We are considering about using SwitchBlox - Small Ethernet Switch (SKU: BB-SWB-E-1) or industrial version (SKU: BB-SWI-C-1) of this switch in our UAV. What is difference from EMC point of view? I can find EMC measurement results only for commercial version.
Do you suggest to shield ethernet cables to get better EMC results?
While SwitchBlox Industrial has not yet undergone EMC testing, SwitchBlox Nano has. SwitchBlox Nano uses the same ethernet transformers as SwitchBlox Industrial, and the circuit is overall quite simple. Below are links to EMC results on both.
SwitchBlox Industrial has more susceptibility to EMC than SwitchBlox. This is because the ethernet transformers are smaller, and more exposed. That all being said, it’s a bit difficult to speak with quantifiably. Can you share what kind of antennas / RF / EMI sources you will expect nearby? Given your application, I imagine it’s BLDCs and radio devices. That’s a pretty common application for SwitchBlox Industrial; I have occasionally seen customers report packet drops due to EMI, which they generally solve by moving the EMI source away from the network switch, some have even solved by adding shielding around the switch.
What we would like to do is to use 5 port switch that is capable to working within temperature operating range of -20 °C to +55 °C and has a smallest EMI “signature” as possible. We are planning to put our UAV under EMC testing and we are trying to use parts that have as small EMI signature as possible. Additionaly, question is - how do you suggest to shield (where to connect shield potential) ethernet cables, since ethernet switch doesn’t have ehternet shield pin.
Understood. In terms of radiation, SwitchBlox Industrial will have a slightly higher signature due to the size of the ethernet transformers, however it will likely be a negligible difference, and will be below the EN55032 Class B limits (which is the more stringent test of EN55032) for CISPR Radiated Emissions 30MHz to 1GHz and CISPR Radiated Emissions 1GHz to 6GHz - CLASS B. For a close approximation, take a look at the results of SwitchBlox Nano (which uses the same chipset and the same transformers as SwitchBlox Industrial and thus is a very good approximation until we complete EMC testing on SwitchBlox Industrial. Note the blue line which is the limit for class B.
30-MHz - 1GHz (Horizontal polarity) Radiated emissions (EN55032) - SwitchBlox Nano
30-MHz - 1GHz (Vertical polarity) Radiated emissions (EN55032) - SwitchBlox Nano
1GHz - 6GHz (Horizontal polarity) Radiated emissions (EN55032) - SwitchBlox Nano
Blue line is average, green line is peak
1GHz - 6GHz (Vertical polarity) Radiated emissions (EN55032) - SwitchBlox Nano
Blue line is average, green line is peak
SwitchBlox Nano is below the class B limit. The worst radiation is around the 200-250MHz mark, with some peaks at 60MHz, 70MHz, 82Mhz and 108MHz. Most of these frequencies are clustered around the fundamental frequencies of 100BASE-T ethernet, which is 125MHz, which makes sense.
If you were really concerned about reducing EMI, you would also consider an EMI shield. This is custom foil or metal shield that surrounds the PCB. Nearly all of our products are bare boards. We sell them like this because most of our customers are putting our products into space/weight limited applications; removing the chassis reduces size and gives more integration flexibility; but it does mean more EMI.
Regarding shielding of cables this is definitely something good to do if you are concerned about EMI. The graph below compares the radiated emissions for a 3 meter S/FTP cable, a 3 meter SF/UTP cable and a U/UTP cable.
Below is an image that shows what each of those cable types is. S/FTP and SF/UTP both contain a shield on the outside, whereas the UTP contains no shield.
You can see the outer shield makes a huge difference, the individual cable pair shield doesn’t really matter.
Now onto your question of where to connect the shield. You are correct that our boards don’t have a simple connection for the shield. Again we did this mostly for size, and the fact that most applications that use our boards have short cables and don’t have huge EMI concerns. The most effective and simplest approach is to connect the shield directly to the ground connection of the board. There is no natural grounding point on SwitchBlox Industrial (the mount holes are floating), so achieving ground of the shield is normally done using a special power cable that breaks out the ground so it can be soldered to the shield drain wires on the shielded data cables. We can provide all these cablesif you let me know the lengths and quantities.
thanks for your very helpful information and examples you sent to me. I will follow your suggestions for sure and I will let you know about the results when EMC prelimanare testing will be done.
Please let me know - do you have any experience about connecting shield of ethernet cable on one side and on both sides? Is usually better to connect shield of ethernet cable only on one side (let say on the side of etherent switch) or on both side (side of ethernet switch and camera for ex.)?
Do you think that non-industrial version of 5 port switch will work fine within -20 °C to 70, 75 °C or it would be better to use industrial version because of such wide temperature range?
My pleasure, keep me in the loop on the EMC side and I will assist.
In general, shielding the cable on one side, or both sides, depends on the application.
Shielding the cable on one side can be good because there is a path for noise to flow on both sides. That’s desired for long cables.
However sometimes you should not ground the shield at both sides. For example if the grounding point on one device will cause issues with the grounding point on another device. In your case, imagine you ground one side of the cable shield to the power ground on SwitchBlox, and then ground the other side of the cable shield to, say, the other device (let’s say it’s a Camera) chassis shield. If the camera connects its chassis shield to its power ground, and the power ground of the camera is the same as SwitchBlox, then it should (there are actually issues with doing this, discussed later) be fine. If the camera’s power ground is a different ground to SwitchBlox’s power ground, then you could have current flowing between both grounds. That’s bad.
However even if both the camera and SwitchBlox are using the same power ground you can have problems. Imagine the camera is closer to some EMI source (like a radio), you would imagine the ground on the camera will pick up that noise, which can flow back into SwitchBlox and cause issues.
Another issue is ground loops. Generally in an automotive system, you want to implement a “star ground”, where there is a single grounding point which prevents ground loops; ie, it prevents currents flowing between different device grounds in a system. If you connect the shield at both ends, and both ends are power grounds, you suddenly create many new paths for ground current to flow, and that causes a lot of EMI issues. See the image below,
In this system you have a single star ground, so all ground currents will flow back in one direction. This is generally advisable for automotive systems (really good reference here).
Now look what happens when you connect the shield on both sides. Suddenly now ground current can flow in the shield between the cameras and SwitchBlox. This is very bad.
For this reason, even if SwitchBlox and the Camera are on the same ground, I wouldn’t recommend connecting the cable shield on both sides. In your application (aircraft), I would only connect the shield on both sides, if both sides connect to a chassis that is not the same as the power ground. However since there is no chassis on SwitchBlox, I would just connect one side of the shield to the power ground on SwitchBlox.
Ground can be quite complex, and so you are right to consider this carefully. As you can see, how you connect the shield depends on your overall grounding scheme, and it’s important to get that right too.
Whoops, I didn’t answer your question on temperature range.
SwitchBlox’s operating temperature range is -10 to +80°C. SwitchBlox Industrial’s operating temperature range is -40°C to 85°C. You need -20 to 75C.
You’ll probably be fine using SwitchBlox to be honest. Most of the challenging side of temperature is on the high end, and you’re in range on that side with SwitchBlox.
It does also come down to risk management. Ultimately, SwitchBlox is not designed to operate below -10C, but many customers do operate it there. They judge the slightly increased risk of failure (usually it’s a reduced MTBF but I don’t have numbers on that), as being worth the reduced cost. If you’re deploying something that has to sit for years and will be expensive to access, don’t take that risk. If you’re deploying something that is life-or-death, don’t take the risk. If you’re deploying something that is attritable, or where the consequences of temperature failure are low, the risk may be just fine. Only you can decide.