IDG500 scale factor

[ramee]

Hi;

I am building a simple IMU with IDG500. I am using 2mv/deg/s outputs from gyro. Problem is when i integrate it i did not get the correct angle. It seams i need to multiply ADC values by about factor 2.1 to get the corret values given by accelerometer. What can be the problem for this. I am using microchip 33fj128MC804 as the processor.

Thanks.

Rameesha

[Tom]

Hi Rameesha,

I really need more info to answer that. What voltage is your dsPic and your IDG500 working with? How is your ADC returning the samples? Signed integer, unsigned, 12-bit, 10-bit, ...?

Is your "dt" correct when integrating?

Tom

[ramee]

Hi Tom ;

My dspic working at 3.3 V .. and using 12 bit ADC for sampling at 1kHz. Compleate kalman calculation runs at 1ms loop (ie 1kHz) ... My sensor board is sparkfum 5DOF IMU (new one) with onboard regulator 2.8 V for IDG500.. Yep i double checked the dt it is 1 ms...

I see the drift (after integration) of IDG300 is too high .. about 1degree per second and drift is seams to be depends on the Gyro orientation. High angle high integral drift..Is this normal...

And also what is the Auto Zero pin of the IDG500 do exactly.. Do we need to use that pin for correct IDG500 interfacing to DSPIC..

Thanks

Rameesha

[Mitch]

The sparkfun sensor boards have a low pass and a high pass analog filter built on the board. What you are describing sounds like the board still has both filters active. Check the spark fun forums. You may need to make some hardware modifications to your sensor board. I'm not sure what the "new board" is today but several of their boards have had this problem. The low pass filter will hold the output cap charged and create an artificial bias. It is impossible to get consistant integration with this filter.

[pizza]

mitch i think you refer to this schematic. there is only a lpf at the xout and yout...do they cause problems?

ramshee can you post the equations that you used to convert your gyroscope values?

this is what i did
gyroX = ((raw_gyroX-gyroXzero)*3.3/4095)/2*1000;
gyroY = ((raw_gyroY-gyroYzero)*3.3/4095)/2*1000;

using a 12bit adc and 3.3V for adc vref...sampling at 800hz so my dt is 1/800.. then i ran a 4th order runge kutta filter before pumping the values into tom's linear kalman filter. using IDG300..

yeah im totally interested in the functions of AZ PTATS and VREF pins of IDG500..like how can they correct the bias?
from the datasheet, we need to pulse(rising edge, more than 2us less than 1500us) the AZ pin for the function to activate, from my understanding, it forces zero rate/bias back to 1.35V, however i do not know how frequently you need to do that, as i have not received my idg500. anybody tried?

[Mitch]

That board should be ok with only the low pass. Some of the boards had both passive lpf and hpf in series. For that case most people have had success by removing the hi pass.

Image1.jpg (80.79 KiB) Viewed 13546 times

[ramee]

@pizza , my equation for gyro x and y are same as you except i put the calculated values in to the equation.

X_angle_dot = (ADCValueGp - 1664) * 0.40283;
Y_angle_dot = -(ADCValueGr - 1664) * 0.40283;

1664 is the zero rate gyro values.

i am using same 3.3 V for ADC ref and 2^12 bit resolution .. dt is 1ms... 1Khz sampling rate.... but when i turn a specific angle say 60 degrees my integration gives about 40 degree so i am putting dt as 2.1 to correct the error.

[pizza]

are X_angle_dot and Y_angle_dot signed integers?

[ramee]

yes they are signed integers..

[pizza]

ok sh!t my sparkfun 5dof(idg500) had just arrived yesterday. it is experiencing the same fate as yours ramee. ;_;