Have you looked at figure 15 in Philips application note AN00022?
Thanks for your answer. I know this application note. I tried the circuit
which is shown in figure 16. The circuit dosn't work properly. One current
peak (negative) is one and a half times higher than the oder (positive).
André
C6 is supposed to guarantee that average currents are equal in both
directions.
Ensure that C4.R17, C5.R18 time constants are much longer than C6.KMZ5
time constant. Only C6 should limit total charge.
Parts used for TR1 and TR2 should have roughly the same saturation
voltage and reasonable gain at the 1A current expected. This allows
KMZ5 impedance (6R) only to determine current pulse amplitude.
R21.C9 time constant should be long compared to rep rate, and C9 must
be a low esr (< 1/10th KMZ5 resistance), high capacity (>10xC6) part.
No imbalance will occur if parts are suitable to the application.
RL