Why Your Second Set Is Faster Than Your First: PAP or Sensor Warm-Up Artifact?

Set 1 on the bench press: 100 kg, 0.51 m/s mean concentric velocity. Set 2, same 100 kg, same rep count, same athlete who insists nothing felt different: 0.58 m/s. That is a 14% jump in bar speed on paper, enough to make a coach reach for more weight next week or write 'great session' in the athlete's log. Before either happens, it is worth asking which of two very different things just occurred. Did the first set genuinely potentiate the nervous system for the second, the way a heavy conditioning contraction is supposed to? Or did the sensor take longer to settle than the athlete did? Both produce an identical-looking number on the screen and lead to completely different coaching decisions. One is a real, modest neuromuscular effect worth programming around. The other is a device artifact that will quietly bias every load-velocity decision built on top of it until someone controls for it.

Before running any test, it helps to know what each explanation should look like if it were the real cause. Post-activation potentiation and sensor warm-up drift leave different fingerprints once you know where to look.

The overlap is what causes the confusion in the field: a typical working set is both a heavy conditioning stimulus and the first time the sensor has taken a real rep since it was mounted. The two effects stack, and nobody sees them separately unless the setup is deliberately controlled.

Post-activation potentiation works at the contractile level: a heavy conditioning contraction phosphorylates the regulatory light chain of myosin, which makes the muscle more sensitive to the calcium released during the next contraction. The practical result is the same neural drive produces a slightly larger, slightly faster contraction for a window of several minutes afterward. Wilson et al. (2013), pooling 147 subjects across multiple conditioning protocols, reported a mean effect of +2.6% jump height and +3.8% sprint velocity, with the effect peaking when rest between the conditioning set and the test fell in a 7-10 minute window - shorter and the fatigue from the conditioning set still dominates, longer and the potentiation has decayed. Seitz and Haff (2016), reviewing 32 studies on jump, sprint, and throw performance, found the same broad pattern and noted the effect was consistently larger in stronger, more trained athletes.

Two limitations matter for interpreting a faster second set specifically. First, the conditioning stimulus has to be heavy enough - a set at 50-60% 1RM is not a strong enough contraction to trigger meaningful RLC phosphorylation, so if your set 1 is a light ramp-up rep rather than genuine working weight, potentiation is not a plausible explanation for what you are seeing. Second, Hamada et al. (2000) found that a meaningful share of athletes, on the order of 30-40% in unselected populations, show no measurable potentiation response or even a net decrement after standardized heavy conditioning. If an athlete has never shown a second-set bump before and suddenly shows a large one today, genuine physiology is a less likely explanation than something that changed in the setup.

Every IMU-based velocity sensor runs on MEMS accelerometers and gyroscopes, and both have a well-documented bias-instability period immediately after power-on: the internal temperature of the chip is still stabilizing, and the reported zero-point drifts slightly during that window before settling. Manufacturers typically spec this stabilization window at somewhere between 60 and 180 seconds. A linear position transducer has its own version of the same problem - cable tension and reel tension take a rep or two to settle into a consistent baseline, and Bluetooth streaming connections often take a few packets to lock into a stable sampling rate after a fresh pairing.

The tell is that this artifact does not care what you actually lifted. If you power on the sensor, mount it, and immediately do an empty-bar rep followed by a second empty-bar rep thirty seconds later, you will often see the same kind of jump between rep 1 and rep 2 that shows up between a heavy set 1 and set 2 - because the mechanism has nothing to do with muscle physiology. In practice, this artifact almost always gets misread as potentiation because the two events happen at the exact same moment in a real session: the sensor gets mounted right before the first working set, so its settling window and the athlete's first heavy contraction land on top of each other.

Rather than guessing which explanation applies, isolate the two variables the same way you would isolate any other measurement question - one at a time, across a few sessions, at a fixed reference load.

The key control is session 2: light load, pre-idled sensor, and if the second-set jump disappears there, it was never about the muscle. If it persists even with a submaximal set 1 and a fully settled sensor, something else is going on - loosening equipment, a fatigue-then-recovery pattern within the set, or genuinely improving technique across reps - and it is worth cross-checking against our noisy velocity readings guide.

A trained lifter (back squat 1RM 150 kg) showed set 1 at 0.61 m/s and set 2 at 0.70 m/s on a 120 kg working load - a 14.8% jump the coach initially logged as a strong potentiation day. Running the isolation protocol: in session 2, with the sensor pre-idled for three minutes and set 1 capped at 65 kg (roughly 43% 1RM), set 1 read 0.94 m/s and set 2 read 0.96 m/s - a 2.1% difference, well inside normal rep-to-rep noise and consistent with no meaningful artifact once the device was settled.

In session 3, with the sensor still pre-idled but set 1 raised to a genuine 128 kg (85% 1RM, 3 reps), the 120 kg reference set afterward read 0.68 m/s against a non-conditioned control set at 0.65 m/s - a 4.6% increase, in line with the magnitude Wilson et al. (2013) reported for trained subjects at a 7-10 minute rest window. Session 4 replicated a 3.9% delta under the same protocol a week later. The conclusion for this athlete: roughly two-thirds of the original 14.8% jump was the sensor still settling, and the remaining piece was a real, modest, and repeatable potentiation effect worth keeping in the program - just not worth the 15% number the raw session 1 data suggested.

If the whole 14.8% swing had been treated as real, the natural coaching response is to add load next week, chasing a number that a settled sensor would not have produced in the first place - and the athlete would likely miss the new target, since the underlying capacity never actually moved that much. Treat it as a sensor artifact instead of genuine adaptation and you risk the opposite error: dismissing a real, if smaller, potentiation effect that could be used deliberately, for example by placing a heavy triple 7-10 minutes before a max-effort attempt or a competition lift. Running the isolation test once per athlete per tracked exercise settles the question permanently for that combination of device, mount point, and lift, and the two-minute pre-idle habit costs nothing to build into a standard warm-up routine going forward.