Analog Filter Design

[jnnewton]

I am planning on interfacing the analog devices ADXL330 imems accelerometer to a PIC uC. I will be moving the sensor at a maximum rate of 18deg/s as a tilt sensor and will need .5 deg resolution. I have a few chioces to make and could use some insight to shorten my trial and error time.

1. The sensor has a ratiometric output. I can get 360 mV per g if i supply the sensor with 3.6V, or as low as 1.95mV/g @ 2V supply. 3.6 V seems to be the way to go for the max resolution. If I am missing anything, let me know.


2. Based on the output of the sensor 0-Vin for +/-3g. I will be using the center of this range (+/- 1g) I am thinking I should amplify only the Area I need and let the rest just saturate at the rails of the op amp (sound good?) This way i get the best resolution. So i take +/-90 deg or +/- 2g or 720mV and amplify to 5V to feed to my PICS ADC

3. So the order of the signal chain should be: sensor, gain amp, filter, uC, right?

4. Because the maximum rate of movement is (i think):
18 deg/s * 1g/90deg * 360mV/g = 72mV/s.

5. i should choose an op amp with a slew rate greater than 72mV/s

6. i should choose a low noise precision op amp,but not in-amp b/c the output from the sensor is not differential..

7. I should set my filter cutoff frequency at somewhere low (i was thinking 100 Hz, should I go lower?)

I am going to put a schematic on here in a few hours, if anyone could reply to let me know i am on the right track, I would appreciate it.

 

[dknguyen]

If you are using it to measure tilt with an accelerometer, the assembly shouldnt be moving. How often do you plan to check it? I'd go for like 10Hz or something for measuring tilt since the thing shouldn't be moving and is just sitting there anyways. I'd use a MAX7401 or MAX7405 and a DS1099 for really low frequency filters. YOu need to sync the filter clock and ADC clock somehow though to prevent beat frequencies.

I wonder what should go before the other, filter or amp?

You should choose a slew rate of 10V/us or more probably so that it can drive your ADC fast enough to sample. Depends on your ADC but that seems to be the standard number. But this is tilt sensing and if it's not very fast then you aren't sampling very fast either so you can have a lower slew rate like 1V/us or something.

 

[jnnewton]

Thank you for your response. I have looked for the part and it is not stocked by digikey, newark or mouser. Are there any other recommendations that you might have for a more readily available active filter? I really like the idea of not having to design a filter and just using one ic. Thanks again

 

[dknguyen]

I think you can get them as samples (don't abuse it though!). I don't know of any others but if you search. If you want to search for more on DIgikey, search for "switched capacitor filter".

 

[jnnewton]

I am unable to find values for op amp resistor components to do what I want and would like to know if it is even possible. I have a signal that is 0-3.6V. I am only interested in approximately the 1.25V to 2.25V range(center at 1.8V) of that original signal and would like to amplify that portion to 0.5-4.5V for input to my ADC using only single supply op-amps at 5V. I can get the signal to center at 2.5V using a difference amp, but cannot then amplify that range to be 0-5V w/o causing a problem (i.e. the circuit below amplifies the range incorrectly)

Difference
V = (V1-V2)*(R2/R1)+Vref(R2/R1)

V1 3.6 2.16 1.8 1.44 0
V2 0 0 0 0 0
R2 1388.888889 1388.888889 1388.888889 1388.888889 1388.888889
R1 1000 1000 1000 1000 1000
Vref 0 0 0 0 0

Vout 5 3 2.5 2 0
This part is good, gets signal centered at 2.5V, now things go bad!

Inverting Gain (Single Supply)
Vout = -(R2/R1)*Vin+(1+R2/R1)*Vbias

Vin 5 3 2.5 1.44 0
Vbias 2.5 2.5 2.5 2.5 2.5
R2 2200 2200 2200 2200 2200
R1 1000 1000 1000 1000 1000

Vout -3 1.4 2.5 4.832 8

Inverting Gain (Single Supply)
Vout = -(R2/R1)*Vin+(1+R2/R1)*Vbias

Vin -3 1.4 2.5 4.832 8
Vbias 3.116 3.116 3.116 3.116 3.116
R2 10000 10000 10000 10000 10000
R1 10000 10000 10000 10000 10000

Vout 9.232 4.832 3.732 1.4 -1.768

 

[ericgibbs]

hi,
This is drawing I did a while ago for a altitude sensor.
I have changed the offset voltage to suit the 1.25V starting point you require.
At less than 1.25V input the output is zero volts, the gain is adjustable to give the maximum range you require of say, +4.5v

It may give you an idea on how to get the adc signal you want.