently under thermal shutdown conditions, it may (at least temporarily) dissipate into the DC-DC converter itself. Again - it's not a mechanism to rely on under normal operating conditions.
Agreed, lack of a suitable connector on the 5V pin is enough reason to avoid it
Brian
It's true that it's strange that a buck converted that it's rated from 3V to 40V, 1.2A and so on, gets so hot at 24V. It's a very poor performance.
1 Like
Correct
At 24V, we are feeding in 1.56W = 0.31A @ 5V
The converter should be 85% efficient, so losses will be 0.23W
At 42°/W, the case should heat up by 9.7° and yet it burns my finger
Something is wrong. Could it be to do with the inductor value?
Data sheet recommends 22uH and the schematic says 10uH (this is the value fitted too)
Brian
So we've got that plus the Renesas core that will draw a little shy of 20mA worst case scenario, not counting whatever is being source by GPIOs.
How do both of your shields manage to draw 290mA combined? Sounds like a lot. Is this consistent with the characteristics of both your shields? I'd expect the ethernet one to be the more power hungry of the two.
Probably a bit less given the fact that the plot is for output load, not input current. Maybe closer to 80%, so that would leave 300mW dissipation in the converter itself. I recognize what you said about the thermal constant - the thermal pad at the bottom should make things easier, too. Still, it's a tiny 3x4mm package.
Depends on the application, but 10uH doesn't seem outlandish if you look at the equation of inductor size and run the numbers through it.
Sorry, doesn't add up. I don't doubt your observations - the only thing I doubt is whether the efficiency calculations are correct. See if you can establish the actual current draw on the 5V rail; see if there's a way to replace R11 with a shunt on a test board (e.g. 100mR) and use that to approximate current draw. Then see how much the converter really dissipates. Might turn out to be a whole lot more than the theoretical 250-300mW.
Thanks rsmls, I agree that something doesn't add up
I have two LEDs to indicate comms, but they are never lit at the same time. Nothing else on GPIO
This is power against supply voltage and shouldn't slope like this, it should be more or less constant with a slight increase towards 24V as the converter efficiency drops:
I have no idea what is going on - I'll contact support about this and update when I get a reply
Brian
Good experiment; well done, this gives an insight.
Almost looks like you've got some kind of Ohmic load going on. Do all boards show this same behavior?
Good call.
Good question: I don’t know for sure, but the four that have been powered up for a while have all desoldered themselves, so very likely
Brian
Yeah, that's a systematic problem for sure.
Do they also do this if you remove the modules (serial, ethernet)? And what if you replace the modules with a similarly-dimensioned Ohmic load (power resistor)?