Maker Pro
Maker Pro

Floating Sensor with Mux and Inamp Design

M

Mark

Jan 1, 1970
0
I'm designing a circuit to read the output from several pressure
transducers, buffer the signal and then digitize it.

The pressure transducer provides a 0-10V signal and the output is stated
as isolated. Contacting the Manufacturer (Taber) they state that they
use a small DC-DC converter within the unit to convert 28V to a lower
voltage to power the electronics inside the unit. Thus, the output is
floating and the low side of the output is not connected to the 28V RTN.

I want to read the signal differentially so I'm using an instrumentation
amplifier. Since the sensor is "floating" I placed a resistor from each
input (+ & -) on the amplifier to the local circuit ground to account
for the input bias currents. The resistors to ground in front of the
in-amp will set the common-mode of the sensor when the mux switch is
closed. So this should work OK for the selected channel through the mux.

But then I realized the other channels feeding the mux, that are not
selected, are still floating and do not have any reference to the mux's
ground.

Is this design acceptable, or should I reference the sensors output to
my local circuit ground on the input side of the mux also?

Thanks,

Mark
 
N

Norm Dresner

Jan 1, 1970
0
Mark said:
I'm designing a circuit to read the output from several pressure
transducers, buffer the signal and then digitize it.

The pressure transducer provides a 0-10V signal and the output is stated
as isolated. Contacting the Manufacturer (Taber) they state that they
use a small DC-DC converter within the unit to convert 28V to a lower
voltage to power the electronics inside the unit. Thus, the output is
floating and the low side of the output is not connected to the 28V RTN.

I want to read the signal differentially so I'm using an instrumentation
amplifier. Since the sensor is "floating" I placed a resistor from each
input (+ & -) on the amplifier to the local circuit ground to account
for the input bias currents. The resistors to ground in front of the
in-amp will set the common-mode of the sensor when the mux switch is
closed. So this should work OK for the selected channel through the mux.

But then I realized the other channels feeding the mux, that are not
selected, are still floating and do not have any reference to the mux's
ground.

Is this design acceptable, or should I reference the sensors output to
my local circuit ground on the input side of the mux also?

Thanks,

Mark


1. How much do a few extra resistors cost?

2. Floating inputs, especially CMOS, are very vulnerable to static -- and
destruction.

3. I've seen a badly designed mux system where the settling time of a
selected input depended on the exact nature of the previous input -- i.e.
floating or grounded or active.

Norm
 
R

Rich Grise

Jan 1, 1970
0
Norm said:
1. How much do a few extra resistors cost?

2. Floating inputs, especially CMOS, are very vulnerable to static -- and
destruction.

3. I've seen a badly designed mux system where the settling time of a
selected input depended on the exact nature of the previous input -- i.e.
floating or grounded or active.

I would even keep the resistors on the input side of the mux, i.e.,
I wouldn't float the sensor between switch positions (or the equivalen). If
you worry that that will change your bias currents, just double the
R value.

Good Luck!
Rich
 
J

John Larkin

Jan 1, 1970
0
I'm designing a circuit to read the output from several pressure
transducers, buffer the signal and then digitize it.

The pressure transducer provides a 0-10V signal and the output is stated
as isolated. Contacting the Manufacturer (Taber) they state that they
use a small DC-DC converter within the unit to convert 28V to a lower
voltage to power the electronics inside the unit. Thus, the output is
floating and the low side of the output is not connected to the 28V RTN.

I want to read the signal differentially so I'm using an instrumentation
amplifier. Since the sensor is "floating" I placed a resistor from each
input (+ & -) on the amplifier to the local circuit ground to account
for the input bias currents. The resistors to ground in front of the
in-amp will set the common-mode of the sensor when the mux switch is
closed. So this should work OK for the selected channel through the mux.

But then I realized the other channels feeding the mux, that are not
selected, are still floating and do not have any reference to the mux's
ground.

Is this design acceptable, or should I reference the sensors output to
my local circuit ground on the input side of the mux also?

Thanks,

Mark


If the source is floating and the mux is differential, you have the
potential for huge common-mode voltages (hum, spikes, stray bias
currents, crud from the DC/DC things in the transducers). If the
deselected channels are all floating and their common-mode voltages
hit the mux rails, many muxes will screw up... the current pushed into
the deselected inputs can blow through to the output. Plus, you can
get latchup problems if you push CMOS mux'es past the power rails.

Since you have the (expensive) luxury of floating sources, why not
just hard-ground the low sides of all the transducers near the mux and
use a single-ended system? That will solve all sorts of common-mode
and EMI problems.

I'd RC lowpass filter the mux inputs, too. It always makes me nervous
to run raw semiconductor mux pins out to the world.

John
 
Top