[email protected] wrote...
Sounds like common mode voltage, you need 2 voltage dividers, one for
each input of the op amp. A much better way is to use an ac input opto
coupler.
Two dividers, yes, but with their low end tied to the measurement-
circuit ground. Inexorx says he's seeking a replica of the AC line
voltage, and an opto-coupler wouldn't be accurate. Nor do I think
a transformer is necessary, provided certain precautions are taken.
The AC power line can have short spikes to many thousands of volts,
and it's important that the resistive divider circuit handle these
without endangering the equipment or any users. Because the subject
is complex, I'll not take the time to here to explore it, except to
say that even a simple version should have many components.
So a transformer may be the easiest solution. If Inexorx needs an
accurate output, he can lightly load the transformer with an RC to
control high-frequency error and still enjoy a wideband low-voltage
replica of the line voltage. For example, let's consider a Signal
Transformer 241-3-24. This is a small 2.4VA split-bobbin type with
excellent input-output isolation, but with the usual losses that we
see in small, cheap transformers. We can analyze a simple model*
to estimate its errors.
primary L_m, 4.6 H (low-level L-meter)
primary L_m, 11.2 H (meas with 120Vac)
primary Rdc, 372 ohms
secondary L_ell, 25.1 mH
secondary Rdc, 23.8 ohms
open-circuit turns ratio, 0.254
First we'll consider the transformer's primary circuit, where its
roughly 11H of (nonlinear) magnetizing inductance has a reactance
of j4.27k ohms at 60Hz, which with 372 ohms of series resistance
and 120V causes a 28mA magnetizing current at a calculated lag of
-90+5 = -85 degrees to the 120 60Hz ac voltage. The transformer
primary's R-L calculated vector loss is 0.4% at 60Hz, dropping to
0.09% at 120Hz, and less at higher harmonics. The 5-degree phase
lag at 60Hz drops to 2.5 degrees at 120Hz and less above that.
Second, we'll consider the scene at the transformer's secondary.
We can add a 100pF output capacitor, which with the 25mH leakage
inductance will limit the high-frequency response to 100kHz, plus
a 15k load resistor to damp the 100kHz L-C resonance. Multiply the
primary's series copper resistance by 0.254^2 and add this to the
secondary's resistance to get the transformer's Zout = 47.8 ohms.
This means our 15k load will drop the output voltage by about 0.3%,
which is fine given we don't know the turns ratio any better than
than anyway.
Lord Inexorx will end up with an accurate voltage transformer up
to 100kHz, with a small 0.4% drop-off at 60Hz (or an above-200Hz
increase of 0.4%, if he prefers to think of it that way). This
small error could be corrected with a few more parts.
* Data from leakage-inductance threads on s.e.d., January, 1998.
Try these threads, Message-ID: <
[email protected]>
Message-ID: <
[email protected]>