EMC compliance help required.

Thread Starter

Nick Bacon

Joined Nov 7, 2016
130
With the help of a couple of guys on the forum I have managed to get very close to a final production unit of my electronic water level switch. There are 2 versions, a modular upto 4 channel unit and a single channel unit. The modular unit has a motherboard that takes the power input, then 4 slots for the daughterboards, then a display board. The single channel unit has onboard power supply with fuse the varistor. Power supply is
Recom 2W, 1 Output, Embedded Switch Mode Power Supply (SMPS), 12V dc, Maximum of 167mA

The units are all solid state, resistors, capacitors, diodes, transistors, leds and a relay. The power supply used on the 4 channel unit is the RS-35-12 SMPS

I only have one large hurdle to overcome, that is CE marking and more specifically EMC testing.

The pictures are of the last prototype and things have been added to the design since. I have ground tracking linking all the mounting holes as they will be mounted to metal chassis plates. chassis plate with have an earth point.

My question is, is there anything else I need to consider with the design? The use is Industrial (power stations etc)
 

Thread Starter

Nick Bacon

Joined Nov 7, 2016
130
Thanks for the post. As yet I have had no help, so I have the basic circuits with no real emc protection.

Am I wrong in thinking that SMPS should already have these in. I know the smaller one does not and seen some circuits for filters.
 

tribbles

Joined Jun 19, 2015
31
I'm in the process of getting my first product through CE marking, and it's been quite illuminating doing the testing. My product is an Intel i7-based CPU module that provides power to other units, so it's quite a bit different to yours, but the principal is the same.

The main testing I've been involved in is conductive and radiated emissions and susceptibility (all under the EN55022 umbrella) - I think that as yours involves water, there'll be some other testing, but as I know a bit about EN55022, that's what I'll concentrate on.

Hopefully this isn't too much of a "egg sucking" email - I know that when I started, I knew very little about EMC and CE testing, and while there are many books on the subject, some of them can be a bit daunting, and without going through it for real, it's difficult to know where to start.

In the testing, there's four aspects:

- Conductive emissions
- Conductive susceptibility
- Radiated emissions
- Radiated susceptibility

"Conductive" basically means "over the wire" - i.e. the power feed. "Radiated" means "over the air" - i.e. bits of your circuit acting like an antenna, and transmitting or receiving signals.

"Emissions" means "stuff you spit out", and "Susceptibility" means "stuff thrown at you".

So the idea is that you don't interfere with the power supply (conductive emissions), if the power supply starts giving you weird signals, then you can still cope with it (conductive susceptibility). Also, you don't interfere with radio signals (radiated emissions), and any radio signals don't cause problems with your equipment (radiated susceptibility).

Conducted problems are mainly as a result of schematic design (but board layout can have a big impact as well), and radiated problems are mainly as a result of board layout.

For CE, conducted is fairly low frequency - 150KHz to 30MHz, and radiated is much higher (25MHz to 1GHz).

Conducted emissions are best solved by input filtering (common mode chokes, inductors, capacitors etc.) - but it's always difficult to know exactly what you're doing (at least that's what I found).

Radiated emissions can be solved by looking at the circuit design, and making sure you don't have any tracks (or loops) whose lengths are a multiple of the SMPS frequency wavelength - if they are, then you can do a number of things such as stitching capacitors along the length.

Susceptibility isn't something I've had problems with, so I'm not sure what to do about them if you find problems!

For conductive testing, both the +ve and -ve are tested separately, and for radiated, the unit is span around, and the height of the antennas are changed to find the 'worst' emission angles. Also, horizontal and vertical polarity is tested.

Emissions can be tested with some pre-compliance kit (which is about £4000) - although it's really more a qualitative result, rather than quantitative. I would recommend doing some pre-screening for the CE testing at a proper lab. The one I use is about £500 for half a day - but you can get a fair amount done in that time (normally just one of the four aspects). If you can get one which will also consult with you as to what you need to do to get your kit to pass, then that's even better.

My first two attempts failed - but not by much - and they talked me through what needed to be done. The third attempt it sailed through, and we went for full CE testing later on (although the final testing will be in a few days time as we had to do a slight board modification). We did have some radiated emission problems, but this is due to a 5V supply line that crosses some power outlet lines (neither of which have stitching capacitors) - but this has been solved by adding some ferrite beads on the power outlets.

Because of the market we're in, we needed to do additional testing which involved 4KV spikes to the power supply (AC - we can work in both DC and AC mode [and all of them need to be tested individually]), and also 100+V spikes to the DC input (which is 10-32V), as well as -ve spikes.

To overcome this, you need to have TVS diodes or varistors on the input to dampen any spikes. For my board, I've used SMBJ36CA (36V bidirectional TVS diode). You should also seriously consider fuses if you've not already fitted them.

We also needed to do static discharge testing, but I'm not sure if you would need to do this.

I hope this helps (anyone, really!)
 

Thread Starter

Nick Bacon

Joined Nov 7, 2016
130
Thanks for the very detailed reply. I guess that I am lucky that my units are all solid state, no micro processors, no digital signals.

I will try and find a company that can do some testing for me. I will probably add some capacitors throughout the circuit but not sure where yet. I think I will add line filters also and maybe so ferrite needs but I need to research where they go. I am no electronics designer.

I have a fuse on one unit and a varistor after the fuse, I will add the same to the other unit for good measure.

I do have inputs from a probe that works via conductivity of water, but well away from the unit. I was wondering about some form of galvanic isolation on the input cables but again not sure what and whether this will affect the resistance levels inside the unit.
 

tribbles

Joined Jun 19, 2015
31
As it's all analogue, and probably fairly low speed at that, you may not need much filtering - if at all. A simple filter is the Murata BNX series, which is available in TH as well as SMT. They don't really work well for high current though (it's more impedence matching).

One thing to be aware of is that as your system is modular, you may need to get it tested in each of the modular forms - i.e. with no modules, one module, two modules, ...

Our system is also modular - but we are testing the modules individually, and using what is known as a "technical argument" about combining these together in different variations (although we will be spot-testing specific combinations).

Also, if your cables are longer than 3m, then there may be additional testing that would need to be done (but it really depends on what the actual tests you need are). This is where it's worth speaking to a lab to get their opinion on the matter (I was fairly lucky in that there are others in my company who have done this before, and they told me what standards we needed to follow - and the lab agreed with them [after a bit of a discussion about some of the nuances of some of them]). I would also look at TVS diode for the input cables as well.

Our system doesn't have long cables like that, but we did test a couple of them that could potentially exceed 3m anyway (the system passed).

Whereabouts are you based? I'm in south England, so the labs I'm aware of are:
  • Hursley EMC (Chandler's Ford)
  • TUV SUD (Fareham)
  • dB Technology (Cambridge)
Further north, I know of York EMC (York, funnily enough - although they seem to have another branch in Bristol).

The funny thing about the susceptibilty test is that you need to define what is a "pass", so you could say "yes, it's supposed to catch fire when this happens" (I suspect that this isn't quite good enough though).

For us, it was the ability for the system to recover in a working state (we have a watchdog on the CPU that reboots it). Unfortunately, during one of the two unexpected reboots that occurred, Windows decided it wanted to check the disk for corruptions, but this was still deemed a pass. You woudn't have this problem, I suspect :)

I've just had a look at the SMPS you're using, and it should pass EN55022 Class B. But as you're using mains voltages (and not an enclosed PSU), there will be some additional safety tests that we didn't need to do. Things like track separation (I'm a bit concerned by your mains tracks where they narrow down - but then I don't really do much with mains. I have a vague recollection that a 5mm gap is required - but that could be between low voltage and high voltage, and there's a different separation for L/N/E).
 

Thread Starter

Nick Bacon

Joined Nov 7, 2016
130
Thanks again. I am in the south too, I live in Fair Oak near Eastleigh so only 10- 15 minutes from chandlers ford.

I have no issues with changing the track widths and gaps. It is likely a line filter will be installed in those tracks so can easily open them up a bit.

The input cables are 10 metres to 30 metres. An alternating current is applied to the probe but at about 6 micro amps.

The SMPS is only one purchased for testing and not vital that it is used. The one channel unit uses an onboard SMPS (plastic enclosure) I will likely add a line filter to this too.
 

ronv

Joined Nov 12, 2008
3,770
Thanks for the post. As yet I have had no help, so I have the basic circuits with no real emc protection.

Am I wrong in thinking that SMPS should already have these in. I know the smaller one does not and seen some circuits for filters.
It looks like they both pass emc, but the little guy has UL pending.
 

tribbles

Joined Jun 19, 2015
31
UL is the Underwriters Laboratory (from memory). The only thing I'm a bit familiar with is the fire safety rating they do (UL94-V0, for example). I don't know what else they do...
 

Thread Starter

Nick Bacon

Joined Nov 7, 2016
130
As it's all analogue, and probably fairly low speed at that, you may not need much filtering - if at all. A simple filter is the Murata BNX series, which is available in TH as well as SMT. They don't really work well for high current though (it's more impedence matching).

One thing to be aware of is that as your system is modular, you may need to get it tested in each of the modular forms - i.e. with no modules, one module, two modules, ...

Our system is also modular - but we are testing the modules individually, and using what is known as a "technical argument" about combining these together in different variations (although we will be spot-testing specific combinations).

Also, if your cables are longer than 3m, then there may be additional testing that would need to be done (but it really depends on what the actual tests you need are). This is where it's worth speaking to a lab to get their opinion on the matter (I was fairly lucky in that there are others in my company who have done this before, and they told me what standards we needed to follow - and the lab agreed with them [after a bit of a discussion about some of the nuances of some of them]). I would also look at TVS diode for the input cables as well.

Our system doesn't have long cables like that, but we did test a couple of them that could potentially exceed 3m anyway (the system passed).

Whereabouts are you based? I'm in south England, so the labs I'm aware of are:
  • Hursley EMC (Chandler's Ford)
  • TUV SUD (Fareham)
  • dB Technology (Cambridge)
Further north, I know of York EMC (York, funnily enough - although they seem to have another branch in Bristol).

The funny thing about the susceptibilty test is that you need to define what is a "pass", so you could say "yes, it's supposed to catch fire when this happens" (I suspect that this isn't quite good enough though).

For us, it was the ability for the system to recover in a working state (we have a watchdog on the CPU that reboots it). Unfortunately, during one of the two unexpected reboots that occurred, Windows decided it wanted to check the disk for corruptions, but this was still deemed a pass. You woudn't have this problem, I suspect :)

I've just had a look at the SMPS you're using, and it should pass EN55022 Class B. But as you're using mains voltages (and not an enclosed PSU), there will be some additional safety tests that we didn't need to do. Things like track separation (I'm a bit concerned by your mains tracks where they narrow down - but then I don't really do much with mains. I have a vague recollection that a 5mm gap is required - but that could be between low voltage and high voltage, and there's a different separation for L/N/E).
With regard to the TVS on the input cables. There are 2 cables that give out a 6V alternating current to the probe, the other 2 cables are the return. When the probe is in water the circuit from cores 1 and 2 to 3 and 4 is complete. What should I use at such low voltage. The potential issues are that the unit receives spikes from workmen that could be welding nearby and the unit is not isolated, which is usually the case. The TLC3702CP and the mosfets are the components likely to be damaged. What ever is used can not change the 6V AC. 2 cores send voltage out and 2 receive it in.
 

Thread Starter

Nick Bacon

Joined Nov 7, 2016
130
I think the CE standard also includes safety, similar to UL (Underwriters Lab) in the US.
Yes it does, at the moment I am concentrating on the etc side of CE as the safety side is a bit more straight forward for me. It includes things like labelling, protection of live parts. This will all be taking care of in the final design, including the type of power supply, the enclosure, chassis earthing etc.

The funny thing with CE is that there is no rules on having it done by a 3rd party, as long as there is a the technical file. I will get the EMC done by a lab and then complete a report according to the low voltage directive for this type of equipment.
 

tribbles

Joined Jun 19, 2015
31
With regard to the TVS on the input cables. There are 2 cables that give out a 6V alternating current to the probe, the other 2 cables are the return. When the probe is in water the circuit from cores 1 and 2 to 3 and 4 is complete. What should I use at such low voltage. The potential issues are that the unit receives spikes from workmen that could be welding nearby and the unit is not isolated, which is usually the case. The TLC3702CP and the mosfets are the components likely to be damaged. What ever is used can not change the 6V AC. 2 cores send voltage out and 2 receive it in.
I would use a high-ish bidirectional TVS diode, say 12V (this means it shouldn't impact the readings), and then some clamp diodes to the supply.

Poor ASCII art follows...

-6V --|>|--[signal]--|<|-- 6V

[signal] --|>|<|-- GND

The TVS diode would catch the higher voltages, and the clamp diodes should catch the rest. This would need to be applied to each of the inputs and outputs.

However, I would say that my domain is digital, and I'm not sure about how this would affect the analogue signals.
 

Thread Starter

Nick Bacon

Joined Nov 7, 2016
130
I have been thinking about the designs and ways of reducing possible issues with CE marking and would like to run some ideas by you. Rather than using a single caged SMPS, I was considering using 2 smaller encapsulated SMPS, one for 2 channels, they will be pcb mount so reducing the need for connecting cables. It will also mean that is one went down then the other would continue to run.

I have found a suitable line filter but they specify for good emc design to use. This is the SMPS. I am unsure about the MOV and NTC

Note: Output filtering capacitor C2 is a electrolytic capacitor, it is recommended to use high frequency and low impedance electrolytic capacitor. For capacitance and current of capacitor please refer to manufacture’s datasheet. Output capacitor withstand voltage derating should be 80% or above. C1 is ceramic capacitor, which is used to filter high-frequency noise. External input NTC is recommended to use 12D-5; External input MOV is recommended to use S14K350.

Looking at the MOV it seems to be a discontinued product and not sure on a replacement. Also not sure what thermistor to look at.
http://uk.rs-online.com/web/p/embedded-switch-mode-power-supplies-smps/6675786/
 

Thread Starter

Nick Bacon

Joined Nov 7, 2016
130
I would use a high-ish bidirectional TVS diode, say 12V (this means it shouldn't impact the readings), and then some clamp diodes to the supply.

Poor ASCII art follows...

-6V --|>|--[signal]--|<|-- 6V

[signal] --|>|<|-- GND

The TVS diode would catch the higher voltages, and the clamp diodes should catch the rest. This would need to be applied to each of the inputs and outputs.

However, I would say that my domain is digital, and I'm not sure about how this would affect the analogue signals.
I have looked through the circuit in more detail and it looks like there are some Zeners there already that should take care of this. I think I will leave it as is so it does not mess with the voltage levels.
 

Papabravo

Joined Feb 24, 2006
21,158
If you get the UL certification you know you have to pay them every three months indefinitely -- right?
Are you sure you can fund this out of your own pocket?
 

Thread Starter

Nick Bacon

Joined Nov 7, 2016
130
If you get the UL certification you know you have to pay them every three months indefinitely -- right?
Are you sure you can fund this out of your own pocket?
The product will only be EMC tested in U.K. For CE marking and that's a one off cost.
 

kubeek

Joined Sep 20, 2005
5,794
Watch out for the leakage current of whatever protection you throw at the analog inputs. TVS and MOVs have pretty large leakage, and some diodes do as well. With micropamps of measurement current, you will probably have issues when subjected to the temperature extremes, which for industrial I believe are -40°C to +70°C. Also check the criteria for bursts and surges so that your device can survive those (i.e. does it need to keep telling the correct state even when subject to fast transients..)
A 1kV surge, even when coupled through the 42ohm and 0.5uF can blow simple I/O protection into a mist of copper and silicon pretty easily - shielded cables might give you a better ability to withstand those due to differnent connections of the test setup.
 
Top