| ??? 01/14/04 22:02 Read: times |
#62578 - RE: Differential lock controller Responding to: ???'s previous message |
I just designed a differential lock controller for 4 wheel drive vehicle, it worked great and the customer is very happy with it (after extensive testing by mechanical engineers thanks god :) ) it is actually going into production.
Just before I make my first production prototype I have few questions which I like your opinion. 1. I sense the position of 5 contacts with respect to ground either open (hi through pullup resistor) or short to ground, the sensors are approximately 3 meters away, they are working fine just connected directly to port pin of the microcontroller OUCH, do I need to use opto-couplers for these switches and contacts? Not necessarily optocouplers, I might use a HCT14 and dual diode protect the inputs. 2. I sense the speed pulses from proximity sensor, I buffer the pulses and send them to interrupt pin directly, do I need to use optocoupler also after the buffer? same as above 3. Do I need to use line filters to my signals on the microcontroller board? I would use a ferrite and a cap. 4. I have a single chip microcontroller, do I need a solid ground plane, although my circuit lives in a grounded metal case with nice military connector? I would strongly recommend it, again how much failure can you handle. No ground plane would probably make, say, one unit in 500 behave strangely because of the vehicle it was installed in. 5. Can I drive LEDs on the dashboard direct of the controller board using a driver transistor? with the correct transistor and series resistor, yes 6. Looks like I have optocouplers on the brains, I thought we only use optocouplers to get red of ground noise in signals from different systems, in this case where I have switches and on off contacts do I need optocouplers at all or is there a cheaper alternative? see above 7. My circuit seems to work nicely without any of the above, I can't raise my cost up because I might risk losing the contract, any suggestions please? It all depends 1) how many errors (missed diff lock or erroneous diff lock) can you handle? 2) with the contact open you effectively have an EMF antenna connected to the port pin I am fairly sure it will blow the uC in some cases. 3) Optpcouplers will be overkill, you can probably handle ground noise if you route both wires fron your controller to the switches and do not rely on chassis ground. BUT as seen above you will need EMF protection. 4) in my expreience uC pins are much easier to blow than HCT pins so throw a $0.18 HCT buffer between the switches and the uC. again, how much failure can you handle is the real question, sticking with your proto it will probably blow in some cases, how many is anybodys guess. A couple of diodes, a ferrite and a cap on each input should not be that expensive. OH, coming to think about it, there may be a cheaper way. Just a series resistor (HCT inputs do not require much) and a SMALL cap. The HCT chips are specified to take 20mA both below ground and above Vcc. Erik |



