Re: [iPAQ] Help : accessing the extension port (FAST IO mode) of the ipaq

Valentin Longchamp wrote:

>
>>
>> Control loop as I used it means: read sensors, calculate actions,
>> apply outputs.
>>
>> Reading the encoders, in my purposed solution, should be done by the
>> microcontroller (at 5KHz or something like that) and the *total count*
>> should be sent to the iPaq at 50Hz.
>>
>> The iPaq should never have to do low-level stuff like reading encoders
>> or generating PWM's.
>>
>> The iPaq should be doing more mid-level (the encoder on motor A moved
>> 23 ticks and on motor B 16 ticks so my current position is <X,Y>) and
>> high-level stuff ("where should I go", "build world map", etc.).
>>
>
> Sorry, I need to precise a little bit : there a PLD counting the "ticks"
> through the Chan A and Chan B from the motors and which increments the
> motor position.
>
> But we want to control the motors with the ipaq on the mid-level as you
> call it : we want to "emulate" a motor controler like the LM 629 with
> our software (the rest of the team doesn't want to use hardware motor
> control, don't ask me why :(), so we have to control the trajectory of
> our robot with the ipaq !!!! That's why I need a big refreshing on the
> motors. We have also noticed that with a 50 Hz (as you propose) refresh
> rate to calculate position, at our speed, the error is too big and we
> have a very big error afer 30 seconds ... that's why we need a big
> refreh rate !!!!

No, no. The control loop frequency only has to do with the "system bandwidth".

I also cooperate with the 5DPO robotic soccer team from the University of Oporto
and we have robots that go at 1.5 m/s with incredible accelerations, controlled
at merely 25Hz.

If you want to increase your odometry precision you can simply let the
microcontroller take more samples from the odometry counters.

For instance, if you want to sample the encoders at 1kHz and have a 50Hz control
loop, you can sample the encoder at 1kHz, store in a local array in the
microcontroller, and at 50Hz send the resulting 20 entry array to the iPaq.

Then the iPaq con integrate the result with 1ms step time to give a more
accurate positioning.

> For sure, we could have done this as you propose it, but for a obscure
> reason my team didn't decide to do so !!!!

Anyway, I wont argue with you anymore, and instead will try and help you go
where you want to go :)

I searched my "personal" archive and found posts to some useful links:

  - http://www.hp.com/developer/iPAQ

      Mechanical drawings, lots of info

  - http://www.imec.be/reconfigurable/gecko.shtml

      This is link to someone that already did something similar to what you
want. Although their project is bigger than what you probably want to build,
maybe you can get some useful information.

  - http://www.foxconn.com/products/ipaq.asp

      You can get some parts here (although there are probably several other
places), like the 100 pin connector

I hope this helps,

-- 
Paulo Marques
Software Development Department - Restinfor, Lda.
Phone: +351 252 290600, Fax: +351 252 290601
Web: www.grupopie.com
"In a world without walls and fences who needs windows and gates?"