I find that ticks aren't enough to aid in real problems, and I always turn on the the grid with 'grid(True)' in matplotlib. Also, the negative number problems must be some kind of anomaly; I've never seen that problem, and I've used several revisions of matplotlib every day for the past couple of years.
Out of curiosity, why do you need a grid? (I need very few ticks myself and almost never a grid. I've found that if the visual message isn't clear enough or needs a different quantitative analysis or display, I'll request the raw data.)
I'm almost always interested in the actual values of things when debugging a part our systems at work. I need to know that the generator exceeded 3kW, or the PDF of the voltage was above 4V, or things went hay-wire at 11:31AM. It is easier to tell at a glance what's happening on a plot if there is a grid.
