virtual ground problem

From: Joerg Hansmann (info_at_jhansmann.de)
Date: 2002-01-18 01:05:48


Hi,

I have done simulations with SWCAD III on the virtual ground circuit
(that buffers Vref/2) of the modularEEG.

Originally the circuit was a TLC272 used as a follower (unity gain).
To get low HF output impedance I have bypassed the output to V-.
The problem however is, that compensated OP-amps (like the TLC272)
with capacitvely bypassed outputs show gain peaking and ringing, because
almost no phase reserve exists (about 1 degree).

The circuit with C5,R5, R6 is somewhat better than the original unity gain follower,
but still exhibits resonance at about 8.5 kHz.

The result seems quite usable to me, however it would obviously be better, if the
phase reserve were greater. Any ideas here how to achieve this without
replacing U1 with an uncompensated type ?
I could imagine that some type of dominant pole compensation
(with a matched zero) should work but have no idea how to realize this.

In the attached graphics a star formation consisting of 3 loads (Rl1..3, e.g. ICs)
connected to the virtual ground (Vref/2) is shown.
At Load1 a current is injected (noise, spikes, etc.) and at Load3 the resulting
voltage is recorded.

Regards,

Joerg



picture

This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:36 BST