From: "Jeff Verive"
Subject: Re: Automotive circuit falsing - urgent help needed please
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
Date: Tue, 31 Dec 2002 11:14:08 -0600
NNTP-Posting-Date: Tue, 31 Dec 2002 11:14:08 CST
At least four types of "filters" must be employed in an automobile:
- Circuit-level and device-level power supply filtering
- I/O conditioning (at the sensor, cable, and micro)
- Circuit-level EMI protection (from radiated noise in the auto environment)
- Circuit-level EMI reduction (making sure your circuit does not radiate or
conduct outward excessive EMI)
For the hobbyist, the most important things to cover are the power supply
filtering and I/O conditioning. You want to ensure that under all
circumstances (including ignition, jump-starting, alternator load-dumps,
etc.) the circuit is protected from damaging high voltage and reverse
voltage spikes, in addition to regulating the power supply voltage so that
it is rock-solid under all of these conditions. All the signal conditioning
in the world will not help if the micro is constantly getting reset or
damaged by EMI. Proper fusing, over-voltage clamps, and generous use of L-C
filters is best. In fact, it is a good idea to treat the power supply
protection as a separate circuit, not just a few capacitors and diodes.
National Semiconductor once sold a handbook on voltage regulators, and the
section on automotive power filtering was outstanding. I bought this
handbook, and keep it in a safe spot to keep other engineers from borrowing
I/O conditioning usually consists of low-pass or high-pass filtering
(depending on the type of sensor), usually communicating with the micro via
a shielded, twisted pair cable. The shield is *usually* grounded at the
micro, to ensure that the shield does not act as an antenna. The twisted
pair goes a long way toward rejecting differential mode noise, and a common
mode choke near the micro helps remove common mode noise. I have almost
always used differential mode and common mode filtering, along with
opto-isolators to protect the micro.
To protect from radiated EMI, consider using a faraday shield around the
micro, with feed-through capacitors as part of I/O connections. A good PI
or T filter network on the power supply and I/O will also keep you from
conducting or radiating excessive EMI to other electronic circuits within
Good luck, and remember to test, test, and test again.
"MikeB." wrote in message
> (Posted this in the other group but no articles show in the last 12
> I probably should have put it here. Any comments welcome!)
> Hello gents,
> I recently made an automotive circuit that needs to detect such things as
> when the brakes and turn signals are applied, when the wipers are turned
> etc. This data is fed into an Atmel for processing. I'm having a problem
> with the unit falsely detecting that the brakes or blinkers have been
> activated. The things I have tried to fix it have not worked.
> From looking at the factory wiring diagram, the brake circuit is simply a
> momentary switch inline with the brake circuit. There are no relays or
> electronics in the circuit. From the diagram, there should be no power in
> the line, and when the brake is applied (switch closes) power flows
> the switch and the brake light, illuminating the light.
> On the pin of the Atmel, I have an NPN transistor and I have the base
> connected to the brake line through a 12K resistor such that when the
> transistor is switched, the pin on the Atmel is grounded. When I try it
> on my bench it works fine - when power is applied to the "sensor" wire,
> Atmel correctly notes the signal. I've used this same setup on many Atmel
> projects and haven't before run into a problem. When installed on the
> vehicle however, it thinks the brakes are being applied even when they
> aren't. I've tried this on several vehicles and even multiple units of
> same vehicle and it's erratic. Sometimes it detects it correctly, some
> times it doesn't.
> I tried using a reverse-biased 5-volt Zener, so that it would take >5V to
> switch the transistor. It did not seem to help, and the Atmel is still
> acting as if the brake is being applied even when it is not. Can there
> really be >5V of "noise" in the lines?
> Does anyone have any ideas? What sort of "filter" should I put on this
> to prevent the circuit from falsely tripping? I know automotive
> systems are very "noisy" and I'm assuming it's this noise that is falsely
> triggering the circuit. I figured the zener with the resistor would do
> job, but it hasn't. What should I use to prevent the falsing?
> Thanks in advance for your insight...