When you first build an OpenArm robot, following the official documentation feels reassuring.
The bill of materials looks reasonable: a 24 V DC power supply, a few hundred watts of rated power, standard wiring.
You assemble everything, power it up, and — it works.
- The arm moves.
- The joints respond.
- Nothing smells hot.
So where is the problem?
The problem usually appears later, when you stop demonstrating the robot and start using it.
The Typical User Story (If You’ve Built Robots, This Will Sound Familiar)
At first, everything is fine:
-
Single-joint motion
-
Low speed
-
No payload or a very light one
Then you try something more realistic:
-
Two or three joints accelerating at the same time
-
Faster trajectories
-
A real payload
-
Maybe even a dual-arm setup
Suddenly you notice:
-
The robot feels “soft” under load
-
Movements slow down unexpectedly
-
The controller reports undervoltage warnings
-
In the worst case: the power supply shuts down and restarts
At this point, many users start questioning motors, gearboxes, or control loops.
In reality, the motors are usually innocent.
What the Official Power Setup Implicitly Assumes
The official OpenArm power recommendations are not wrong — but they are optimized for a very specific operating window.
Implicit assumptions include:
-
Mostly single-axis motion
-
Low acceleration
-
Small payloads
-
Educational or demonstrator use
-
Low duty cycle
As long as you stay within that window, the system behaves well.
The moment you leave it, the electrical design becomes the limiting factor.
This is not unique to OpenArm — it is a classic transition problem from demo robot to real robot.
Robot Power Is a Transient Problem, Not an Average One
One of the most common misunderstandings in robotics power design is focusing on average power.
Robots do not consume power smoothly.
They consume power in bursts.
Let’s look at the basic mechanical relationship:
P=τ⋅ω
Torque times speed.
Now imagine a realistic scenario:
-
Several joints start moving within the same control cycle
-
Each joint briefly demands high torque to overcome inertia
-
Gear reductions amplify motor current demand
Even if each joint only needs, say, 200–300 W electrically for a short moment, four joints starting together already create a 1 kW transient load.
This peak may last only 50–200 ms —
but that is long enough to:
-
Trigger current limiting
-
Cause voltage sag
-
Activate protection circuits
Power supplies are usually rated for continuous power, not for robotic acceleration peaks.
A Real Community Example: Worst-Case Power in Dual-Arm OpenArm Systems
A particularly illustrative example can be found in an official OpenArm GitHub discussion, where community members attempted to estimate the worst-case power consumption of a dual-arm OpenArm setup.
In this discussion, one contributor started from the actual power supplies used in the reference setup — two independent AC/DC supplies rated at approximately 360 W each. From this, they concluded that even in a pessimistic scenario, the total electrical power available to the system would realistically remain below ~720 W, simply because the power supplies themselves define a hard upper bound.
Another participant tried a different approach:
they summed up the rated currents of all joint motors, multiplied them by the bus voltage, and arrived at a number close to 2500 W. However, the author explicitly noted that this result was clearly unrealistic and massively overestimated the real operating conditions.
This contrast highlights a common engineering pitfall:
Summing nominal motor ratings does not describe the real power behavior of a robotic system.
Motor nameplate values typically describe:
-
maximum continuous current, or
-
worst-case thermal limits
They do not represent:
-
simultaneous operating points of all joints,
-
transient behavior under coordinated motion, or
-
the actual limits imposed by the power supply and DC bus.
The key takeaway from this discussion is highly relevant for system design:
In a robot like OpenArm, the power supply architecture defines the practical upper limit of system power, not the theoretical sum of individual motor ratings.
This community exchange reinforces an important principle already well known in industrial robotics:
real system behavior must be evaluated at the system level, not by aggregating component datasheet maxima.
Once this distinction between component ratings and system limits is understood, the role of wiring and DC-bus design becomes obvious.
Reference: OpenArm GitHub Issue #229 – “Worst case power consumption of OpenArm with dual-arm configuration”
Why 24 V Makes Everything Harder Than It Looks
At 24 V, current rises very quickly with power demand:
-
500 W → ~21 A
-
1000 W → ~42 A
-
2000 W → ~83 A
Those numbers are not theoretical — they appear in real OpenArm systems under dynamic motion.
High current has very practical consequences:
-
Voltage drop along cables
-
Heating of connectors and crimps
-
Reduced voltage at the motor drive input
And here is the critical part:
Less DC bus voltage means less available motor power and torque.
This is why users often describe the symptom as
“the arm feels weak”
instead of
“the power supply is collapsing”.
The physics is simply hidden inside the wiring.
A Subtle but Critical Topic: Regenerative Energy
There is another effect that is rarely discussed in open-source robot documentation.
When a robot joint decelerates — or lowers a payload — it does not consume energy.
It produces energy.
The motor becomes a generator.
In industrial servo systems, this is well known and carefully managed.
In many hobby- or research-grade setups, it is simply ignored.
Most AC/DC power supplies:
-
Cannot absorb energy from the DC bus
-
Are designed for one-directional power flow
-
React to rising DC voltage with over-voltage protection
So what happens during a fast deceleration?
-
Energy flows back into the DC bus
-
Bus voltage rises
-
The power supply shuts down
-
The entire robot loses power
From the user’s perspective, this looks random.
From an engineering perspective, it is completely predictable.
The Engineering Conclusion
The official OpenArm power design is not wrong.
It is simply minimal.
It assumes:
-
Low dynamics
-
Low concurrency
-
Low regenerative stress
Once you move beyond those assumptions, the system must be redesigned as a real DC power system, not just a “supply plus wires”.
That redesign typically includes:
-
More realistic power headroom
-
Proper wiring for high current
-
Thoughtful DC bus architecture
-
Regenerative energy handling
What Comes Next
In the next article, we will zoom in on a topic that looks boring — but decides everything:
👉 Episode 2:
“Why Wiring Determines Performance in 24 V OpenArm Systems”
We will talk about:
-
Cable resistance and voltage drop
-
Why “thick enough” is often not enough
-
How wiring silently limits torque and speed