MG996R issues with Arduino

Question

I am working on a robot dog, and using 12 MG996R servos. I am powering the servos via a 2S LiPo battery, and I have the battery's negative tied to the Arduino's ground to prevent ground loop issues.

Upon powering up the battery, I immediately noticed that some servos twitched even though there was no program running (the Arduino was turned off at the time). I attempted to move the servos by hand, and while the others still moved freely, these four had a noticeable resistance to turning.

Note that I had not yet used servo.attach(); so the motors should still be able to rotate freely.

In addition, one of the four began smoking. I unplugged the battery quickly. Now, all the motors rotate freely by hand, except the one which smoked is very hard to turn.

Could someone tell me what happened? I double-checked my wiring and everything is good. Positives to positives, negatives to negatives, and signal wires to Arduino pins.

I've run MG996Rs on 2S LiPos in the past with no issues, so I'm not sure what would cause the problem now.

I bought the servos from several different sellers. There are a few from Tower Pro (these three work fine), some from a company called DeeGoo-FPV (The one that smoked came from this company, and three more), and some unlabeled ones (The other three broken ones, plus two working ones.)

EDIT: I opened the burned servo and measured the motor output as shorted. One of the chips was burned as well (probably the one that made the smoke)

Answer 1

tl;dr: use 6 V

Those rascals are rated for 6 volts, not the nominal 7.4 V from 2S LiPos. You could imagine a Shenzhen manufacturer's design/value engineer zealously shaving a penny cost per unit by reducing safety factors and overstressing cheaper components, such that the servos would (mostly) work at 6 V but not necessarily at 7.4 V.

I once tried to run MG996Rs with 2S and, while not smoking out, they acted erratic and crazy. Retesting at 6 V, those same servos behaved correctly.

This all suggests a market opportunity for high-efficiency buck regulator modules to turn 7.4 V into 6.0 V. Better, of course, would be a line of 2S-rated servos for us to use.