geoffr
Posts: 85
Joined: Wed Aug 22, 2012 11:25 am
Location: Melbourne, VIC

EMI suppression for inductive AC loads

Fri Oct 14, 2016 4:12 am

One of the interesting challenges I am running into with building an irrigation controller is the EMI generated as a result of switching the solenoid valves.
There are really two sets of inductive loads in the project: the relays which switch the AC circuits (here I have flyback diodes integrated into the ULN2803a Darlington transistor arrays which switch the relays) and the solenoids themselves (which are running on 24VAC).
It is the EMI from the solenoids (mainly when they switch off) that I still need to deal with. The solution I have found, looking around online, is to put a MOV in parallel with the load, and possibly also a small resistor (47 Ohm in one example, 100 Ohm in another) in series with a small capacitor (100 nF to 220nF) in parallel with the load.

I was wondering if I am missing anything or if anyone has a better suggestion.

Regards,
Geoff.

IanS
Posts: 232
Joined: Wed Jun 20, 2012 2:51 pm
Location: Southampton, England

Re: EMI suppression for inductive AC loads

Mon Oct 17, 2016 11:18 am

One further possibility is to only switch the relay when the AC voltage passes through 0. You might be able to track the phase via an voltage divider and ADC and time your controls to the zero crossing, or you can get relays that will do it for you automatically. Search for 'zero crossing relays'.
This will also prolong the life of your relays.

ewaller
Posts: 32
Joined: Tue Oct 04, 2016 5:24 pm

Re: EMI suppression for inductive AC loads

Mon Oct 17, 2016 4:15 pm

IanS wrote:One further possibility is to only switch the relay when the AC voltage passes through 0. You might be able to track the phase via an voltage divider and ADC and time your controls to the zero crossing, or you can get relays that will do it for you automatically. Search for 'zero crossing relays'.
This will also prolong the life of your relays.
The only problem with this is that "zero crossing relays" usually switch when the voltage crosses zero, not when the current crosses zero. With a purely inductive load (a pretty good model for a relay coil) the current is 90° out of phase with the voltage; the current is at a peak at the same time the voltage reaches zero. This means you will still get an inductive spike, and depending on the phase, that pulse can be positive or negative. So, see if you can find a "relay" that switches when the current is at zero. Another solution would to use a positive DC voltage to drive the coil instead of an AC voltage. You could drive it with a single FET, and should use a Schottky diode across the load to catch the negative pulse when the FET turns off.

ddahms
Posts: 67
Joined: Tue Mar 18, 2014 3:38 pm

Re: EMI suppression for inductive AC loads

Mon Oct 17, 2016 7:32 pm

Although my commercial sprinkler controller uses triacs, it has a MOV across each one. Its part number is S10K30 and it has a 47V breakdown voltage. They are available from DigiKey.

You could try a 0.1uF capacitor across the load, without any series resistor.

You do have the DC power to the Pi separate from the 24VAC power to the relays and valves, right?

I am curious about what problems are being caused by the EMI.

geoffr
Posts: 85
Joined: Wed Aug 22, 2012 11:25 am
Location: Melbourne, VIC

Re: EMI suppression for inductive AC loads

Mon Oct 17, 2016 7:38 pm

ddahms wrote:Although my commercial sprinkler controller uses triacs, it has a MOV across each one. Its part number is S10K30 and it has a 47V breakdown voltage. They are available from DigiKey.

You could try a 0.1uF capacitor across the load, without any series resistor.

You do have the DC power to the Pi separate from the 24VAC power to the relays and valves, right?

I am curious about what problems are being caused by the EMI.
When I was initially testing the solenoid valves, I carelessly ran a console cable near the wiring to the solenoids. Often when the solenoids switched off, the MCP23017 which is controlling them would disappear off the I2C bus. I needed to reset the MCP23017 to get it working again.
One simple solution is to just keep the cabling physically separated. The problem I have however is that I am going to be running cabling to a flow sensor in the same duct as the cabling for the solenoids. I expect that the problem may resurface when I do that.

Moe
Posts: 230
Joined: Sun Jan 25, 2015 2:44 pm

Re: EMI suppression for inductive AC loads

Tue Oct 18, 2016 10:42 pm

If I were trying to solve this problem at work I'd probably focus on isolation between the electronics and the loads, rather than trying to silence the solenoids. Solenoids are always going to be noisy; MOVs will help but you'll never get rid of the transients completely.

As the aim here is to prevent those transients coupling onto signal wires, the best approach is NOT to run the cables together. However, as you've already said this is not an option :-) you need to make the susceptor cables less susceptible. Shielding might help a little bit; probably not much.

If the susceptor is carrying the I2C then you really need to convert it to something else, e.g. a differential signal over twisted pair (e.g. ethernet, DSL, RS422) or ideally fibre with line drivers at each end. I2C is single-ended, intended for short range comms and incredibly sensitive to noise. That noise could easily be conducted down the DC power supply or other cables to the Pi and 'break' the bus even if the actual I2C wires are nowhere near the source.

Another possibility - your system already knows exactly when the interference is going to happen, so you could selectively ignore certain sensor readings.
Submarine communication systems engineer and amateur robot enthusiast.

geoffr
Posts: 85
Joined: Wed Aug 22, 2012 11:25 am
Location: Melbourne, VIC

Re: EMI suppression for inductive AC loads

Thu Oct 20, 2016 9:31 am

Moe wrote:If I were trying to solve this problem at work I'd probably focus on isolation between the electronics and the loads, rather than trying to silence the solenoids. Solenoids are always going to be noisy; MOVs will help but you'll never get rid of the transients completely.

....
My initial tests using a MOV have been successful. The sensor wiring that I am going to run parallel to the solenoid wiring is for a Hall Effect sensor (flow sensor). In the real installation, the length of cabling that will be running parallel is greater, so my experience in the real world may be different.
I will basically need to put a small enclosure with the MOVs in near the solenoids. I was planning to put an IP67-rated enclosure in place to do all my connections, so it can also house the MOVs. Fortunately, my plumbing is such that all my solenoid valves are in one location. That makes things simpler.

As you say, it would be preferable to use different cabling routes. Logistically that is however rather difficult for me.

geoffr
Posts: 85
Joined: Wed Aug 22, 2012 11:25 am
Location: Melbourne, VIC

Re: EMI suppression for inductive AC loads

Thu Oct 20, 2016 9:37 am

ddahms wrote:Although my commercial sprinkler controller uses triacs, it has a MOV across each one. Its part number is S10K30 and it has a 47V breakdown voltage. They are available from DigiKey.

You could try a 0.1uF capacitor across the load, without any series resistor.

You do have the DC power to the Pi separate from the 24VAC power to the relays and valves, right?

I am curious about what problems are being caused by the EMI.
The problem I am seeing is that the noise which results when the irrigation solenoids switch off interferes with the MCP23017 that I am using. Sometimes it disappears from the I2C bus altogether.
I am using a bridge rectifier and a buck converter to get my 5VDC supply from my 24VAC supply. I have a fairly large smoothing capacitor between the rectifier and the buck as well.
The solution using a MOV and capacitor seems to work in initial testing. The real world may be different where my two sets of cabling run parallel for a longer distance.

Return to “Automation, sensing and robotics”