a race car system development

Yikes, a pic18, use a PC for developement and implement using a PC104 card

While it is possible to run a kernel on a PIC18 (FreeRTOS.org has a PIC18 port for example), it is a bit of a round peg in a square hole. There are plenty of other low cost development environments you can choose that would give you more freedom - if you are set on Microchip then take a look at their 16 and 32bit offerings.

What is your motivation for choosing the PIC18?

--
Regards,
Richard.

+ http://www.FreeRTOS.org & http://www.FreeRTOS.org/shop
17 official architecture ports, more than 6000 downloads per month.

+ http://www.SafeRTOS.com
Certified by TÜV as meeting the requirements for safety related systems.

Don't decide on the answer (a single PIC18 + RTOS) before you have a good handle on what the question is.

It's natural (and reasonable) to tend to use processors and solutions that you're familiar with; helps to cut down on the time needed to get up to speed in a new environment. If you want to go the microcontroller route, though, I'd suggest using many, specialized, cooperating uCs rather than trying to do it all with a single one.

Data acquisition. What sensors? How many? How interfaced?

CAN proto. Is this roll yer own or do you have to speak to CANOpen, CIA, CAN Kingdom, ...?

USB data logger. Does it have to be a USB stick or can you use something like an SD card? The latter is somewhat easier to implement on a uC, although Atmel has some USB-flavored AVRs that work well as small hosts.

Gear shifting. Again, what sensors? What actuators?

Telemetry. Range? Data rate?

ECU. Any interfacing required?

Dashboard controls and displays. What type? How many?

--
Rich Webb     Norfolk, VA

I hope you don't plan on exceeding about 20 mph with the race car.

If it is possible then please guide us for the system centralisation

Only "centralize" the parts that need to be centralized. In particular, make sure that the data acquisition system lives even if something else dies.

They got free samples.

>

Maybe you should reconsider a single MCU. This is not the first year the Formula Student Cars are beeing built. I know of projects using up to 12 AT90CANUSB128.

--
Best Regards,
Ulf Samuelsson
This is intended to be my personal opinion which may,
or may not be shared by my employer Atmel Nordic AB

I apologize for replying so late!! I participated as junior team member in the competition last year few days after my first post and saw the systems of other teams , no team centralized their systems ever so even I am thinking of leaving the idea of systems centralization and I am running short of time for that. After that I was more into implementing all the basic stuff required for different modules and learning a lot for the same!

(Till date I have got good success in CAN communication and GLCD for the dashboard, data acquisition has about 10 to 11 sensors which I have implemented satisfactorily)

But whatever little I could read and learn about RTOS in all these days I think it would make sense to implement it in Automatic gear shifting system.

Now what I aim to implement here is :

1) This systems will use data from two important sensors required for gear shifts Gear Position Sensor and RPM sensor.

2)Since the system has two sensors in it I intend to use it also as a node for the CAN network in the car.(the CAN network has PIC18f and mcp2551 for the physical layer , am not using CAN kingdom or CANopen and just communicating between different nodes by CAN protocol)So it will also send the sensor data whenever requested by the LCD controller on the dashboard.

3)It will "continuously monitor" the RPM from the engine(for this there is already a VR sensor on the engine ) and make upshift or downshift according to the algorithm .

4)I am using solenoid actuator for the gear lever and for clutch I haven't got any actuator yet but mostly I will be using a DC motor with threaded shaft for linear gradual actuation which is required for clutch.

5)If I use the gear actuator (which i think is requisite for making the system fully automatic)then I also intend use a potentiometric type sensor to give feedback to the controller about the actuation of the clutch (I dont want to just depend on the delay routine in the code and assume the that the clutch lever must have actuated)

Please tell me that will it be a good idea to implement a RTOS here (or something like scheduler) since the more I work on the algo and think about all different possibilities of automatic gear shifts in a race car I feel that all the actions must be taken by the controller at critical timings.

Or should I just stick to serial method of checking the RPM and shifting and then again start checking it ?

I am using mikroC these days but I am ready to shift to mplab anytime if it is required.

(please keep in mind while answering that am very new with RTOS and know only about some basics stuff from wiki, rest I have a got good experience of PIC18f458 in all these days)

If you want to post a followup via groups.google.com (or anywhere else), ensure you quote enough for the article to make sense. Google is only an interface to Usenet; it's not Usenet itself. Don't assume your readers can, or ever will, see any previous articles.

More details at:

--
 [mail]: Chuck F (cbfalconer at maineline dot net) 
 [page]: 
            Try the download section.

You may want to feed other information into the shift selection like accelerator and braking. There was a lot of material published on the Saturn automatic transmission shift algorithms although the implementation used fuzzy logic it identified what was important.

I am not sure your clutch proposal will be fast enough.

You can use a RTOS but the software in production cars is using event driven software either using interrupt driven threads or a software implementation of the same thing.

Regards,

-- Walter Banks Byte Craft Limited

what is usenet?

--
Rich Webb     Norfolk, VA

Doesn't something need to be done with the accelerator? Shifting gears with the throttle wide open would not seem to be a good idea.

Formula 1 cars seem to all have some sort of automated shifting process which is push-button controlled. Hasn't anything been published about that?

... snip ...

This system is Usenet. You are interfacing with it using googles flawed reader system. I am interfacing with it using another reader and a news-server system. Usenet connects news systems worldwide.

Some free news servers. I use motzarella and gmane. (1 time charge) (free) (free) (free) (free) (free) (via://dotsrc.org/usenet/)(free) (mail-lists via news) (free) (pay)

Accelerator control would be a good idea. Reset the engine speed to match the road speed. There are lots of papers on shifting but the Saturn ones are probably the most interesting because it describes a production car. Look at the patents on this transmission as well.

There is very little published literature on race cars. The push button shift sequencer on them is similar to a motorcycle up/down shifters.

Some of the race cars have anti-slip type accelerators that use a technology similar to anti lock brakes to allow maximum acceleration without wheel slip.

Regards,

-- Walter Banks Byte Craft Limited

Hey, Walter!!!

I agree with Walter on all counts. I don't think an electric motor with screw drive will be anything near fast enough for race-car type shifts. Ok for launching, NG for anything else.

Also, you absolutely need a throttle input. Probably also a brake sensor. The computer needs to know if you're speeding up, slowing down, or if the race is over.

The thing is, race car drivers use downshifts for engine braking to help them slow down for a corner, and also to be in the right gear for the exit.

Maybe I'm missing something. How is the driver supposed to tell the computer if he needs to shift up or down? Does he have a paddle or lever?

Back to the clutch: There _IS_ another option, which is not to use it at all, at least during upshifts. Motorcycle riders often shift without the clutch (the transmission is built to allow it). Just tap the kill switch to remove power for an instant. I think that's the way F1 cars work, also. But you definitely need the clutch for downshifts.

Jack

See also Nissan's new system for the 350Z (?). It sounds great in the ads .

That's right. Formula 1 builders aren't likely to publish their secrets. They guard them more tightly than Oak Ridge does.

Not anymore. F1 banned traction control for 2008.

Jack

As Walter said, race car builders aren't likely to publish their speed secrets.

Re the F1 shifter: It shifts in something like 100 ms or less. I think maybe a _LOT_ less. Each builder works hard to make them shift as fast as possible. At 200 mph, every ms gained is a foot more ahead of the other guy. Over thousands of shifts during a race, that's a mile or so. But those systems are either hydraulic or pneumatic, and with 600 horsepower to drive the pump -- and $100,000,000 to burn -- you can do a lot of stuff.

Jack

It can be done with large diesel-engine vehicles as well during upshifting. Let off the throttle, shift to neutral, wait for the tach to drop to the proper value, engage the next gear. It's not quite so easy shifting down, but it can be done.

I once drove my Saab a few miles without a clutch when the hydraulic system on it failed. It wasn't easy nor pretty, but I got to a friend's house with the crippled beast.

You can handle 'no clutch' on both motorcycles and cars. At least I have. The key is to shift into neutral, match engine speed with road speed for the next gear (either up or down), and then shift. A couple of hundred miles home like this is enough to encourage repairs.

The major problem is getting started in a car. For this, a hill is useful. The bike I could usually push fast enough to dump it into low.

Now I've got a slush box, and no way to handle failures!! :-)

--
 [mail]: Chuck F (cbfalconer at maineline dot net) 
 [page]: 
            Try the download section.