Isolated DC voltage sensor

[Alexis]

Hi,

I need a voltage sensor that reads a voltage going from 0V to 60V DC. I don't
want to pick up the noise present in the power stage in the signal, so I think
that it should be isolated.

Any idea about what I could use or how I can do it? Of course I don't want to
pay a three figures price for a sensor...

Thanks,

Alex

 

[CFoley1064]

Hi,

I need a voltage sensor that reads a voltage going from 0V to 60V DC. I don't

want to pick up the noise present in the power stage in the signal, so I
think
that it should be isolated.

Any idea about what I could use or how I can do it? Of course I don't want to

pay a three figures price for a sensor...

Thanks,

Alex

Questions:
How accurately do you need to read the voltage?
How frequently do you need to need to have data ?
What (or who) is reading this -- do you want a display, or is the data going
somewhere else, like a PIC or a PC? If it's data you're looking for instead of
just a display, how do you want it -- frequency, pulse width, serial, parallel
data bits?
What kind of power supplies do you have available on both sides of the
isolation barrier?
What skill level are you bringing to this project?
Sorry, but anyone reading your initial question really won't be able to help
you, because you haven't given enough information to allow a good answer.

Good luck.
Chris

 

[Mario Trams]

I need a voltage sensor that reads a voltage going from 0V to 60V DC. I
don't want to pick up the noise present in the power stage in the signal,
so I think that it should be isolated.

Any idea about what I could use or how I can do it? Of course I don't want
to pay a three figures price for a sensor...

You could use an opto coupler where the diode is driven with the voltage
you want to measure (with an appropriate series-resistor, of course).

The output of the opto coupler will be proportional to the voltage.
The transmission curve won't be very linear, but you can fix this
in software (when you are using some sort of micro controller).
There are also some troubles with the lower voltage-end where the
current through the opto coupler diode is not sufficient to cause any
effect. Here it should help to do some active preconditioning
of your signal to measure. I.e. you could scale it down to a smaller
range with an OpAmp (perhaps with an additional voltage divider as
prescaler) and then add an offset voltage (as well with an OpAmp) so that
the input voltage of 0V marks the lower bound where the opto coupler
starts to work.
But do not forget: For this active signal conditioning, you must not use
the power supply that is used for the back-end measurement circuitry
(i.e. AD-converters etc.). Instead, you have to use a power supply from
the system under measurement. When this is not possible (i.e. all you
got are two wires with ground and the 0-60V signal), you could use a
DC-DC-converter for generating an isolated power supply voltage.

Regards,
Mario

 

[Theo]

You could use an opto coupler where the diode is driven with the voltage
you want to measure (with an appropriate series-resistor, of course).

The output of the opto coupler will be proportional to the voltage.
The transmission curve won't be very linear, but you can fix this
in software (when you are using some sort of micro controller).

Just take a linear opto coupler.There are at least two manufacturers that
produce them.

regards,
Theo.

 

[Rene Tschaggelar]

Hi,

I need a voltage sensor that reads a voltage going from 0V to 60V DC. I don't
want to pick up the noise present in the power stage in the signal, so I think
that it should be isolated.

Any idea about what I could use or how I can do it? Of course I don't want to
pay a three figures price for a sensor...

There are voltage to frequency converters.
Their signal is a pulsetrain to be sent over a optocoupler.

Rene

 

[Fred Bloggs]

Tony Williams wrote:
[snip u of quebec twinkie qweschun..]

An expensive, but easy to implement, method would be to use an LEM
hall-effect isolated current-current sensor. The most sensitive
they produce has an input full-scale of 10mA into a 250 ohm input
resistance. It needs no power on the primary side and 15-0-15v on
the secondary side. Cost is less than 3 figures, but only just.

At the other end of the scale, the cheapest and crudest method
would be to use a 2PCO relay and a flying capacitor. Power is only
needed on the secondary side. Very easy to implement, very slow
update rate, and with the life being limited to 1E8 to 1E9 relay
operations. Even with the life disadvantage I have seen this
method used quite successfully in an industrial 4000-way
thermocouple scanner.

If power is available on both sides then a variety of methods can
be used, usually using an optocoupler, either in a linear mode, or
modulated in some way. The Infineon IL300 is a linear optocoupler
with one LED and a pair of matched photodiodes. A simple servo
loop on the primary side produces a matching output current on the
secondary side. Not difficult to implement, good bandwidth, and
the resulting stability and linearity is good.

High isolation DC-DC converters at 250mW under $10 make synchronous,
highly accurate VFCs like the AD7740 very feasible. Since the output is
ratiometric to clock frequency, ac-couple the DC-DC converter ripple
into comparator for VFC clock source and transmit the gated output burst
across OC- analog mux multiple inputs and/or use other way OC to switch
in 0 and FSV for gain /offset calibration.

 

[Yzordderex]

Their signal is a pulsetrain to be sent over a optocoupler.

Rene

HCPL-7840 is Agilent solution. Uses I think a serial link inside.
40KhZ bandwidth. I use them on 600vac drives for short circuit
current sensing on motor leads. NO PROB with high dv/dt. Watch out
for the TIL300 solution. Has no dv/dt immunity. Sucks if one side of
opto moving in relation to other. A little messy to implement too.

Why not just use a differential amplifier? Put op-amp at earth
ground. (I assume your control ckts at earth.) A couple of from say
100k to 1MEG resistors from your bus to the op-amp inputs and just
figure what gain you want for feedback resistors. This has been a
standard method for many years. Got a litttle noise you want filtered
- just place a couple small caps across the feedback resistors to form
pole.

73
Bob
N9NEO

 

[Tony Williams]

Why not just use a differential amplifier? Put op-amp at earth
ground. (I assume your control ckts at earth.) A couple of from
say 100k to 1MEG resistors from your bus to the op-amp inputs and
just figure what gain you want for feedback resistors. This has
been a standard method for many years. Got a litttle noise you
want filtered - just place a couple small caps across the
feedback resistors to form pole.

Since the OP is starting with such a high differential
voltage that is probably quite an acceptable solution.