Hi All,
I've written quite a large program (it could be rationalised) and there are no observable issues with the program at all. When I first turn on the Arduino, invariably it works fine.
However, after a few seconds, the third PCA9685 controlling my servo starts producing weird results. I do have a lot of kit on my Arduino Mega including two thirty LED NeoPixels and so far around 36 servos on three PCA9685s, but the first PCA9685 has it's own power supply going to it, which is then daisy chained to the next two.
It's weird because I can invariably operate the servos attached to the third PCA9685 a couple of times before it starts behaving weirdly, but it's very inconsistent and sometimes it rubbish straight off. It's like the servos have a fit sometimes.... yet the first two PCA9685s are working perfectly.
Moreover, occasionally, the pressing of one of the switches that operate the third servo PCA9685 will force it all to reset.
Power wise, I have a 30w plug, that has a 2A supply for DC and 2.1A for the USB, the USB goes to the USB on the Mega and the DC goes to the supply on the first PCA9685.
Is it likely a power issue?
I ask as I'm not really sure whether I should put another power supply to the second or third PCA9685 or buy a bigger PSU?
The servo power supply needs to provide at least 1 Ampere per servo for small servos like the SG90, and at least 2.5 Amperes per servo for larger ones like the MG996R.
So, for 36 servos at 5V, you will need a 180W power supply, minimum.
You need to do a power budget for your servos, and for your overall project. 2A is way too low, but what you really need for a power supply is very dependent on what your doing with all your attachments.
C
Unfortunately you can't count on that. Any noise in the servo input signal line can cause it to start moving, and when it does, it briefly draws the stall current.
If the power supply is inadequate, this causes a voltage drop on the line, which in turn causes other servos to see false signals. The result is a twitching mess.
It's like the servos have a fit sometimes.
If you look through the forum, someone posts about this exact problem nearly every day, almost always due to an inadequate servo power supply.
Avoid that by choosing a power supply following the recommendations above.
Okay. Let's put it this way. Pick one servo. Run it towards one end of travel, but mechanically block it from getting to the chosen point. It's now in stall. Because it hasn't gotten to the end point, you can do virtually nothing to prevent it. Even if you remove your drive signal, I can't tell you if your particular unit will cease drawing stall current, because that's not specified. This is one of the hidden gotchas with servos.
Is this a model railroad application? If so, I can tell you what I've done to help with some of the issues that can arise with servos used for turnouts - message me off-forum, if you wish.
C
What gauge of wire would I need to connect the PCAs?
I mean, 5v 40A is a lot of current and the 'standard' breadboard style wire is surely going to be insufficient?
Sorry - long time since I've done electronics
Breadboards are for temporary experiments with low power logic circuits, and cannot be used for motors and servos -- the tracks burn.
For servos, most people either solder the connections or use a servo power distribution PCB. You can buy the latter, or make your own: https://diyodemag.com/projects/servo_power
You still haven't told us what servos you are using. At the very least, look up the stall current.
All my wiring is via, well, wire not a breadboard.
It uses solid core, not sure about the AWG though. I will check it when I power it up, but it's definitely 'fatter' than normal breadboard wire.
Also, the wiring from the PCA to the Servo is always directly via the servo cable, but sometimes with an extension cable too.
It is for model railways.
Firstly, you'll run multiple wires from the power supply to each PCA board. That means you'll only have the current of 16 servos in that wire, not the whole thing.
Secondly, most screw terminals will take 14, or at least 16 AWG wire. Unless all your servos are in stall simultaneously, that will be sufficient.
Note, the power to the PCA boards must be from your other supply, not the supply feeding the Arduino. Keep them separate, and add only one ground wire between the Arduino and the PCA supply as a common reference.
HTH
C
Without a strong indication of future plans, that will certainly cover your current needs(groan). I do wonder if you might be better served with multiple smaller units, but that's a layout-topology thing. I also wonder how far you're extending your I2C, as that can get problematic. If everything is centralised, that supply is likely a good answer, especially if you have more LED strings or more servos down the road.
A comment on power distribution - you're playing with significant amperage outputs here. That translates to the potential for fire. I'd advise fusing each run to each PCA board separately, as well as any other distribution runs. If you have a short at the end of a run of wire, you don't want the internal protection of the supply to deliver, worst-case, 59.5 amperes into a short circuit somewhere in the wooden structure of your layout; much better to fuse appropriately so that your fuse dies well before heat develops somewhere unpredicted.
I've used something similar to these in the past:
Fuses can never protect your electronic components. They are there to stop your copper connections from over heating. That includes the copper traces on your circuit boards. If all your individual power connections use similar currents, then one size fuse can be used in EACH of the power circuits. Know the amps used and select a fuse one step higher.