Here is CAN protection circuitry as suggested by MORNSUN in their datasheet of CAN transceiver with integrated power. I have few questions regarding few components used there:
I will answer some of your questions:
What is the purpose of R3 and C1?
You have a digital circuit that its ground (i.e. digital ground) should/can not be connected to the earth (chassis ground) directly. And you have high-speed signals flowing around which may cause EMI problems due to high-speed switching. To manage this and to prevent other systems connected to your circuit to be affected, you want the excessive noise to go to the earth. Hence the capacitor C1: Provides the shortest path from digital ground to earth at very high frequencies (i.e. noise) by its nature. C1 is open at low frequencies or DC, but you want a path from your digital ground to the earth. Hence the resistor R3. 1M/1n pair is widely used for that purpose (e.g. ethernet chassis to earth).
What is the purpose of R1 and R2?
They are protection resistors. Since there are TVS diodes across the CAN lines, excessive clamping currents may flow through them when there are spikes (e.g. lightning) 1. Those resistors simply limit the clamping current so that the bus does not be latched up by the transceiver by any chance. Their values should be low enough not to affect the line impedance (120 Ohms).
1 In some applications where the spike immunity requirement is higher, MOVs are used instead of TVS diodes.
What is the purpose of D1 and D2?
I don't know. I've never seen such an implementation. Hope someone answers.
Why aren't TVS from CANH and CANL to GND recommended?
Who says that? There are many applications having uni- or bi-directional TVS diodes from CAN lines to GND.
I think that the protection circuit has to be treated with suspicion. I say this because the circuit has a GDT as the first line of protection and, that doesn't mean that it's there to protect against puny levels of ESD but, "more muscular" indirect lightning surges (as per the words below the circuit).
Now, indirect lightning surges are a totally different game when it comes to rating components and, the GDT used is a good example of a "muscle" device. Once triggered (it can take several microseconds to activate) it will clamp down to about 8 volts (see red below): -
To trigger the device takes a DC value of about 68 volts to 112 volts (purple box above) and here's where the whole protection scheme is on shaky ground. For a few microseconds there might be a massive current surge into the TVS diode. This is because a GDT just does not respond that quickly. During this short time the terminal voltage might reach several hundreds of volts. The GDT data sheet suggests that the impulse spark-over voltage might be as high as 700 volts for an impulse rising at 1 kV per μs.
So somewhere between an impulse of 100 volts per second and 1 kV per microsecond you will see a terminal voltage that might be many hundreds of volts. I could make an estimation of 400 volts for 4 μs. During that time the peak current into the TVS will only be limited by the 2 watt 2.7 Ω resistors in each leg. Given that the TVS has a maximum clamping voltage of 48.4 volts (at a peak current of 12.4 amps) means about 350 volts across the two 2.7 Ω resistors.
That's a peak current of about 65 amps: -
So, the TVS can handle it but, the series 1N4007 diodes offer no evidence that they can handle that sort of current. So, when you ask this: -
What is the purpose of D1 and D2?
I have to say that it looks like some bodge to fix something and I doubt that they will survive any meaningful lightning protection test. As a guess, I would say that the self-capacitance of the TVS of several nano farads might disrupt data sufficiently and, the forward voltage drop of the 1N4007 gives just enough extra headroom on data amplitude to ensure that it works. The 1N4007 has a reverse capacitance of only a few pF so it could improve data integrity against the backdrop of excessive natural capacitance in the TVS diode: -
What is the purpose of R1 and R2?
I've explained that earlier i.e. they limit current into the TVS and protect it but, you need to ensure that those resistors can take a massive surge power of over 10 kW each for a few microseconds.
And finally, with any over-voltage protection scheme offered, there should be a statement about what it is intended to protect against i.e. the threat should be named explicitly (as per EN 61000-4-5 for instance). The words below the Mornsun circuit are these: -
And, they are insufficient in my opinion.
I think that the 1N4007 diodes can handle the surge current because they will operate in forward mode always. And in forward mode their IFSM is 30 A for a surge duration of 8.3 ms (sine-wave) which is significantly long as a surge duration. As a comparison, the SMBJ30CA is rated at around 230W at 8 ms so by taking a clamping around 55V (best case), the surge current IPP ~ 4 A
https://www.vishay.com/docs/88392/smbj.pdf
So the SMBJ30CA will burn before the 1N4007. This kind of topology is quite common for communication bus protection : 2 low capacitance diodes working in forward mode in series with a TVS. As Andy Aka mentioned, the 2 low capacitance 1N4007 will drive the overall capacitance of the protection module.
This being said, the IEC61000-4-5 is meant to modelize surges on power mains and we often consider that these surges can affect the datalines by coupling. But obviously with much less energy than on power lines directly connected to the mains (conduction mode). That’s the reason why the IEC61000-4-5 requires a 42-ohm serial resistance between the surge generator and the DUT (Device Under Test). To be accurate it is a 40-ohm resistor as the internal resistor of the surge generator is 2 ohm. By the way the current waveform of the IEC61000-4-5 is an exponential waveform called 8/20 µs. Meaning that half of the peak current is reached at tp=20 µs.
This is the tp mentioned in the power derating curves from Vishay shown above. The 10/1000 µs that you will find in all the TVS datasheet comes from telecom standards (GR 1089) and is used as industry standard to qualify the TVS; For example the SMBJ30CA is a 600W TVS based on 10/1000µs surge. But this kind of surges are not representative of what happens in industrial environments and is only relevant for wireline.
https://www.st.com/resource/en/datasheet/smbj30ca.pdf
Back to the initial topology of MORNSUN, for me it seems oversized. Semiconductor vendors offer integrated solutions in SOT23 or SOT323 that can do the job. ST does not recommend SM15T but ESDCAN series connected to GND with no GDT and no resistor. This can cover system ESD up to 30 kV. System ESD is based on IEC 61000-4-2 for industrial devices and on ISO 10605 for automotive devices. This is much more stringent than HBM ESD ratings of ICs (simply because the purpose is not the same).
EarthLord,
For automotive, you have to consider various standards like indeed reverse battery, jump start.
This is explained in: CAN-bus-protection-ST-ESDCAN-series presentation
As mentioned by Andy, you can have surges coming from alternators called “load dump” but again they might appear only by coupling to CAN datalines and most of the time there is a centralized load dump protection in the cars clamping these transients at 35V (for 12 V battery vehicles).
I hope it clarifies.
