Hello.

I work in the incubator product(like a smart farm ) development team.
Product development was recently completed and we commissioned a national agency capable of EMC testing.
We have been judged nonconforming several times, and we are currently debugging.
However, there was something I couldn't understand, so I asked this question.

First, the product is composed as follows.

1. 12V 6A Peltier Module (For Control Temperature)
2. Lan Chip (Our product can communicate with the outside with a wired LAN cable. Not WiFi)
3. Main Controller Board (For overall product control)
4. 12V 2A Led Module (For supplying light sources to plants)
5. Touch display (For easy user control)
6. Temperature Senser
7. 12V 10A SMPS Adapter (220VAC 60Hz To 12VDC 10A)

8. The maximum clock for the processor built into the main controller is 80Mhz.
9. The maximum clock for the touch display is 300Mhz.
10. The maximum clock of Lan Chip is 60Mhz.

There is no part that generates more clock than this in our product.

Q1. In the attached EMC test result, electromagnetic waves are leaking even at 400Mhz ~ 1Ghz.
There are no components in our product that generate such high frequencies, how can this result?

Q2. If you have any doubts, how would you recommend taking action?

My goal is to have the dB of all bands down to around 20dB.

 

They are all harmonics of 50MHz, with a large peak at 50MHz.
100Mbit/second has a maximum frequency of 50MHz - Was it tested with the LAN output properly terminated?
or Is the main processor or the LAN actually running at 50MHz not 60MHz or 80Mhz?

 

Sometimes things that aren't supposed to oscillate do so, for example MOSFETs in power supplies if they turn on or off too slowly or TOO FAST in some circuits because they cause parasitic resonances to ring, or what are supposed linear amplifiers oscillating because of too much phase shift in the feedback or improperly bypassed power supply leads. One product I inherited from the previous engineer has a monolithic regulator that oscillated in the RF range. A linear amplifier stage oscillated because the power supply bypass capacitors were too far away.

There are consultants who have been around the block a number of times and can recognize likely EMC sources from the schematic and the PCB layout.

 

This is a graph of radiated emissions. These are all harmonics of internal switching and digital signals. You need to contain the equipment in a grounded metal enclosure if possible. If plastic, add foil or a conductive coating internally. Wiring exiting this enclosure needs filtering. As a minimum, use ferrite sleeves on all cables entering/exiting the equipment.

Internal wiring needs to be kept as short as possible and bundled together so as to reduce any loops. Once again ferrite sleeves are useful. Hopefully, your PCBs use ground planes.

Is the SMPS adapter separate from the equipment? If so, hopefully it should meet international EMC standards and have markings confirming this. If so you could assume that it already meets standards and does not need to be included in the test. Try testing the equipment using a temporary linear power supply.

 

Q1. In the attached EMC test result, electromagnetic waves are leaking even at 400Mhz ~ 1Ghz.

The Fourier series of a square wave contains all odd harmonics. https://www.khanacademy.org/science...ies/v/ee-visualize-fourier-series-square-wave

 

As Ian0 says the EMC spikes are all harmonics of 50MHz. Marley offers a number of possible solutions which you could try.

My advice is to try and reduce the noise at source (the 50MHz) – the spikes are essentially due to the fast rise time of this signal radiating off the PCB tracks/wiring.

If the 50MHz source is loaded with a suitable value capacitor to reduce the fast rise time – this can be a very simple solution. But too large a capacitor will stop the thing working and too small a value will not reduce the harmonics sufficiently.

In my experience the capacitor value need only be of the order of a few nF of a type having a low impedance at high frequency (e.g. ceramic type).

 

This is a graph of radiated emissions. (Graph emitted)

Can you please post that graph?

 

As Ian0 says the EMC spikes are all harmonics of 50MHz. Marley offers a number of possible solutions which you could try.

My advice is to try and reduce the noise at source (the 50MHz) – the spikes are essentially due to the fast rise time of this signal radiating off the PCB tracks/wiring.

If the 50MHz source is loaded with a suitable value capacitor to reduce the fast rise time – this can be a very simple solution. But too large a capacitor will stop the thing working and too small a value will not reduce the harmonics sufficiently.

In my experience the capacitor value need only be of the order of a few nF of a type having a low impedance at high frequency (e.g. ceramic type).

My suspicion is that the product was tested without the LAN cable connected, and thus the 50MHz source was operating but unterminated.
Not sure that adding capacitance would help - might just make more current flow when the output changes state. However, a termination resistor would help to suppress reflections and radiation.

 

Thank you for letting me know where I can find a solution.

I live in Korea and will arrive at work about 3 hours from now.

After going to work, I will try to solve it in detail based on the various directions you have given me.

It may take some time if you order additional parts and materials to solve the problem, but I will share the current situation for those who answered me and for those who have similar problems.

Thank you.

 

EMC compliance is a bit of a black art – minor product changes can have a dramatic affect.

I once spent many hours believing that modifications I was making to a product was affecting the EMC performance – eventually I realised that the position/routing of the mains supply cable was dramatically affecting the EMC emissions.

A good EMC test facility will offer assistance in fixing non-compliant products, but how quickly they can effect a fix will depend on how good their EMC engineers are and how badly your product is failing to meet the EMC limits.

 

1. 12V 6A Peltier Module (For Control Temperature)
2. Lan Chip (Our product can communicate with the outside with a wired LAN cable. Not WiFi)
3. Main Controller Board (For overall product control)
4. 12V 2A Led Module (For supplying light sources to plants)
5. Touch display (For easy user control)
6. Temperature Senser
7. 12V 10A SMPS Adapter (220VAC 60Hz To 12VDC 10A)

If you can disconnect any of the separate modules and still operate the device it would be worthwhile to try that to narrow it down to the offending module. Don't disconnect while it's running. See if it will operate from power up without the modules.

Some possible places to look:
-Displays are notorious emitters because you can't shield the front of a display.
-Other replies have mentioned the LAN frequency, so try disconnecting that and see if the problems still exist.
-Are your LED's PWM controlled? There's another clock source that could be the culprit.
-Also mentioned is the sample frequency. See if disconnecting the sensor helps.

Other suggestions:
-On your cables twist signals wires with their signal returns. You want current to a module to match the return current and for the two to be twisted together to reduce space between them.
-Shield cables if necessary. Do be careful to avoid creating a second return current path with the shield. With that in mind sometimes it's necessary to connect the shield only at one end.
-If you try shielding the enclosure, ensure that the shield is continuous around the whole device. If it has a lid, you may need to connect the lid electrically to the rest of the enclosure, i.e. with a ground strap.

 

Regarding displays, I'm reminded of a situation I ran into many years ago. Our device was a lottery ticket sales terminal. It had a 10" 640x480 pixel monochrome LCD display. Our software developers had set the background on the main screen to be gray. It seemed like a good idea at the time. When we went to test the unit for RF emissions, it failed with a spike at half of the display pixel update frequency. After playing around with it for a while, we discovered that a white background or a black background didn't create the spike. Changing that background allowed us to pass the testing. What was happening was that the gray background had one pixel on then the next off and so on for the whole 307,000 pixels. With the 50% gray that we had we had the worst case emissions for that display!

 

Hello.

I work in the incubator product(like a smart farm ) development team.
Product development was recently completed and we commissioned a national agency capable of EMC testing.
We have been judged nonconforming several times, and we are currently debugging.
However, there was something I couldn't understand, so I asked this question.

First, the product is composed as follows.

1. 12V 6A Peltier Module (For Control Temperature)
2. Lan Chip (Our product can communicate with the outside with a wired LAN cable. Not WiFi)
3. Main Controller Board (For overall product control)
4. 12V 2A Led Module (For supplying light sources to plants)
5. Touch display (For easy user control)
6. Temperature Senser
7. 12V 10A SMPS Adapter (220VAC 60Hz To 12VDC 10A)

8. The maximum clock for the processor built into the main controller is 80Mhz.
9. The maximum clock for the touch display is 300Mhz.
10. The maximum clock of Lan Chip is 60Mhz.

There is no part that generates more clock than this in our product.

Q1. In the attached EMC test result, electromagnetic waves are leaking even at 400Mhz ~ 1Ghz.
There are no components in our product that generate such high frequencies, how can this result?

Q2. If you have any doubts, how would you recommend taking action?

My goal is to have the dB of all bands down to around 20dB.

Did you do conducted emissions first? Was that a pass or a fail? Do you have that data? If these frequencies are coming out the power lines they will radiate so may be a good place to start.