I'm a student of bioengineering at the University of California, San Diego, and I've been presented with a problem in the control of the partial pressure of oxygen in a nitrogen-oxygen gas mixture.

The basic structure of the problem is that we're doing trials using anesthetized fruit flies under hypoxic conditions. We have a PPO sensor already installed and communicating with our computer program via a DAQ board; however, as much as our predecessors have tried to automate the process, it remains that someone (usually a hapless undergrad) needs to sit with the machine while it's running and manually adjust the PPO as it drifts off of true--usually pretty rapidly and usually through a range at least a couple of points away from our target value.

We've already begun investigating PID controllers (and theory) under the guidance of our advisor, but to be frank, we're a little in the dark (and a little scared) of all the information we have to process and integrate. We HAVE found some actuated systems used for industrial applications that may be able to do what we need--but either they are too expensive, too large, or both. All we really need is a very small system for a very limited application--so where should we start looking, in terms of companies and manufacturers who might have the required materials?

Thanks so much in advance!

 

I'd suggest starting with on-off control, for simplicity's sake, rather than PID control

ON-OFF control:
Just like your thermostat at home. If the temperature (process variable or PV) is below the thermostat setting (setpoint or SP), the thermostat "calls for" heat. Once the temperature (PV) increases to match the SP, the thermostat tells the furnace to turn off. The furnace is either full on or full off. The ON-OFF controller is the switch that tells it to turn on or off.

In your case, you condition (get the signal compatible with the controller input) and feed your PPO signal into an ON-OFF controller. You establish a setpoint (what you want) on the controller and the controller will compare the two and turn its output on or off.

The output is a switch (ordinarily a relay; power for the valve gets connected to the contacts of the relay. The controller controls the coil of the relay). You connect the controller output to a power source and the controller switches power on or off to a solenoid valve.

A solenoid valve can be either NO (normally open) or NC (normally closed). NO means when the valve gets power, the valve closes. NC means that when the valve gets power, the valve opens. It seems backwards, but that's the terminology. You pick the valve for what you need to have happen when the valve doesn't have power. Does it need be closed or open when there's no power applied to it? For instance, if you want to maintain the pressure (valve closed) without applying power to the solenoid, then you'd have an NC valve, Normally Closed without power.

Usually the controller can be setup for direct or reverse action, which flips the output mode, so that if the output mode valve is opening when it should close, then the controller's action gets changed.

On-off controllers look like this:
http://www.lesman.com/unleashd/catalog/control/control_udc2500.html

You don't mention how you manually maintain the pressure now, probably with a needle valve on a pressure line from a regulator on a pressure tank. I suspect you'll need a tiny little direct acting solenoid (don't use a "pilot operated" valves that require a pressure drop if you a supplying gas to maintain pressure.) Something like
http://www.lesman.com/unleashd/catalog/valves/valves_8262.html

Bud

 

Thank you for your help! That does sound like it would be a great deal easier to implement than the PID controller (which would not necessarily be my first choice going into this, as I know <i>nothing</i> about the theory behind them; but our advisor was sold on them, so it was my first stop for research), and probably a lot cheaper, too. I'll try to get more information on the exact problem he's trying to solve, then see what we can do about getting a valve for it. (Probably NO--as we have more problems with the pressure plummeting than we do with it creeping over our target value.)

Thanks again!

Kim