John said:
That's a good point.You could put a transistor buffer on the opamp
output, or, in a one-direction application like this, you could hang a
resistor to the minus supply, pulling 10mA out. Then the opamp could
switch between sinking and sourcing 10 mA.
Do you think using a series resistor with diodes to prevent excursion
more than V+ plus .7 or V- or ground minus 0.7 would prevent that
problem? With a hookup like that, wouldn't an early start of the sensor
supply at worst just try to power up your circuit?
That seems like the best solution if you have 24V available from the
sensors.
Hi, John. Depending on Eric's application, everything you suggested
*could* work -- it all depends on the specifics of his application.
Many sensors with current output are self-powered, and you just take
the current output wire referenced to it's COM wire for the 4-20mA
output. Some require the mA output be referenced to the V+ side (view
in fixed font or M$ Notepad):
|
| VCC VCC
| .-------. + .--------. o-----.
| | | | | | | |
| | +o-------' | +o----------' .-.
| | | | | 250| |
| | | | | | |
| | OUTo-------------. | OUTo ------------. '-'
| | | | | | | |
| | | .-. | | | |
| | COMo-------. 250| | | COMo----------. '--'
| | | | | | | | |
| '-------' | '-' '--------' |
| | | |
| OUT SOURCING o-----' OUT SINKING |
| | |
| === ===
| GND GND
Fig. 1
(created by AACircuit v1.28.6 beta 04/19/05
www.tech-chat.de)
Some have built-in isolated floating power supplies, and just have two
output wires -- you can tie either side to the PC common without worry.
And many aren't self-powered at all -- you apply a minimum voltage
between the two wires, and the sensor powers from that, with total
current draw guaranteed to be 4-20mA, like this:
|
| VCC
| .-------. +
| | | |
| | +o-------'
| | |
| | |
| | OUTo--------o Vo
| | | |
| | | .-.
| | | 250| |
| | | | |
| '-------' '-'
| |
| ===
| GND
|
|
Fig. 2
(created by AACircuit v1.28.6 beta 04/19/05
www.tech-chat.de)
This won't work for a 0-20mA output (they do exist), so it's more
likely it's one of the other types of outputs unless the OP
mis-described his app.
The difficulty comes when interfacing to a PC or other DAQ system where
you have fixed voltages referenced to PC ground. It's trivially easy
to smoke a board input, and I've gotten a lot more cautious over the
years about this particular problem, in part because you occasionally
have to supply a solution before you know all the details (much like
Eric's post).
For the two wire solution shown in Fig. 2 above, if the +15V of the DAC
board is sufficient, he'd need no input protection at all. Many
sensors require more than 15V, though. With a higher power supply
voltage (say, +24V), Eric should just be able to put a 12V zener across
the 250 ohm resistor for protection.
But using a combination of voltage dividers at the input, followed by a
diff amp with enough gain to compensate for the voltage divider, is
good as a universal input which can accept common mode voltages quite a
bit above the power supply rails.
The concept I mentioned above is expressed as a one-IC solution from
Analog Devices. The AD629 is a high common mode voltage diff amp which
has laser trimming to achieve minimum offset and gain error. For a few
bucks per channel, he can have a universal front end that can handle
common mode voltages of up to +/-270V with the +/-15V supply commonly
available on DAQ boards. The low end version of the AD629 specs 1mV
max offset and 0.05% max gain error, which is a small price to pay:
http://www.analog.com/uploadedfiles/data_sheets/225410897ad629_a.pdf
Eric can also implement the high common mode voltage diff amp with
discrete components and standard op amps, just as shown in the
suggested wiring diagram on the data sheet. Resistor matching is a bit
of a pain though, especially with multiple channels. If the OP wants
to go that route (been there, done that, got the tee-shirt), having a
factory cal cycle and/or autocal cycles can help with offset and gain
non-linearities. And after all, the OP didn't specify accuracy or
precision here -- he was just asking. And I don't remember needing
more than 12 or 13 bit resolution on any position or displacement
sensor output signal -- actually, 10 or even 9 bits (0.1% or 0.2%)
would almost always have been sufficient.
If the OP wants to look at the wide variety of current sense solutions,
he might want to look at Linear App Note 105, the Current Sense Circuit
Collection.
http://www.linear.com
Punch in AN105 on search, and right-click "Save Target As..." (long
..pdf file alert). They don't cheat too much toward their product, and
one of those solutions should be right for the OP.
Cheers
Chris
