Could use some input on a formula or any other
experience.......
Using a GyroChip-II "Rate-Sensor" and performing
an A-D (12-bit). The data is combined with another
16-bit source (joystick-output) to produce a command
position value. The value is then sent over to a PMD
(Performance Motion Devices) to operate the motor.
The joystick serves to position the axis, while the Rate-
Sensor acts to stabilize the axis (vibration, etc.)
I'm having a difficult time determining what calculation
I could use (within a PIC16F873) .......in assembly lang.
>>
Using a GyroChip-II "Rate-Sensor" and performing
an A-D (12-bit). The data is combined with another
16-bit source (joystick-output) to produce a command
position value. The value is then sent over to a PMD
(Performance Motion Devices) to operate the motor.
The joystick serves to position the axis, while the Rate-
Sensor acts to stabilize the axis (vibration, etc.)
I'm having a difficult time determining what calculation
I could use (within a PIC16F873) .......in assembly lang.
Any algorithms out there for such a task ?
<<
One of my current projects envolves dealing with real time gyro and
accellerometer data on a 16F877. I do various analysis and filtering on the
data both in integer and floating point. Beyond that I don't know what to
tell you because I don't understand your question. Perhaps you could be
more specific.
********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, olinKILLspamembedinc.com,http://www.embedinc.com
>
> >>
> Using a GyroChip-II "Rate-Sensor" and performing
> an A-D (12-bit). The data is combined with another
> 16-bit source (joystick-output) to produce a command
> position value. The value is then sent over to a PMD
> (Performance Motion Devices) to operate the motor.
>
> The joystick serves to position the axis, while the Rate-
> Sensor acts to stabilize the axis (vibration, etc.)
>
> I'm having a difficult time determining what calculation
> I could use (within a PIC16F873) .......in assembly lang.
>
> Any algorithms out there for such a task ?
> <<
>
> One of my current projects envolves dealing with real time gyro and
> accellerometer data on a 16F877. I do various analysis and filtering on the
> data both in integer and floating point. Beyond that I don't know what to
> tell you because I don't understand your question. Perhaps you could be
> more specific.
It sounds like a controls question. Although that depends on why the PIC
is processing the inputs, which isn't very clear.
You're right, I left out alot of details, because much
more is happening with the PIC's, that's not related
to the data-integration for producing the position
error.
The PIC utilizing the equation (or in need of one) is
actually working within a Master/Slave topology.
It happens to be the Slave and handles USART
& SPI data, as well as performing the integration.
The joystick data is received at 19.2K baud over
an async.-serial-port, using a 4-byte packet. The
packet breakout is not important here, but it does
contain 16-bits of "position-information" that is used
to operate "one" axis through a PMD-Chipset.....
which then drives (via Sign-Mag) an LMD18200
H-bridge (as a Servo-motor).
The rate-sensor is located remotely and uses a
GyroChip-II Sensor, conditioned via 3rd, order
Butterworth (100hz. cutoff) and then passes to
a MAX-187 12-bit A-D. We sample at 6250hz.
and the Slave gathers SPI data at 1.2mbs
Both the Rate-Sensor data and the joystick data
must be integrated to provide a single "position
error". With the outcome being "two" independant
error sources, which when integrated have an
equal effect on the Servo's position error. A scaling
of the two errors is necessary too.
Hope this shines more light.........it's about all that
I can disclose.
My "main" problem is lacking the ability to develop
assembly routines that perform much more than the
"basic-four", within a narrow time-frame of ~ 10 to 15
uSEC. while operating at 20mhz. clock.
Also, I'm one of those that happens to be "C-challenged",
meaning ....... can't easily read/write it. Thus, I work with
assembly only. Of course, when my clone is ready, I'll be
sending him to school again......full-time.
So, now we come to the "nutshell"........
**********************************************************************
"I'm looking for assembly based algorithms, that can provide
integration functions and exceed the level of only basic math".
**********************************************************************
Regards,
Steve Kelley
P.S. Where can I pick-up some singing potatoes ?
They sound like easy "pets" to maintain 8^)
I would recommend giving yourself more time. With your anti-aliasing filter
as it is, you have 72dB attenuation at 1600 Hz, which means that a 3200 Hz
sample rate will guarantee that all aliases are smaller than one
least-significant bit.
One more op-amp can get you to a 6th-order filter, and then you can get your
sample rate down to 800 Hz.
All this is assuming that your 100-Hz cutoff actually implies that you are
only interested in data below 100 Hz, which is probably pretty reasonable
for a gyro application.
I would heartily recommend downloading the FilterLab program from the
Microchip website.
When doing signal processing with a relatively arithmetically-challenged
processor like a PIC, step 1 must be to do a little work on the front end to
get the sample rate as low as possible.
protobyte.com
