There is a certain normal waveform to be expected from the IABP, when it is correctly timed.
A good overview of what happens can be found in an educational propaganda document by Arrow.
The balloon itself has a pressure transducer, and it generates a waveform.
About 40 milliseconds before the dicrotic notch, the IABP balloon inflates. This is timed with the ECG, usually - the end of the T wave is used as a marker that systole has finished. Why the delay? because even the best IABP pistons require a few milliseconds to shoot some helium into the balloon. Balloon deflation (which is also very rapid) is timed with the R wave.
The aortic pressure waveform generated by the deflation and inflation of the balloon demonstrates some of the hemodynamic effects of IABP counterpulsation.
The waveform in the diagram above is slightly exaggerated, in order to simplify the explanations.
In that diagram the graphical representation of the events occurring around the balloon deflation are "stretched" slightly, giving the impression that there is a long time between the deflation of the balloon and the beginning of the assisted systole.
Now, the diagram above is perhaps even more stylised, and is far from being a faithful representation of reality.
However, the basic principles are there.
Inflation of the balloon is triggered by the beginning of diastole, which correlates with the middle of the T-wave. The balloon is timed to deflate at the very end of diastole. This correlates with the R-wave on the ECG, and this is the most commonly used trigger for balloon deflation.
In atrial fibrillation, the ECG trigger is timed to deflate on the R wave as usual, but the R-R interval (which governs the timing of the balloon remaining inflated) varies. The R wave timing can also be of the "pattern" type, where normal QRS morphology is expected, or "peak" type where - if your QRS is monstrously misshapen - the IABP will choose the maximum voltage peak and use that instead.
Apart from ECG triggering, other methods exist:
Pacemaker timing: this is a well-practised technique (it seems to have first been described in a letter to the editor of The Annals of Thoracic Surgery by John Kratz, 1986). In short, there may be situations where the ECG measurement is either unreliable or unfeasible (eg. your open chest patient has literally no surface for the electrodes, or they are covered in a film of sweat which causes them to fall off). It is possible to slave the pump to the temporary pacemaker pulse generator, to time the deflation of the balloon according to the pacemaker pulse instead of the R wave. Clever modern pumps have "atrial" and "ventricular" pacing trigger settings, with appropriate timing offsets.
One minor issue with this is the possibility that you don't want the pump to be pacing-timed, but it forms a treasonous alliance with the pacemaker, against you and the patient. This can happen when you turn off the high-pass filter on your ECG monitor to see the pacing spikes (Reade, 2007). The IABP then mistakes these for R wave complexes and deflates the balloon. The resulting early deflation is usually not an issue because the ventricular pacing spike and the QRS are fairly close together. However, if the IABP decides to time deflation with the atrial pacing spike, all the benefits of systolic augmentation can be lost.
Arterial pulse timing is for situations when the patient is not paced, nor is the ECG any good. It is a poor second to ECG timing because of a noticeable delay to the balloon inflation. Ideally you'd expect the balloon to start inflating about 40 msec before the dicrotic notch (to compensate for the fact that even helium doesn't flow instantly). By using pressure trigger, one relies on the propagation of the pressure wave, which - though brisk, ~10m/sec - is not as fast as the electrical signals. The delay was measured by Pantalos et al (2003), who simultaneously measured the aortic root and the IABP machine-end lumen pressure. Delays ranging from 60-119 msec were seen. l This has the effect of decreasing diastolic augmentation and increasing afterload, which could be disastrous.
Asynchronous timing is also an option. The pump defaults to a regular rate of 80 bpm, irrespective of what the myocardium is doing. In many ways this is the philosophical opposite of "timing", i.e. the inflations aren't timed to the cardiac cycle in any sense. Obviously this is only useful if there is no cardiac cycle, i.e. the patient is asystolic.