Accurate measurement of signals

[seyyah]

I'm working on dc motor control, some simple controls or dc/dc converters etc. But obtaining accurate signals from the designed system is the most difficult part according to me. All my designs are µcontroller based. The signals which i measure are sometimes an adc reading of voltage or current, sometimes an optical encoder reading, sometimes only a switch or button. But generally all of them. I'm using simple techniques like RC filters to suppress noise. But generally it doesn't suffice. So i'm using some delays or taking several measurments in software. But this reduces sensitivity and accuracy and prevents me to make good controls. There must be a common or formal way of this job. Can you please suggest me something so i can overcome these difficulties.

 

[nicleo]

Is your system on a well-designed PCB or just on prototyping board?

I encountered such problem before. But I believe, if I could (which needs $$$) place the analog ADC circuit on a well-design PCB, the noise of currenct or voltage sensing will be reduced.

Do you mind to share your system specifications (e.g. speed of your microcontrol, resolution of your ADC circuit, base speed of your dc motor, type of control, etc)?

 

[IanP]

Try active filters for analog signals and take, say, 10 readings and average them in software.
Digital signals: RC fiter first andl conditioning gates (schmit) to sharpen edges. Sometimes optical isolation also helps. 8)

 

[pico]

To avoid or minimize signal conditioning or processing for noise reduction, I think you have to use the shielding and grounding and component arrangement and wiring knowledge to protect or avoid interference.

 

[echo47]

Can you see the bad signals on an oscilloscope? Do you recognize what's causing the bad signals?

I'm guessing your power switching circuits are somehow inducing glitches into your measurement inputs. You may need better isolation, better grounding, smaller ground loops, or something like that. Difficult to guess without seeing your project schematic and physical construction.

In general, it's smarter to reduce the interference than to filter the measurement inputs.

 

[seyyah]

Actually my question is not for one project i asked it for general. Because i face the same difficulties everytime. I may solve the problems by trying for my current project but in future projects i'll face them again. So my next project must be to learn this subject: How to reduce noise, how to measure accurately and fast enough etc.

After several experiences i think i design better pcbs. But all my work is a bit practical. I want to learn more deeply and be more sure when i do something. For example i use a simple low pass filter made by using R and C. I calculate the frequency that must be allowed and not allowed. And i use it in my system but it doesn't work. May be the calculation is right but other factors affects it and when i redesign it for 10 times of my frequency, it works but not still very good.

I think you should've understand what my problem is.

 

[nicleo]

For example i use a simple low pass filter made by using R and C. I calculate the frequency that must be allowed and not allowed. And i use it in my system but it doesn't work. May be the calculation is right but other factors affects it and when i redesign it for 10 times of my frequency, it works but not still very good.

RC filter is 1st order filter, meaning the amplitude after cut off frequency drops at the rate of 20dB. It's also important in the selection of capacitor. Usually low ESR type should be choosen for filter design. Of course higher order filter will provide sharper attenuation, but it tends to introduce more delay.

What kind of motor have you controlled before? If you're interested, we can start discuss on one type motor first. I'm also interested and wish to learn more too.

 

[seyyah]

This is my first motor control and it's a pm dc motor. my system is not very complex so closed loop control is not a must. At the beginning i planned to complete it with open loop than go into closed loop. But i faced difficulties in loop when a measurement is needed. I solved them but the solutions are not good enough for a closed loop control. Because if i want to measure accurately it takes ltoo much time, if i want to measure fastly than the accuracy is bad. I don't know if there are advanced software techniques but i think it must be solved by hardware mainly.

 

[nicleo]

This is my first motor control and it's a pm dc motor. my system is not very complex so closed loop control is not a must. At the beginning i planned to complete it with open loop than go into closed loop. But i faced difficulties in loop when a measurement is needed. I solved them but the solutions are not good enough for a closed loop control. Because if i want to measure accurately it takes ltoo much time, if i want to measure fastly than the accuracy is bad. I don't know if there are advanced software techniques but i think it must be solved by hardware mainly.

Interesting ... probably we could start from pm dc motor.
1) Is your pm dc motor 'brushed' or 'brushless' type of dc motor?
2) What is your application?
3) What is the controller (microcontroller or DSP controller)? What is the processing speed of your controller?
4) If close loop, what kind of control you wish to apply? Speed control? Torque control? Or, position control (servo application)?

I think both, i.e. hardware and software, can be improved to measure more accurately. Noise (inteference, ground loop, etc) reduction is important, especially from the power switching drive.

 

[seyyah]

Interesting ... probably we could start from pm dc motor.
1) Is your pm dc motor 'brushed' or 'brushless' type of dc motor?
2) What is your application?
3) What is the controller (microcontroller or DSP controller)? What is the processing speed of your controller?
4) If close loop, what kind of control you wish to apply? Speed control? Torque control? Or, position control (servo application)?

I think both, i.e. hardware and software, can be improved to measure more accurately. Noise (inteference, ground loop, etc) reduction is important, especially from the power switching drive.

1) Magnets are on the stator side. I don't excite the phases. I only control the voltage by pwm. Is it pm brushed motor? I think so.
2)I plan to use it for opening/and closing a door(or a similar thing. )
3)It's 20 Mhz pic ucontroller (5 mips)
4)I would apply speed control. For this reason i was planning to estimate the speed from the back emf but when i returned to open loop i used a homemade (not suitable for closed loop speed control) optical encoder to obtain the position of the object(door etc) to determine the slow down point.