Hi, Priyank. I'd still like a bit of information on the class you're
in -- it's hard to give practical advice without knowing where you're
coming from. From the message string I'd guess you're enrolled at VU,
but I'm not clear if you're in a TAFE sequence, the standard EE
program, or a biomed undergrad.
Again, this isn't hard -- at least to get a start on things. You've
got a sensor and two ICs, one mounted in the pressure cuff (IA) and the
other mounted close to your DAQ board (filter/amplifier op amp).
You've got the standard IA configuration for a gain of 100, and you can
cut&paste the filter from Freescale AN-1571, Fig. 1 (V(offset) = GND if
you're using a split supply and an LM741)
The question of electrical noise is always a tough one. There are many
ways to work on this problem, which is central to a lot of biomed
applications. From the way I read your post, it sounds like you've
just hung a cap at the DAQ and filter/amp input -- I'd like to offer
the suggestion that having the instrumentation amp drive a cap directly
is just so wrong.
| .------.
| | | UNWISE CHOICE! |
| | +10Vo--------------o--------------------------o +10V
| | | | |
| | | |
| | | | |
| | | o------.
| | | | +| |
| | Vo+o---o------' | ---
| | | | |10uF --- |
| | | .--. | |\| |
| | | | | '-|-\ AD | |
| | | .-. | | \ 622 |
| | | | | '---o \ | |
| | | | | | >---)--------------o----o Vin
| | | '-' '---o / | | | To DAQ and
| | | | | | /| | +|22uF Filter/Amp
| | | '--' .-|+/ | | | ---
| | | | |/| | | ---
| | | | | | | | |
| | | | | | | |
| | | | | | | | |
| | | | | '--o-o--------------o-----o GND
| | | | | +| |
| | | | | ---
| | | | | ---
| | Vo-o---o------' | |10uF |
| | | o----'
| | | | |
| | -10V o--------------o--------------------------o -10V
| | | |
| '------'
| Freescale 4 Conductor
| Pressure Shielded
| Sensor
|
(created by AACircuit v1.28.5 beta 02/06/05
www.tech-chat.de)
Let's count the ways this setup is asking for problems. To start with,
while the AD622 is current limited and can handle an indefinite short
circuit, it's not made to directly drive a capacitive load. You're
almost guaranteed to have the IA oscillate like heck. It will also get
hot with the current spikes while it's oscillating, which is going to
degrade its performance pretty severely if it lives. Mercy on the poor
Analog Devices chip? Sadly, no.
Now the power supply current is going to be bypassed at the IA by two
tantalum caps. This, again, is a good thing. But you're also going to
have the power supply current variations impressed on the GND line
coming back to the DAQ board. If you're looking for a millivolt-level
signal, this is a prescription for problems. Having the bypass caps
there will just tend to even out the spikes, and also lower their
frequencies, bringing them closer to your frequency of interest. This
becomes a bad thing. It gets worse when you start looking at the poles
in the feedback loop of your power supply ripple. This also increases
the noise, and the chances of oscillation.
If I were doing this, the first thing I'd say is that you have to
separate the power supply lines from the signal. That would mean a
separate 3-conductor for the split supply, and a two-conductor twisted
pair shielded for the signal. I would connect the two-conductor at the
IA itself (pin 6 +, pin 5 -), and have that go directly to the DAQ and
filter/amp board. At the filter board, I'd just use a load resistor at
the input, and use a diff amp to separate out the signal itself from
the noise generated by the power supply lines, having that output go to
the Fig. 1 filter.
This should at least be a start on getting where you want to go:
| .------.
| | | WISER CHOICE |
| | +10Vo--------------o--------------------------o +10V
| | | | |
| | | |
| | | | |
| | | o------.
| | | | +| |
| | Vo+o---o------' | ---
| | | | |10uF --- |
| | | .--. | |\| | .---->To DAQ
| | | | | '-|-\ AD | | |
| | | .-. | | \ 622 | |
| | | | | '---o \ | | | |
| | | | | | >---)-------------------------o---->+
| | | '-' '---o / | | | |
| | | | | | /o----)------------------. .-.To Diff Amp
| | | '--' .-|+/ | | | | | 2.2K| |then Filter
| | | | |/| | | 2-conductor | | |
| | | | | | | | Shielded | '-'
| | | | | | | | |
| | | | | | | | '------o---->-
| | | | | '--o-o-------------------o GND
| | | | | +| |
| | | | | ---
| | | | | ---
| | Vo-o---o------' | |10uF |
| | | o----'
| | | | |
| | -10V o--------------o--------------------------o -10V
| | | |
| '------'
| Freescale 3 Conductor
| Pressure Cable
| Sensor
|
(created by AACircuit v1.28.5 beta 02/06/05
www.tech-chat.de)
Note that the twisted pair negative wire should be connected as close
as possible to pin 5 of the AD622 IA. Of secondary importance would be
keeping the trace between pin 5 and the power supply GND as low
impedance as possible.
The diff amp prior to the filter will do the most to knock out noise
(any antenna pickup should be impressed equally on both lines, and will
be cancelled by the diff amp). The voltage drop caused by the 2.2K
resistor is a linear function of the current, and while it will tend to
attenuate the signal a little, it shouldnt affect the basic signal
fidelity.
In the "real world", of course, this circuit has the deficiency of
being pitifully unprotected from ESD and the other challenges of
putting electronics in the hands of people. Op amp inputs and outputs
floating in the breeze, indeed. But the main difference between this
circuit and anything you'd make for real would be the utterly essential
requirement of providing galvanic isolation between the patient and
your electronics. That changes a lot.
Just for yuks, a biomed engineer would probably be looking at something
like using a galvanically isolated DC-to-DC converter at the box end,
and then use a precision optoisolated V-F conversion to read the gross
pressure (DC level), and a linear isolation amplifier to cull the
millivolt signal from the DC level. Added expense and complexity --
you can see how doing this stuff in a medical environment can add
costs. Also, you would probably want a dedicated DSP to do
post-filtering on the oscillometric information, which makes this a
real project and a half. But if this were a free-standing box, the DSP
could also do the pump control, user interface and display, too --
there's not much to do while the conversion is taking place, and many
DSPs have ability to spare. And if that isn't enough, you'd have to
look at radiated emissions (stringent medical requirements to avoid
interfering with other kewl stuff).
Feel free to post again if you've got problems, but I'd guess that the
WISER CHOICE setup above (cost: 1 additional op amp for the diff amp,
and a second 2-conductor twisted pair shielded wire) should be pretty
much what the doctor ordered, at least for your classroom setting.
Good luck
Chris
