Relay Switching - generating Noise

[sdelima]

Hey all,

I need some help in designing a MCU system. I have a set of 4 relays and one of them switches an inverter. I have a MCU monitoring the system and so everytime the relay switches the MCU resets itself (noise). I am working on different PCB layouts so as to keep the relay as far away from the MCU as possible. I have tried caps ... but have not been successful. I would greatly appreciate any suggestions.

Thank you
SD

 

[cbarn24050]

Hi, the usuall cause is poor power supply arrangements, inadequate decoupling at your micro and load currents flowing in the micro's 0v circuit.If your relays are driving other inductive loads, such as a contactor, then you need to put snubber network across the contacts and a VDR accros the contactor coil.

 

[BrianG]

Silly Question: You mention trying capacitors, but have you got the usual diode across the relay coil?

 

[sdelima]

I have the diode across the relay.

The load (a lamp load) is primary ont he normally-closed contacts (no load on the normally open)
When the relay switches from normally open to normally closed, it does not reset the micro but when it switches from normally closed back to normally open it causes the mcu to reset.

i appreciate your help

SD

 

[buzzp]

This is definately the problem based on when it resets. Have you monitored the 5V and the master clear pin with a scope? You will likely see it drop when you switch or you may see ground bounce when this switching happens. Make sure you have a bypass cap on master clear and the power supply pin for the microcontroller right NEXT to the pins. Make sure the leads (thru hole) are as short as possible. For starters, use a 10u 6.3 volt tantalum kluged onto the above pins. If this fixes your problem and you already have caps on these pins then the problem is with the location of the existing caps and/or board layout. Also, tie master clear hard to the 5 volt bus (no resistor in between). Is the micro operating at lower than 5V? If so, I would consider operating at 5 volts because of the noise being generated by the inductive spike. This will make the micro less immune to these types of events. You may also consider a bidirectional TVSS on the 5volt bus to ground. This should clamp any voltages outside of the TVSS voltage. This could be a lot of different things: board layout, capacitor size and location, etc.. You may also think about ferrite cores on the control wires to the lamp. Let us know what you have found and I will be happy to give more advise.

 

[BrianR]

There are several things you have not defined but check these points:
1. Diode across relay coil. (You've done that).
2. The surge current when relay closes, energising the load momentarly drives the power supply into overload and it briefly drops its output voltage. If the MCU is connected to the same supply it will reset. Solution is to use a seperate power source and seperate wiring for the load.
3. Ensure that there are no common current paths for the MCU power and ground, and the load power and ground. The ground paths are just as important as the power paths because when the load energises there may be significant momentary voltage spikes on the power AND ground leads.

 

[sdelima]

I apprciate your help....

I think I have figured out the problem. Let me begin my explaing what i have... I have 3 circuits.... A charging circuit (charges the battery), and inverter circuit (in a black box.. I cannot redesign that) and a daughter board (this is where the microcontroller is.. and once again i cannot redesign that... )

The only thing i have access to (to redesign ) is the charging circuit. A 14.4 V battery is used to power the lamps (in an emergency) and after droping down the voltage (with a zener)it is fed into the MCU circuit.

Now this is what I did.. I disconnected the 14.4 V battery to the to the MCU board and I used a seperate battery to power the board.. and now I don't have any resets ...

Here is the problem I need to use the same battery that powers the inverter to power the MCU board. There is a little connector and after that is closed will the MCU board and the inverter be connected. Hence I can't get the power to the MCU board as close to the battery as possible....

I have a 0.01uF cap at the pins between the charger board and the MCU board.. Do y'all have any ideas on cleaning that power supply

Thank you once again.. I appreciate it

SD

 

[sdelima]

Also i forgot to mention the person who designed the existing MCU board made the operating voltage 4.0V .. (don't ask me why )

 

[melone]

Try putting MUCH MORE CAPACITANCE on your board (in the range of 100-1000 uF). Also, place 0.1uF CERAMIC capacitors in parallel with your 0.01 uF (hopefully ceramic) caps. The large C will help smooth out noise on the incoming power, and the small C will act like a current pool for your tranistioning micro. You might want to put some bleed off caps to help dissapate your energy from the caps once the system powers down.

Good luck and keep us posted!

 

[yesway]

Hi guys, just wanted to check how the MCU drives the relay(s)? If it drives them directly (doesnt sound likely in this case?) or via a bipolar transistor it may use enough current from the probably weak zenervoltage for VCC to drop? In such a case you may consider:
0. If possible increase series R to the relay driver transistor(MUST be checked that transistor saturates correctly or fire may occur!)
1. More powerful regulator for MCU, like a 7805 or 78L05
2. Put some buffer stage between the MCU and the relay driving circuitry
3. Exchange the bipolar relay driver with a FET (with some 10k gate pull down in case of MCU fail)

best of luck

 

[cbarn24050]

Hi, the lamps are a virual short when they are cold and cause a voltage dip when you connect them.Put a diode in series with the supply to the micro voltage regulator to stop the res cap discharging during the voltage dip.

 

[BrianR]

cbarn24050's comment is right on the ball. It is standard practice in military systems to have a series diode in the power line, followed by a large electro and a 0.1uF multilayer ceramic cap to ground. This holds up the power (to MCU in this case) during momentary dropouts.

These dropouts are also a major cause of failure of 3 terminal voltage regulators which find themselved momentarily reverse biased when power input droops but the output is held high by a cap in the load. They do not like being reverse biased.

 

[buzz41]

Some good suggestion, so far. I flagged this, though:

A 14.4 V battery is used to power the lamps (in an emergency) and after droping down the voltage (with a zener)it is fed into the MCU circuit.

If that is a series zener feeding the micro, I would consider that sub-standard design in a power ckt. Dump the zener and use a 78xxx as suggested, even if other measures cure the immediate problem. A series diode may be good to add as well, though it will use up a bit of voltage head room.

 

[sdelima]

Guys.. here is what i did and I think it fixed the problem.

I stuck with the zener and used a decoupling circuit to decouple my power supply into the MCU circuitry. I moved the relay away from the MCU circuit and that helped too.

When I get my new PCB i'll keep y'all informed. Thanks a bunch

SD