Arduino Mega and Power

Karin_FL Aug 30, 2020

  1. Karin_FL

    Karin_FL New Member

    5
    1
    1
    Hello:
    First post for me. I've been looking at the DCC++ system, and it intrigues me. I have an Arduino Mega 2560 rev.3 and the Arduino Motor Shield. I have powered up just the Mega 2560 with downloaded code
    and it appears to send packets all ok, by monitoring the pins with an oscilloscope.
    A basic question for the assembled masses. Even though the Arduino docs say it is recommended to
    not exceed 12 volts DC being fed to the Mega board. Has anyone actually tried +15 volts DC????
    The spec sheet for the NCP1117 indicates max safe input voltage at +20 volts DC. I haven't measured
    the current being drawn by just CPU board by itself yet, so I am guessing that the caveat about exceeding
    12 volts DC is due to the heat sink capability of the CPU board. The only load that the +12 volt buss from the two pin power connector sees is the NCP1117 regulator which then feeds the rest of the +5 volt loads on the CPU board. Please get a little technical about the response, as I am degree retired Electrical Engineer and have tons of experience designing and building embedded computer and communication systems. The only other draw back to providing CPU board power via the Motor Shield would be possible current spikes being impressed on the CPU input power. Regards, Karin
     
  2. FlightRisk

    FlightRisk TrainBoard Member

    548
    237
    14
    They do fail. Assuming the SOT package. Even with the thermal shutdown circuit which puts the regulator into this oscillating mode, You are wasting 10V that turns into heat. You can do the math at 500mA you have to dissipate 5 Watts of power. If you put a motor shield and a wifi board on it, you will normally have enough current at 15V to overheat the chip. You can check the thermal characteristics on the datasheet and the board layout to see there is very little heat sinking on the board for that regulator. There are a lot of variables as you know, what else is being driven off the board, ambient temperature, heat sink and if it is it sinking or sourcing current to pins, etc.

    So even at a meager 200mA, that is (15-5)*.2 = 2Watts. At around 80 degrees C per watt trying to follow the ambient temp / junction temp graph, 80*2 = 160C. That's hot :) It goes into shutdown at 175C. I hate to use the "I think I remember" phrase, but apologies in advance, I think the SOT version of the chip can only handle 1.5W with no heat sink with the copper on that board. But let's say it can dissipate 2W, then I = P/E = 2/10 = 200mA before thermal shutdown. You wouldn't normally melt one to do short-term tests if you want to play, but they do seem to fail when they are run hot.
     
  3. BigJake

    BigJake TrainBoard Member

    3,259
    6,172
    70
    Welcome aboard, Karin! I am a retired EE too.

    Adding to what FlightRisk said, linear (including LDO "Low Drop Out") regulators draw the same input current as they output. Therefore if power in is the same current, times the input voltage, and power out is the same current times the lower output voltage, then the difference between power in and power out is dissipated as heat by the regulator.

    According to the datasheet, the regulator can supply as much as 800 mA at 5V with as low as a 6.5V, or as high as 12V supply over its recommended operating temperature. Data sheet maximum ratings (given without other conditions) are usually damage avoidance limits, not operational limits. Just because the actual voltage is below the overall maximum does NOT mean it will operate as specified elsewhere.

    Reliability goes down (failure rate goes up) rapidly with increasing component operating temperature, so to limit the power dissipation and heat in that regulator, I would prefer a power supply with probably an 8-10V output. Most wall warts don't have excellent voltage regulation over load, temperature and/or age, so that gives you a little room for its output voltage (the input voltage to the regulator) to droop before the regulator stops maintaining Vout=5V.
     
  4. S t e f a n

    S t e f a n TrainBoard Member

    167
    93
    6
    I'm new here too. This is interesting to me as well, but for the Uno. That one has an AMS1117 in the SOT package, and there is a note in the datasheet saying
    "SOT-223 package ϕ JA= 90°C/W* * With package soldering to copper area over backside ground plane or internal power plane ϕ JA can vary from 46°C/W to >90°C/W depending on mounting technique and the size of the copper area." (from http://www.advanced-monolithic.com/pdf/ds1117.pdf )
    it also says the thermal resistance junction to case is 'only' 15 C/W, and thermal protection shuts down at 165C junction temperature. On my board I see some thermal vias over an area of 17x10mm^2, so maybe the situation is not quite as bad Flightrisk and BigJake are putting it.
    i don't know; somebody should test it.

    Just be aware that (unlike most people on this board ;)) I'm not a retired EE; I like to think that my job was to break the stuff the EEs designed and the technicians built...:ROFLMAO:

    Regarding the need for 15V (I have read posts saying that 14 or 15V is the optimal voltage for N scale): Currently I'm running my Arduino and the shield with a 12V/5A (nominally) supply, and that seems fully sufficient. i achieve realistic top speeds at 50% throttle. But I'm still at the testing stage, with just two (Kato E9) engines and eight lit passenger cars.
     
    Last edited: Sep 21, 2020
  5. Karin_FL

    Karin_FL New Member

    5
    1
    1
    One thought on this issue. And this is probably going to be sacrilege to most of you. A modification to the Arduino Mega2560 board could be done.
    As far as I can tell by reviewing the schematics for the Mega2560 and the Motor Shield is that the signal labeled VIN on the Mega2560 only goes to the input power to the voltage regulator refdes IC1 (NCP1117), and a compartor IC7B. This regulator is what develops +5 volt for the Mega2560 board. On the Motor Shield board VIN comes from a jumper trace that can be cut. One side of the jumper trace is labeled VIN and the other side labeled VMOT, which is the motor input supply voltage (+15 volt). Now to the sacrilege. Texas Instruments p/n TPSM84205 is a small switching regulator board with a TO-220 pinout. One could remove the NCP1117 on the Mega2560 board and replace it with the TPSM84205. Then you could provide +15 volt to the motor shield board without cutting the jumper trace. The Motor Shield would have +15 volt to drive the H-bridge chip, the +15 volt is supplied to the Mega2560 via the jumper trace and its connection via the connector labeled POWER (8x1F-H8.5) on the Mega2560 board. The only other connection of +15 volt on the Mega2560 board is to IC7B through a 10k-10k voltage divider, which controls logic to select either board +5 volt or USB 5 volt sourcing. Now this is not for the faint of heart to accomplish, but in my opinion this is an option if one wishes to create some flexibility in powering options.
    Karin
     
    OlyPen likes this.
  6. BigJake

    BigJake TrainBoard Member

    3,259
    6,172
    70
    One thing to double check is whether the TPSM84205 has a minimum output current to regulate, and whether the Mega board will continuously draw at least that current, unless powered off by the 15V supply. Most switch mode supplies today do not require a minimum load to regulate, but a few still do, so just check and make sure.

    Also ensure that any required input and/or output capacitance for the new regulator are present, or can be added. They are often significantly higher than for linear regulators (LDOs), but the TPSM84205 may already provide what it needs. Again, just check and make sure.

    Stefan, it is likely that the amount of heatsinking capacity provided was only that designed for the power dissipation at the specified input voltage range, and therefore might not be capable of rejecting enough heat at higher than specified input supply voltage. In high volume manufacturing, vias and copper cost money (copper that is etched off during manufacturing can be recycled), so they may or may not have optimized the vias and copper to suit only their specified operating environment and supply voltage.
     
    OlyPen likes this.
  7. Karin_FL

    Karin_FL New Member

    5
    1
    1
    From the data sheet, the TPSM84205 has no minimum current. I've used them a lot and they are pretty stable and neat devices.
    Karin
     
  8. BigJake

    BigJake TrainBoard Member

    3,259
    6,172
    70
    Then as long as the minimum input and output capacitance is met, you should be in good shape!
     

Share This Page