OpAmp PID controllers?

[DennisCNC]

Any one here build a opamp based PID controller for some analogue control project?
What Im working on is, will have 0 to 75mVdc for feedback signal this will come from a current shunt resistor that a 90Vdc motor is connected to. The output signal from the PID controller will be 0 to 10Vdc to control a voltage to air pressure regulator/transducer. And have a POT to set the value it needs to hold as the load on the motor changes the controller needs to compensate to maintain set amp load value by raising or lower the air pressure.

I did find quite a bit on google but most of it is temperature based.

Any tips welcome!

Thanks

[keebler303]

How quick and/or good does the control need to be? I ask because you can simplify things drastically by utilizing just a P (proportional) controller. You will lose a little bit of the nice quick response of a PID controller but save yourself a lot of headaches. I am not sure how much you know about PID control but the biggest problem with making an analog PID controller is the D (derivative) part. Any real world signal has noise. The noise generally looks like this: /\/\/\/\/\/\/\/\/\/\/\/\. When you take the derivative of that signal you get something rapidly jumping from very positive to very negative. This can wreak havoc on your process control. To avoid this you have to add low pass filtering to your signals so that yo can smooth out the noise.

I'll stop for now, if you have more questions, ask away.

Matt

[keebler303]

Here is some homework

http://www.ecircuitcenter.com/Circuits/pid1/pid1.htm

The circuit for a PID controller is unchanged no matter what it is controlling. The examples you have found for temperature are just the same as what you need, you will just have a little bit different signal conditioning before you run it through the PID controller.

Something like an amplifier to scale your 75mA signal to 10V and then a difference amplifier to calculate the error between the measured current and the POT setpoint. This error is what is fed into the PID controller.

Matt

[DennisCNC]

Matt,

Thats for the help, I think for this project just the P will be plenty as the response can be slow +/- .5 seconds. The load change will happed gradually on the motor.
So your saying to scale the 75mV signal to 10V and then use it for the opamp input, this way it will be easier to have the 0-10V output signal?

What I know about PID loops is when it comes to servo motor tuning, just change the numbers up and down till you have a good response

Dennis

[keebler303]

75 mV is your full scale input. To get a 0-10V output, you need to do one of two things:

1) Scale your 0-75mV signal up to 0-10V then compare with your 0-10V command signal.

2) Use a 0-75mV signal as your command and compare it with your 75mV current feedback signal. Then after you perform your comparison between where you want to be (command) and where you are now (feedback), you can amplify the error up to 0-10V.

I don't know enough to say which would be a better route but I would probably go with scaling the input and then doing everything else on a 0-10V scale. It is easier to see small changes on a bigger voltage and your signal to noise ratio would be better.

You might also need a low pass filter on your feedback signal so that the noise does not get amplified with the signal.

When it comes to tuning, P does the hard work. More P gain makes it respond faster at the cost of overshoot. D slows the response to reduce overshoot. I removes steady state error. The link above explains it pretty well.

Matt

[DennisCNC]

Matt,

I was looking at the feedback signal today with a scope and it looks pretty clean very little noise. What does a schematic look like for voltage scaling with a opamp, Im googling for a while and not sure what its supposed to look like.

I haven't made much analogue projects, I mostly mess with digital stuff.

Thanks

[keebler303]

Dennis, you'll have to check for noise after you amplify it. To get a 10 volt signal you need a gain of 133. That means that your noise will be amplified by 133 times as well as your signal. It still may be ok, I guess you can cross that bridge when you get to it.

What you need is a non-inverting amplifier with a gain of around 133. Google is littered with circuits for these, here is one, with the equation necessary to find the correct resistors.

http://www.play-hookey.com/analog/no...amplifier.html

Matt

Afterthought: If one side of your current shunt is not at 0V potential, you will have to use a differential amplifier. Hopefully your current shunt is tied between ground and your motor negative so that one side is ground referenced?

[Andre' B]

Do you even need an opamp?
Figure 20 on page 6 shows a single transistor motor speed control I have used several times.
http://www.sdp-si.com/D757/pdf/D757P367.pdf

I have used it with both BJTs and MOSfets, whatever was in the junk box.
It does require the use of a motor with an analog tachometer so not sure if it will work with what you are doing.

[keebler303]

I don't think that will work. His feedback signal is current, not speed. I am quite sure that you could use some other circuit using only transistors. The way I know how to do it, and I think the easiest way, is to just use op amps. Op amps are simple, versatile, have high input impedance, etc. etc. To realize a proportional closed loop control with only discrete transistors, you would have to call in an electrical engineer, not such a simple task.

Matt