Great, well grounded talk as usual. Lightning strikes can be fun too; something like 1000v / m differential so your shielded serial cable (I'm an old timer) to your shack out back may be a source of sparks in a storm.
@CHSElectronics
3 жыл бұрын
My solution that has worked and continues to work is to use isolated power supplies to power isolated communication drivers like for RS485, RS422, and CANed networks. Each isolated DC Ground can be tied together as a common reference point for both types of signals.
@robertbox5399
3 жыл бұрын
Lots of talk and pen lid clicking but little in terms of solutions. Do you short the planes to the case? Do you ground the screws? Do you need 00 AWG screened cable to survive fault currents? Why is ethernet OK? What about USB? Are phonos between rooms safe? Do I fit GND resistors to prevent loops? What effect does grounding the case have on EMC? Does Altium have an equivalent to Hyperlynx for less than $20,000?
@DiegoColl44
3 жыл бұрын
What bibliography do you recommend reading about grounding? thanks for the tips..!!
@Zachariah-Peterson
3 жыл бұрын
One excellent textbook is Grounds for Grounding: A Circuit to System Handbook
@DiegoColl44
3 жыл бұрын
Thanks @@Zachariah-Peterson ..!!
@mohammadrezaadib8999
6 ай бұрын
I have 10 VFDs with metal enclosure. they all have same PE and their GNDs are connected through PLC connector. Nothing bad has happed, may it get problematic later?
@alperkaplan1839
3 жыл бұрын
thanks for video.I have one question.If cable data will be sent to long distance but there is a potential difference.What is the solution?
@Hneel65
2 жыл бұрын
What if there's a substantial potential difference between the 2 GND's (let's say 10V), then what about the signal connections? They will be way too high or too low either, or not?
@Zachariah-Peterson
2 жыл бұрын
HI Hans, This was actually one of the original reasons for the development of differential pairs; they don't need a ground reference, the signal levels on each trace in the pair are referenced to each other. The other reason was to support high speed/high serial data rate digital signaling with a conductor configuration that only radiates differential mode noise and that can withstand strong common-mode noise. One guideline I have seen in the past when shielding is needed between two enclosures is to use a capacitor to couple the shields between the two grounds in the enclosures. This is better than grounding the cable at only one end from an EMC perspective because that cable shield might act as a resonator or antenna. The cap would block DC offsets and strong noise at power frequencies, but it allows high frequency (RF) noise to dissipate in either side. The cap and the native DC shielding in the cable basically make a high-pass filter, so you should size the cap accordingly.
@Parvi_
3 жыл бұрын
What if two different boards powered through two different power supplies communicate on Serial protocol, eg, UART , and face issues due to mentioned problem in video? Should we make the TX/RX line btw two boards differential?
@Zachariah-Peterson
3 жыл бұрын
Hi Raza, In a practical situation, you would generally not do this with two different power supplies unless you can tie the grounds together so that any GND offset is removed between the two boards. What you are describing, and the need to use differential TX/RX or bidirectional differential pairs happens in larger installations. For example, automotive data connections use CAN bus to transfer data within a vehicle over differential pairs. Ground offsets and noise are very common in an automobile, so it makes sense to use a low-speed differential protocol like CAN bus to deal with both problems. CAN bus is also used in industrial computing installations as part of distributed process control systems. This protocol can easily withstand about 5 V ground offset with most transceivers, although larger offsets are possible on some components.
@myetis1990
3 жыл бұрын
PE is connected to chassis of the atx case and the GND of the mainboard is connected to chassis through mount holes. Is it a right practise?
@Zachariah-Peterson
3 жыл бұрын
Yes that is an appropriate practice in many cases, it just depends on the specific noise or safety problem you need to deal with. That type of practice normally targets RF noise by giving it a very low impedance path to reach GND at multiple points on the board. The other reason you might do that on a computer specifically is to make a connection between some exposed metal (like a connector) and your safety ground (the chassis in this case) in case an ESD event occurs near that connector. That way, any current that gets excited in the board during ESD goes into the chassis (your safety ground) instead of into your circuits. However, you should ask yourself, do you know for sure those mounting holes are grounded? They might be grounded to the case, but not necessarily to the board. Just because you see exposed copper on a mounting hole, it doesn't mean that the copper on the mounting hole is connected to the internal GND plane. Normally it is grounded like this in computer motherboards, but this is just something to think about in general.... As an example from industrial Ethernet design, the exposed metal shroud on the connectors can be a place where an ESD event causes an electrical pulse to enter the board, and so these points can be grounded directly to the chassis. There is a good Renesas application note that shows how to do this (see page 11 and 31): www.renesas.com/us/en/document/apn/industrial-ethernet-phy-r19an0015ed0102-rev102?language=en In this system the portion of the board near the power input would also be grounded back to the frame, so you have at least two points where the board and frame grounds are bridged. This could create a big ground loop, but then what they do in that system is separate the GND plane between the Ethernet PHY and the magnetics circuit/RJ45 connector, so the loop is broken because the GND planes are separate. In that case, the only signal you should route over the gap in that plane is the Ethernet traces as they are differential pairs, and even then you should only do this at low speed (10/100) because the carrier frequency is pretty low. I also don't agree with splitting the AGND and DGND sections of the ground layer as shown in page 31; as soon as you route a digital signal over the gap between DGND and AGND, you can create strong radiated emission. There is also a lot of debate as to whether GND should be cleared from beneath discrete magnetics circuits. I discussed some of this in a Signal Integrity Journal article last year: www.signalintegrityjournal.com/articles/1808-ethernet-connectors-and-routing-above-ground-planes
@Parvi_
3 жыл бұрын
Please also make a video about "How Copper thickness is selected for different layers in a board stackup". Thanks.
@Zachariah-Peterson
2 жыл бұрын
Hi Raza, we just published something on copper thickness, but it was less about selecting for specific layers. I'll add something like this to our list.
@Parvi_
2 жыл бұрын
I already watched that very informative video of yours, titled: "Copper thickness Deep Dive". Enjoying your video series, Thanks for sharing.
@rafalzasada8826
3 жыл бұрын
This video does not show how both PCB are powered up so it is quite difficult to understand how you could get 10V difference in grounds.
Пікірлер: 20