EMI filter for 24VAC

[boylesg]

Can some one give me an idea if an EMI filter normally used for 240VAC would work for 24VAC?

And if not how would I go about designing one for 24VAC?

 

[betwixt]

I think this may be a classic case of trying to over-engineer a relatively simple problem. In the real world, almost any filter will do something and trying to be very specific about what it has to eliminate when the cause isn't quantified is pointless. Unless the cause of the problem can be identified, the best strategy is a general purpose filter so lets look at what it has to do:

1. let the 24V AC pass through with as little difficulty as possible,
2. block as much of every other signal, assuming it is the interference, as possible.

A narrow band filter, centered on 50Hz would be ideal but also very bulky and very expensive. I'm thinking here of lots of large iron-cored inductors and capacitors, not in the slightest a practical solution.

So lets look at what is different between the wanted power and the unwanted interference. There are two parameters, voltage and frequency that will be different. It is unlikely to be an under-voltage or increase in the 24V that causes your fault because you would see it in other equipment. Far more likely is a rapid spike created by some external inductive load being switched on or off.

Spikes can be eliminated by two methods - by clamping or by filtering and often both are used. The problem with a clamp is what to do with the excess energy it has to absorb. Imagine this scenario (don't try this at home!), you want a 5V supply from a car battery so you connect a 5V Zener diode across it. The result is a puff of smoke from where the Zener used to be because it tried to sink many many Amps from the battery to limit its voltage. However, if you had added a suitable resistor in series with the Zener it would drop the excess in the resistor and give you the 5V you wanted. The same applies to clamping across 24V, usually a MOV would be used instead of a Zener but the principle would be the same, it still needs something in series to take up the difference between the actual voltage (with the spike) and the 24V you want. You have to be careful if you use a resistor because it will also limit the current to your load. Also note that 24V is probably an RMS voltage so the natural peaks will be about 1.4 times higher and you don't want to clip those off as well.

A better solution is an inductor and capacitor filter, it will have increasing attenuation as the frequency rises so the 50Hz, being low frequency is relatively unhindered but anything with faster rising and falling edges must, by definition, contain higher frequencies. Remember the reactance (= resistance to AC) of a capacitor falls as frequency rises and the reactance of an inductor increases with frequency so by adding C in parallel and L in series make passage of high frequencies harder in two ways in a conventional line filter.

The bottom line is that ANY line filter is better than none and you have to make some compromises between complexity, size, cost and performance. The filter you pictured will still work at 24V and you can still use it. The only thing to note is the MOVs are probably rated for 275VAC clamping so at 24V it's very unlikely they will ever operate, you would have to change them to 27.5V ones to get the same protection.

I would question where your 24V comes from, if it is dropped by a transformer, it may be more advantageous to use the original filter on the line side of the transformer.

Brian.

 

[boylesg]

Something like this I guess you mean?
Most of it is on the shields so not under my control re wiring.

 

[KlausST]

Hi,

Your picture shows the datapath. This is not muchbof interest.

To find the source of the problem we need to see the power paths and the GND paths.
Here a complete sketch or a photo is good information.

We need to see whether there are twisted or shielded wires..
We also need to see whether there are ground-loops.

Without values and specifications, it's impossible to calculate.
Then you just can do "trial and error".
But with trial and error you can never be sure if your solution is reliable and how good the safety margin is.
There is a goid chance that it works on one place but not at the other place ... but this is what you have already.

Klaus

 

[boylesg]

Is this what you are asking for Brian?

Is this enough?

 

[betwixt]

It was Klaus that asked for the power arrangements but it does help to some degree.

Question: You show one '24VAC plug' feeding both the relays and the bridge rectifier, I presume this means the relays and solenoids are part of the same assembly and a single cable carries the 24VAC from the plug to all the other blocks in your diagram. Is that correct? If it is, can you show more detail of which parts of the wiring are common to the relays/solenoids and bridge rectifier.

Purely as a guess at this stage and assuming the bridge feed wiring doesn't share solenoid current, you should add a filter only in the link between the plug and rectifier. However, the load after the bridge rectifier looks to be quite well stabilized so your problem may be more to do with ground loops than the supply voltage itself.

Brian.

 

[boylesg]

Hi,

Your picture shows the datapath. This is not muchbof interest.

To find the source of the problem we need to see the power paths and the GND paths.
Here a complete sketch or a photo is good information.

We need to see whether there are twisted or shielded wires..
We also need to see whether there are ground-loops.

Without values and specifications, it's impossible to calculate.
Then you just can do "trial and error".
But with trial and error you can never be sure if your solution is reliable and how good the safety margin is.
There is a goid chance that it works on one place but not at the other place ... but this is what you have already.

Klaus

Why don't you show me an example of what you want? Because I just don't understand what it is exactly you are looking for.

 

[FvM]

Some observations:
The hardware descriptions in post #22 and #24 are completely different. Do you have all the peripherals listed in both block diagrams? If so how is the actual shield configuration? Are you even sure that the shields cooperate without disturbing each other?

The only (rather scarce) failure report we got is in post #19, SD card failure. Is this the whole story? To get a failure caused by hardware interferences, there must be either voltage drops or massive common mode currents running through the shield pack. It could be also well software related, or e.g. triggered by the WiFi antenna next to the data logger shield.

 

[KlausST]

Hi,

O.K.
Let´s start a new.

place your circuit on the bench. Power it up. Try to force the error you are talking about.
When the error happens:
Don´t move anything. Don't connect or disconnect any wire/cable. Don´t disconnect the scope. ..
--> just upload one photo of all your circuit with all connections. An overview.
including power supply, wires, every pcb/display/sensor...

That´s it.

We will check your photo. Maybe after that we ask for additional more detailled photos.

*****
It´s no problem if you don´t understand what we are talking about. Just tell us what you don´t understand.

But I aked a lot of - in my eyes - rather clear questions, like: Solenoid count, solenoid current, how many solenoids are ON at the same time. ...

I asked about shielded / twisted cables....

It´s not important that you understand why I ask for this. But you should be able to give those informations.

*****
Your Sketch.. is a good start.
I asked about GND wiring. Complete GND wiring.

I don´t know your circuit. So I have to guess.
And I guees the sketch is more like "you want it to be" but not like it is in reality.

Just see the box "Relays". It just has an AC input, no GND so far. But I guess there needs to be some GND connection as well as some DC supply to operate the relay coil. Just gessing.... You know whether my guessing is right or not.

A photo shows the real condition. (as long as you don´t move, connect or disconnect something)

Klaus

 

[boylesg]

Does this sort of diagram show you what you want to see?

 

[KlausST]

No.

you said:
"I just don't understand what it is exactly you are looking for."

I said:
"--> just upload one photo of all your circuit "

*****
a little explanation:
Such problems are not that simple. It´s not simple to detect the root cause. It´s not simple to find a solution.
There are many causes. So many I can´t tell you all of them.
Thus there is no "exact" what I am looking for. It´s a whole bunch of things I am looking for. And it´s difficult to explain every single possible problem you have to look for.

It´s impossble for you within one forum thread to understand this all. I´m still learning after decades of intensive electronics development.

****
Thus I tried to get a photo. It is objective. It shows the real world, not what you like it to be. It gives us an overview. We can see the wiring, the sizes, the cable length, possible GND loops, possible sources of noise, possible PCB problems, and countless other possible things.... even things I haven´t thought about yet.

I failed to get this photo.

So, what´s your idea? How to go on? Tell us what you expect from us.

And mind: I´m not interested on you to feel bad. I´m interested in solving an electrical problem.

Klaus

 

[boylesg]

 

[boylesg]

 

[Taihung]

You need clear your purpose when using EMI filter.
The filter can reduce noise from your board to 24Vac supply bus and vice versa, from 24V bus to your board.
EMI have two sides: conduct & radiate.
The most in-line filter is more effective for conduct EMI (150kHz - 30Mhz).
Radiate EMI is noise with high frequency 30Mhz - 3Ghz, so normal inductor, capacitor is very hard to filter. We have to come back the design: select right solution, damper for high dv/dt point, damper for ringing, layout of PCB.
You want to filter EMI, but how much and why need ?
Measurement with spectrum analyzer to know what exactly how your board generate EMI and compare with which EMI standard you want to meet.
After you have measurement, we wil know which point(frequency) is high and we design/select filter component for damping it. EMI filter is not universal for all application. It is specific design for each circuit.
An universal EMI filter still have, but it is multiple stage to cover overall range from 150Khz - 30Mhz. It is only for measurement equipment and not suitable for mass product.
Your board use LM2596/LM2576 module, 24Vac is low voltage, as my estimation the EMI is not so high. You can add a small low pass LC filter is enough.
You need protection for over voltage than EMI filter. TVS or MOV is good choice.
If VAC power supply swing large value, you should add more input capacitor to maintenaince input voltage.
When on/off relay for AC power supply and inductive load will generate spike, this is the most trouble make noise and sometime restart your circuit. Some bad case, MCu can lost flash. Add RC snubber for relay contact, using zero voltage detect to close contact and zero current detect to open contact. Remember contact change state take time when apply control voltage. It follows the kind of relay, vary from few ms. Close and open contact delay time is different.