If the periods are roughly divided in three periods, from 1976 to 1986, from 1987 to 2008 and from 2008 onwards, pretty accurately on each period an exponential function can be fitted (excel does the sum of squares calculations within a split second). Thus, the following equations result:
for the first period:
y1 = 2602,6e0,1181(x-1975)
for the second period:
y2 = 4331,4e0,0744(x-1975)
and for the third period:
y3 = 50021e0,0014(x-1975)
wherein y1, y2 and y3 is the total debt for each period, x is the year in question.
The turning point, the time of change of one "debt regime" into the other can be calculated by solving for x under the condition y1 equals y2 and under the condition y2 equals y3.
The first turning point is calculated to be about late July 1987, the second turning point is calculated to be about early December 2008.
From these relative rough excell generated exponential fits, and from the relatively rough "end of year only" total debt figures, these two turning points pretty accurately reflect the 1987 crash panic moment (October 19, 1987) and the credit crunch panic moment (October 2008).
Apparently, what is needed for a major change of regime in total debt development, is a "panic moment", after which debt increment appears to slow down considerably.
If we consider the debt development of the last four years (2011 still being an estimate), total debt has virtually stabilized, despite the fact that vast amounts of money have been injected into the US and other economies.
Scary enough, it appears that we are only one further panic moment away from a seriously declining total debt. When this happens, money is virtually evacuated from economy to reduce debts, inherently resulting in a deflation period.
I very much hope that I am wrong!
 Federal reserve statistical release, Flow of Funds Accounts of the United States, June 9, 2011, Z1 D3 p. 9.