Ask: Why SP485REN-L/TR is frequently damaged

chenzhouyu

Ask: Why SP485REN-L/TR is frequently damaged [Copy link]

 

Hi everyone, I have a question about 485.

The whole device is a metal shell, which is used to monitor the status of the generator of the marine generator set. The metal shell is connected to a protective ground wire through the connector on the main device PCB

Connected to the outside, no electrical connection with the power ground and signal ground.
The 485 network has five nodes, two of which are configured with 120 ohm terminal resistors, A on each node is added with a 1K pull-down resistor, and B is connected to a 1K pull-up resistor.
The wiring of each 485 node is very short, no more than 50 cm (since the nodes are all in a metal shell, no isolation solution is used), and the power supply of the four slave devices is

Autonomous equipment.

tagetage

Is the 485 line a shielded line with single-ended grounding?

chenzhouyu

tagetage posted on 2022-10-26 23:54 Is the 485 line a shielded line, with single-end grounding?

The wire is very short, inside a metal casing, and no shielding is used

bigbat

It is recommended to add a voltage regulator to the power supply end of all SP485. I suspect that the fault is caused by the main power supply system, not the fault introduced by the 485 line.

langtuodianzi

Enhanced protection.

呜呼哀哉

I encountered the same problem. My circuit structure is the same as the original poster's. The post I posted is a comparison of circuit structures. The original poster can also take a look to see if our problems have anything in common.

呜呼哀哉

bigbat posted on 2022-10-27 09:09 It is recommended to add a voltage regulator to the power supply end of all SP485. I suspect that the fault is caused by the main power supply system, not the fault introduced by the 485 line

Can you explain this in detail? I have the same problem, circuit structure comparison, and I also considered it because the RS485 chip was damaged.

bigbat

Alas, posted on 2022-10-27 09:36 Can you explain this in detail? I have the same problem, and compared the circuit structure. I also considered it because the RS485 chip was damaged.

Differential transmission does not require a common ground. Theoretically, as long as there is a voltage difference between AB, transmission can be performed. However, the OP's device has a common ground and a common power supply. It is better to have a common ground. If the power supply is: different devices may have different power supply requirements. In a complicated situation, some devices even have a negative power supply. This can result in multiple "reasonable power supplies", which is not interference. There should be as many potential differences as possible. The two pull-up and pull-down resistors R201 and R207 are introduced into the power supply system of the chip. At this time, the chip may have a higher negative voltage, thereby damaging the chip. Therefore, I suggest that the entire 485 communication system use an "independent" power supply system. It is best to isolate it. If it is isolated, use a clamping system for power supply. Treat the entire system as a signal. Of course, this method may damage the MCU main system. The most recommended method is to isolate the 485 system independently.

Gen_X

1. Due to the use of 5V circuit, the 485 driving ability is limited, it is recommended to pull up and down to 4.7k or more

2. If the distance is short or the speed is not high, you can use 150 ohms instead of 120.

3. It is recommended to replace the TVS with a 6V one.

4. Ferrite beads or printed inductors can be considered for RS485 inlet or outlet.

chunyang

There is strong electromagnetic interference near the generator. With a 50cm spacing and 5 nodes, the bus length should be about 2 meters. Proper protection must be considered in a strong electromagnetic environment. However, the information provided by the original poster is not enough. The design of all cable routing and PCB design must be examined. The original poster probably did not realize that a routing of tens of centimeters in a strong electromagnetic environment may induce a high enough voltage.

At least, there is a design error in the original poster's circuit. The bias resistor should generally be arranged on the host side and calculated using Ohm's law with a 200mV threshold. It can of course be arranged at each node when no expansion is required, but the resistance value is different from that of a single-point bias. Pay attention to the synthetic impedance.

In addition, TVS protection should also be added between the AB lines to prevent excessive differential mode voltage. Strong differential mode voltage is entirely possible in a strong electromagnetic environment.

At the same time, please note that bus cables in strong electromagnetic environments should use shielded wires, 3-core or 4-core, depending on whether remote power supply is required. If it is required, use 4-core, if not, use 3-core. Except for the AB line, the other lines are power and ground. When the remote load is heavy, use cables with more cores, and the power and ground are connected in parallel. The shielding layer of the shielded wire should be connected to the protective ground.

tagetage

chenzhouyu posted on 2022-10-27 07:49 The wire is very short, in a metal casing, and no shielding is used

Then consider the conducted interference, change the power supply to an anti-interference one. The interference of the generator on the ship is very large, larger than you can imagine.

呜呼哀哉

Friends, which of these answers do you prefer? Do you have any ideas? I feel sorry for you. I really don't understand the problem.

chenzhouyu

tagetage posted on 2022-10-27 13:51 Then it is considered to be conducted interference. Switch to an anti-interference power supply. The interference of the generator on the ship is very large, larger than you can imagine.

It is possible that it was not exposed during reliability testing within the company.

chenzhouyu

chunyang posted on 2022-10-27 13:43 Strong electromagnetic interference cannot be ruled out near the generator. With a spacing of 50 cm and 5 nodes, the bus length should be about 2 meters. Proper consideration must be given to strong electromagnetic environments...

谢谢chunyang版主，看了您的回复我还有几个问题需向您请教。
1、我看了产品的内部结构，电线比较多，不规整。PCB的保护地和信号地间距很小。
2、偏置电阻我重新调整下，我目前每个节点都有偏置电阻，并放置两个终端电阻，是否会因过载导致芯片损坏？
3、增加AB线间的TVS管，SMBJ6.0CA。
4、在A、B线上分别增加两个箝位二极管连接到5V和GND,同时在5V电源线上增加SMBJ5.0CA的TVS管，对感应电压能否起到保护作用？

chenzhouyu

Gen_X published on 2022-10-27 11:26 1. Due to the use of 5V circuit, the driving capability of 485 is limited. It is recommended to use 4.7k or more for pull-up and pull-down. 2. If the distance is short or the speed is not high, 120 can be used instead. ...

Ok, thanks for your suggestion.

damiaa

If the chip is damaged due to line interference of ab line, you can look at the left circuit of the 5th floor, the circuit on the right is weaker.

I have encountered 485 chip damage before, and it is a headache, and it often breaks down. Using a protection circuit can basically solve the problem.

chunyang

chenzhouyu posted on 2022-10-28 08:15 Thank you, moderator chunyang. After reading your reply, I still have a few questions to ask you. 1. I looked at the internal structure of the product. There are many wires, not...

A better way to protect the bus is to use Schottky diodes to clamp the bus power supply and bus signal reference ground respectively, add a bidirectional TVS between the AB lines, and the action threshold level should be lower than the tolerance limit of the differential mode voltage of the interface device, and higher than the normal AB line voltage difference of the bus. The closer to the normal voltage difference, the better. The bias resistor must be adjusted to meet the 200mV threshold of 485. Excessive load generally does not cause damage to the interface chip, but only increases unnecessary power consumption. However, when there is strong external electromagnetic coupling, heavy load will increase the probability of chip damage, so the bias should not be excessive.

chunyang

Also note that if the bus topology is a forked star as shown in Figure 14, the terminal resistance is set to 120 ohms on the host side, and it must be added to each point on the terminal side, but the resistance value must be greatly increased, and the combined impedance cannot be lower than 60 ohms. If the bus signal transmission rate is low enough, the terminal side does not need a terminal resistance, after all, the bus is not long. The condition for not using a terminal resistance on the terminal side is that the bus length is less than 1/10 of the wavelength corresponding to the highest signal transmission rate.

This shows that your biasing and terminal processing are not appropriate. Be sure to change to single-point biasing and determine the appropriate terminal matching method.

chenzhouyu

chunyang posted on 2022-10-28 12:53 Also note that if the bus topology is like the bifurcated star in Figure 14, the terminal resistance is set to 120 ohms on the host side, and it should be added to each point on the terminal side, but the resistance value...

Thank you very much for your patience.

chunyang

You'd better change the bus routing to host-terminal 1-terminal 2-terminal 3-terminal 4, and then add terminal resistors at the host and terminal 4. In addition to the many improvement solutions I mentioned in my previous post, your problem is not a big deal and can be easily corrected. I'm afraid that you will take a chance and not strictly follow the specifications, then the consequences will be worrying, and you can only hope that the blind cat will catch a dead mouse and you will be in trouble.