Circuitry for Tachometer 5

[FeX32]

Hello all,

I am in the process of designing a control system for a DC motor. The motor has a voltage tachometer installed. My question is, what kind of circuitry is best suited to use between the tachometer itself and the A/D interface so that I can use this as an accurate feedback.
I read about voltage followers/buffers. Are these suitable? If so, what are the advantages/disadvantages of these and how can I construct one?
Thanks for any input.

Fe

 

[BobM3]

If you want to convert the frequency of the tach output to an analog voltage then take a look at National's LM2907 chip. Its data sheets will most likely have some circuits shown on it.

 

[FeX32]

Thanks BobM3.
However, the tach's output is in DC voltage in which it is proportional to the angular speed of the motor. About 1V potential per increase of 1000rpm or so.

Fe

 

[melone]

So what is the range on your A/D? What is the range on the output from the tach? If max output is higher than max input, use a voltage divider. If max input is higher than max output, use an amplification circuit.

 

[FeX32]

I am using D-space. So the range on the input to the A/D is +-10V. The tach's range is about +-5V.
Do I have to worry about resistance differences in the tach as compared to the A/D converter?
In what case would I need a voltage follower?

Also, the tach demonstrates regular minor reading fluctuations I believe due to the brushes. Is there a way to normalize this?
Thanks for your help.

Fe

 

[BobM3]

What resolution do you need? If you can live with 1/2 the A/D's resolution then +-5.0 volts is o.k. You may need to protect the A/D against overvoltages.

The A/D should have a high input resistance and the tachometer a low output resistance. This is what you want. No need to worry about resistance mismatches.

The noise on the signal may have to be filtered with a capacitor. I'd try just a .1 microfarad cap on the output of the tach for starters.

 

[FeX32]

Thanks.
I can live with 1/2 the accuracy of the A/D.
I will try the capacitor, see how it goes.



Fe

 

[FeX32]

Another question.

Do I have to worry about drawing too much current from the tach? Will this effect my controller?

Fe

 

[BobM3]

A/D's have megaohm(s) input resistances so you will be drawing only microamps from your tachometer - no problem.

 

[FeX32]

Thanks BobM3.



Fe

 

[itsmoked]

Put a cap across as suggested and add a ~7V Zener.

I'd also add a 10ohm resistor in seres with the tach in front of the Zener and the cap. Since the A/D will have a high input impedance the resistor will help protect the A/D and provide a small amount of filtering. Frankly, I'd use the highest value I could that showed no apparent voltage drop across it. That could be a 1k.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

DC tachs usually can deliver several tens of mA before having linerity issues. So, the standard interface is a voltage divider. Make it 10 + 10 kohms to have the tach full range correspond to the A/D full range.

The smoothing time constant is usually kept at 10 - 50 ms, depending on mechanical issues mostly. Since your voltage divider has a 5 kohms internal resistance, your capacitor to GND shall be 2 - 10 microfarads.

Same reasoning if your tach voltage is 200 V - or any other value.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[IRstuff]

Why does he need a divider? His tach range is +/- 5V and his ADC range is +/- 10V

TTFN

FAQ731-376

 

[FeX32]

Thanks for the replies.
Its nice to see that everyone is along the same lines and can add to each other.
I roughly drew up the diagram that I interpret from itsmoked's reply.
Please let me know if this is what you guys mean.

P.S. won't adding the resistor reduce the current in the line? Or is this negligible compared to the added security and filtering it provides?

Fe

 

[BobM3]

Not quite right with the diode. Put it in parallel with the capacitor across the A/D. Make sure it is a zener as Keith said. The "arrow" of the zener would point to the + output of the tach.

It's not a plus/minus 5 volt output from the tach, is it?

Another thought, are you doing any frequency domain work (Fourier Transforms) after you digitize the data? If so the filtering requirements go up.

 

[Skogsgurra]

Right IR, I got it "umgekehrt". No divider.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[FeX32]

Thanks for the correction BobM3.
I will put the zener in parallel with the capacitor and the A/D.
The tach is just a small DC motor generating current at +5V when the shaft spins in one direction and -5V when the shaft spins in the other.
I will not be doing any frequency work on this signal as I am planning to just use it as a feedback for the speed controller.
However, this motor is only a part of the setup and I will be required to analyze some other data in the frequency domain.
Is a low-pass filter sufficient for this?



Fe

 

[BobM3]

For +/- voltages the single zener will not work. You could put 2 zeners (each 7 volts) in "anti-series" (pointing in opposite directions) or look at using a MOV.

If you are doing work in the frequency domain you may need a better filter than the simple RC filter proposed here. It's possible you may need an anti-aliasing filter (real good low pass filter). It might be included with your card though - look at the specs for the card and see if it is mentioned. Also, the zeners might be included with your card. Again check the specs to see if they are offered as protection for the A/Ds.

 

[FeX32]

Thanks, I will look into a MOV. Although, I should have mentioned that my application will only require me to run the motor in one direction.
From the specs of my card I can see that I doesn't have an anti-aliasing low pass filter. I'm not sure about the zeners though.
I tried the RC filter today, and I do get fairly ok results. Thus far, I have only simulated it with a PI controller. However, the voltage still bounces vary little. I will see what I can do about the small bounces.
Thanks again,

Fe

 

[PHovnanian]

Where are you seeing these 'bounces'? In the digital data following the ADC? Its possible that your tachometer is producing some ripple (you could check it with an oscilloscope) at a frequency above one half of the ADC sampling rate and you are getting some aliasing. If so, you'll need a lower time constant RC filter (bigger C and/or R).