Why is voltage drooping so much when driving an RC servo (also ripple problem)?

[Involute]

I'm controlling an RC servo from a 5V Arduino (Pro Trinket). I'm supplying 5V to the entire system from a 5A bench supply. scope_0.jpg shows the supply output when the motor's idle. Scope_1.jpg shows the output when the motor is not quite stalled, at which point it's drawing about 800 mA. I have the current limited to 1.5A at the supply. Why is the voltage drooping so much, and why is the ripple so high? I would think a 5A supply should be able to handle this without breaking a sweat, but it's a Korad, not a Keysight, so maybe not.

I tried a variety of caps across the supply output, but didn't see much effect until I got to 2200 uF (the biggest I have on hand).  Scope_2.jpg shows the unloaded output, and scope_3.jpg shows it loaded.  The ripple's a lot better in both, but the droop is still there under load.

Explanations/suggestions greatly appreciated.

[stmdude]

Wild guess: Tiny/thin wires involved.  0.8A*5V is quite a bit of power, meaning that the resistance in wires and traces become a big factor.

[Involute]

Great idea!  I'll beef up the related wires and report back.  Thanks. 

[Mechatrommer]

add big capacitor near the servo

[Involute]

Upgraded the wire, no difference.  Substituted a power supply harvested from somewhere that I had lying around and that solved the problem.  Sounds like I need to ditch my Korad bench supply.  Too bad.  I really liked it until this.

[kurt]

add big capacitor near the servo

Or a "right size" capacitor near the servo.

Looks like the current waveform have a very short rise time. So a big capacitor might have to much inductance to do anything about it. That might also be the reason the filter caps in the PSU can't deal with it.

The servo can probably both have short rise times and relatively long pulse duration, so it could be that two caps are needed. A small low ESL to filter the sharp edges and a larger bulk capacitor to float it over the rest of the pulse. Then again, it could be that the PSU caps can handle the latter if the sharp edges are filtered enough by a small low ESL cap.

[Involute]

How small and how low ESL?