Author Topic: Isolated CAN ground connection  (Read 13360 times)

0 Members and 1 Guest are viewing this topic.

Offline Mad IDTopic starter

  • Regular Contributor
  • *
  • Posts: 167
  • Country: 00
Isolated CAN ground connection
« on: March 18, 2012, 03:21:11 pm »
Hi everyone.

In my current project I must use isolated CAN bus. Transceiver I'm using is from Analog Devices ADM3053 which includes an isolated DC/DC converter.
I'm confused by the datasheet (Figure 32) where a bus connector is depicted without a ground wire connection, i.e. only CAN+ and CAN- signals present.

I know CAN is differential but my opinion was that ground is needed especially with isolated CAN transceivers. (e.q. RS-485 standard clearly states that a ground wire should be present in the communication cable).

This is how I think it should be done: All CAN nodes have to be isolated except for one. This one provides the single-ground (earth) connection for the whole bus. Ground wire is carried along with the twisted pair to each node. No ground loop is created.

Please help me clarify this. Do I need a ground connection or not? Individual nodes are not! referenced to earth, if so isolation would be a waste of money.

Thank you!
 
 

Offline Performa01

  • Super Contributor
  • ***
  • Posts: 1729
  • Country: at
Re: Isolated CAN ground connection
« Reply #1 on: March 18, 2012, 08:39:25 pm »
Differential interfaces have a rather limited common mode voltage range. This is of course depending on the receiver, yet many CAN transceivers can handle just -2 to +7V.
On some sites, ground potential differences of several volts can easily build up over longer distances. Thus it certainly is a good idea to have all receiver grounds connected together by means of the ground wire of the communication cable. This does not create a ground loop since no signal is referenced to ground. That’s (one part of) the real beauty of differential pairs.
There might be some compensation current flowing through the ground wire, this should be low enough to not produce any significant voltage drop.

In noisy environments, shielded cabling is good practice anyway. This also helps to improve ESD immunity. The shield can only be effective when there is a low impedance path to the ground of the circuit that shall be protected. One single ground connection in an installation with several tenths of meters cable running is certainly not low impedance for higher frequencies. If galvanic isolation is really required for some reason, then each node should at least be connected to the shield by a HV capacitor (e.g. 10nF/2kV).

Well, that’s how I handle that very issue in my current project – it’s not proved in practice so far, as I’m still busy getting one single device working with all its features, so external connections over longer distances are designed in (CAN being one of them), yet of no relevance right now.

Hope this helps a bit to make up your mind! ;)
 

Offline gxti

  • Frequent Contributor
  • **
  • Posts: 507
  • Country: us
Re: Isolated CAN ground connection
« Reply #2 on: March 18, 2012, 08:43:16 pm »
That's a very neat part, It costs twice as much as a potted DC-DC converter in small quantities, but still damn cool. If you want to save a few bucks maybe look into using a seperate transceiver, isolator, and isolated power supply -- it seems to be a bit cheaper for small quantities. If you need a small footprint though, this chip can't be beat.

I'm confused by the datasheet (Figure 32) where a bus connector is depicted without a ground wire connection, i.e. only CAN+ and CAN- signals present.

I know CAN is differential but my opinion was that ground is needed especially with isolated CAN transceivers. (e.q. RS-485 standard clearly states that a ground wire should be present in the communication cable).
I'm not sure either. There's no mention either way in the datasheet other than that diagram, so I would go with your intuition and use a ground wire.

Quote
This is how I think it should be done: All CAN nodes have to be isolated except for one. This one provides the single-ground (earth) connection for the whole bus. Ground wire is carried along with the twisted pair to each node. No ground loop is created.
I don't see much point in strongly tying the bus ground to anything. Use the isolated transceivers everywhere, and if you're feeling frisky, use a 1M resistor to reference bus ground to system ground at exactly one point, just to keep it from floating around.
« Last Edit: March 18, 2012, 08:47:03 pm by gxti »
 

Offline Mad IDTopic starter

  • Regular Contributor
  • *
  • Posts: 167
  • Country: 00
Re: Isolated CAN ground connection
« Reply #3 on: March 19, 2012, 07:34:11 am »
That's a very neat part, It costs twice as much as a potted DC-DC converter in small quantities, but still damn cool. If you want to save a few bucks maybe look into using a seperate transceiver, isolator, and isolated power supply -- it seems to be a bit cheaper for small quantities. If you need a small footprint though, this chip can't be beat.

Not sure if save is possible. The cheapest isolated DC-DC converted I've been able to find is 6$ (http://search.digikey.com/us/en/products/NTE0505MC/811-1679-5-ND/1927240) @ 500 pcs. On top of that I need a isolated transceiver which is at least 3$ (e.q. ISO3080 from T.Instruments).

With Analog Devices IC I have two completely isolated interfaces i.e. they have seperate power supplies and each can communicate with different remote equipment.
Please recommend a cheaper solution if there is one :) ?

I don't see much point in strongly tying the bus ground to anything. Use the isolated transceivers everywhere, and if you're feeling frisky, use a 1M resistor to reference bus ground to system ground at exactly one point, just to keep it from floating around.

Take a look at this app note from TI.
http://www.ti.com/general/docs/lit/getliterature.tsp?literatureNumber=slla268&fileType=pdf

Page 7, paragraph in the middle.

Thanks for your help.
 

Offline Short Circuit

  • Frequent Contributor
  • **
  • Posts: 439
  • Country: nl
    • White Bream electronics R&D
Re: Isolated CAN ground connection
« Reply #4 on: March 19, 2012, 11:54:42 am »
The CAN transceiver has relative low imput impedance and a 2~3V recessive bus state, so it will always track the bus CM voltage.

If you read the datasheet, you'll notice there is no mention of common mode voltage range, which implies no ground connection.
Basicly the CM range is the 2.5kV isolation voltage, with a limit of 25kV/us voltage change rate.
 

Offline gxti

  • Frequent Contributor
  • **
  • Posts: 507
  • Country: us
Re: Isolated CAN ground connection
« Reply #5 on: March 20, 2012, 03:12:48 am »
Not sure if save is possible. The cheapest isolated DC-DC converted I've been able to find is 6$ (http://search.digikey.com/us/en/products/NTE0505MC/811-1679-5-ND/1927240) @ 500 pcs. On top of that I need a isolated transceiver which is at least 3$ (e.q. ISO3080 from T.Instruments).

With Analog Devices IC I have two completely isolated interfaces i.e. they have seperate power supplies and each can communicate with different remote equipment.
Please recommend a cheaper solution if there is one :) ?
$5.24 Power supply: http://search.digikey.com/us/en/products/VBSD1-S5-S5-SIP/102-1360-ND/989815
$1.55 digital isolator: http://search.digikey.com/us/en/products/SI8431AB-D-IS/336-1758-5-ND/2170665
$1.12 CAN transceiver: http://search.digikey.com/us/en/products/MCP2551-I%2FSN/MCP2551-I%2FSN-ND/509452
All in singles. That's 7.91 USD total vs 11.83USD for the AD part. More parts obviously, and I didn't bother checking if they're actually compatible, but that's what I meant. You can also use different kinds of transceiver (RS-485, etc.). The one thing you won't get is safety isolation -- that DC-DC converter, as well as the one you found, do not have regulatory approvals and don't even have enough separation between the pins to serve that purpose. I didn't look very long for the cheapest safety converter, but obviously it's going to cost quite a bit more. Whether you need it or not depends on your application. Quite impressive that the ADM3053 manages to get that regulatory approval at such a low cost. By all means use it if it does what you want.

 

Offline Mad IDTopic starter

  • Regular Contributor
  • *
  • Posts: 167
  • Country: 00
Re: Isolated CAN ground connection
« Reply #6 on: March 21, 2012, 07:40:11 am »
The CAN transceiver has relative low imput impedance and a 2~3V recessive bus state, so it will always track the bus CM voltage.
If you read the datasheet, you'll notice there is no mention of common mode voltage range, which implies no ground connection.
Basicly the CM range is the 2.5kV isolation voltage, with a limit of 25kV/us voltage change rate.

Thank you. This answer is correct. I received similar answer (i.e. ground not needed) directly from Analog Devices support forum.
I will use this opportunity to ask you the same question but for RS-485. Does isolated RS-485 need ground? There is a lot of confusion on this matter on the internet. I have at least 5 application notes (e.q. from TI) which state that ground is needed but still know a lot of working systems which don't use the ground wire.

What's your opinion? BTW TIA/EIA RS_485 standard says to use the ground wire. Weird how people omit it.
 

Offline Short Circuit

  • Frequent Contributor
  • **
  • Posts: 439
  • Country: nl
    • White Bream electronics R&D
Re: Isolated CAN ground connection
« Reply #7 on: March 21, 2012, 09:27:30 am »
I believe RS485 should have a ground connection to the (isolated) transceiver. That's because the TX-off state is high impedance, which allows the bus to float out of CM range. However, some RS485 nodes employ a biasing network consisting of a fairly high valued 'termination' and pullup + pulldowns to power rails. Older RS485 transceivers don't have defined RX condition and this allowed for a known bus state with no active drivers. The same network will also bias an isolated transceiver to the bus CM voltage, so things should work nicely. And even without such biasing, imput resistance will likely decrease when the imputs drift outside Vin range, therby biasing before the receiver is actually damaged. But I would not consider this a very reliable and failproof mechanism...
 

Offline AndersJ

  • Frequent Contributor
  • **
  • Posts: 415
  • Country: se
Re: Isolated CAN ground connection
« Reply #8 on: September 04, 2019, 06:16:52 pm »
Does anyone have any experience, good or bad,
using the ADM3053 isolated CAN interface mentioned in this thread?
Any grounding issues?
Other issues?
Is the chip a good choice for new designs?
Are there other similar ”all in one” chips available?
"It should work"
R.N.Naidoo
 

Offline Jeroen3

  • Super Contributor
  • ***
  • Posts: 4209
  • Country: nl
  • Embedded Engineer
    • jeroen3.nl
Re: Isolated CAN ground connection
« Reply #9 on: September 04, 2019, 06:46:59 pm »
Any further standardized application CAN bus does require signal ground, CAN_GND next to CAN_H and CAN_L.
Common mode range for this transceiver is 36 volts.
Common mode range for the isolator is 2.5kV.
You could theoretically use split termination and not use ground. But you shouldn't.
On your desk CAN without ground is fine. With 20 nodes, it is not.

Use a small Murata NXF1 for low quantity, and SN6501 driver with Wurth transformer for high quantity. These all in ones are risky and not worth it, poor lead times. Don't go there for long time production.
Those all in ones are literally three (or more) dies in one package.
« Last Edit: September 04, 2019, 06:56:24 pm by Jeroen3 »
 
The following users thanked this post: AndersJ

Offline AndersJ

  • Frequent Contributor
  • **
  • Posts: 415
  • Country: se
Re: Isolated CAN ground connection
« Reply #10 on: September 04, 2019, 07:27:54 pm »
Thanks,
exactly the type of comment I was hoping for.

It's interesting that you mention the other options.
I have been considering the:
* Texas SN6501 driver
* Wurth 760390014 transformer
* Texas ISO7721 digital isolator

The above components are taken from Texas Instruments App note TIDA-01487.
Yes, single source, sort of, but not as single as the ADM3053.
"It should work"
R.N.Naidoo
 

Offline tru3533

  • Regular Contributor
  • *
  • Posts: 130
  • Country: no
Re: Isolated CAN ground connection
« Reply #11 on: September 04, 2019, 09:31:47 pm »
CAN bus signals have nothing to do with ground!

The CAN bus is an differential bus, meaning the voltage forming the bits is measured between the CANH and CANL wires.

So if noise interference happens, it will be implemented on both lines, and this will not cause any problems as we are measuring between the CANH and CANL wires. (same thing happens on RS485)

RS232 is referenced to ground, here noise is catastrophic as the ground is not floating.

Also notice the twisting of the CAN wires, this is very important as this will cancel the noise. 
 

Offline Jeroen3

  • Super Contributor
  • ***
  • Posts: 4209
  • Country: nl
  • Embedded Engineer
    • jeroen3.nl
Re: Isolated CAN ground connection
« Reply #12 on: September 05, 2019, 05:58:43 am »
Check the Silicon Laboratories Si8622 isolators.
 
The following users thanked this post: AndersJ

Offline AndersJ

  • Frequent Contributor
  • **
  • Posts: 415
  • Country: se
Re: Isolated CAN ground connection
« Reply #13 on: September 05, 2019, 08:54:59 am »
CAN bus signals have nothing to do with ground!

The CAN bus is an differential bus, meaning the voltage forming the bits is measured between the CANH and CANL wires.

So if noise interference happens, it will be implemented on both lines, and this will not cause any problems as we are measuring between the CANH and CANL wires. (same thing happens on RS485)

Grounding does not always concern only noise.
It is also needed to keep tranceivers within reasonable common mode levels,
especially on long cables between remote locations.
"It should work"
R.N.Naidoo
 

Offline tru3533

  • Regular Contributor
  • *
  • Posts: 130
  • Country: no
Re: Isolated CAN ground connection
« Reply #14 on: September 05, 2019, 10:40:28 am »
From my point of view you only need isolated CAN bus for your test equipment.
 

Offline AndersJ

  • Frequent Contributor
  • **
  • Posts: 415
  • Country: se
Re: Isolated CAN ground connection
« Reply #15 on: September 05, 2019, 10:48:23 am »
From my point of view you only need isolated CAN bus for your test equipment.

Why only for test equipment?
Why is isolated CAN not needed when communicating between distant locations
such as in industrial environments and/or between buildings with differing ground potentials.

"It should work"
R.N.Naidoo
 

Offline tru3533

  • Regular Contributor
  • *
  • Posts: 130
  • Country: no
Re: Isolated CAN ground connection
« Reply #16 on: September 05, 2019, 05:13:07 pm »
Quote
Why is isolated CAN not needed when communicating between distant locations

Maximum length due to Signal Propagation Time on CAN bus
1Mbps = maximum 50m
500kbps = maximum 120m

Quote
industrial environments

Why do you think isolated CAN bus don't exist in a car with hundreds of sensors on CAN bus?
Answer: It works perfectly without.


« Last Edit: September 05, 2019, 05:45:20 pm by tru3533 »
 

Offline AndersJ

  • Frequent Contributor
  • **
  • Posts: 415
  • Country: se
Re: Isolated CAN ground connection
« Reply #17 on: September 05, 2019, 05:58:08 pm »
Quote
Why is isolated CAN not needed when communicating between distant locations

Maximum length due to Signal Propagation Time on CAN bus
1Mbps = maximum 50m
500kbps = maximum 120m

Quote
industrial environments

Why do you think isolated CAN bus don't exist in a car with hundreds of sensors on CAN bus?
Answer: It works perfectly without.

1. Isolation is not used to resolve timing problems. It is used to avoid problems caused by common mode voltages.
2. A car has a common well defined ground. Isolation is not needed.

"It should work"
R.N.Naidoo
 

Offline tru3533

  • Regular Contributor
  • *
  • Posts: 130
  • Country: no
Re: Isolated CAN ground connection
« Reply #18 on: September 05, 2019, 07:43:17 pm »
1. If your distant house/building is more than 50m away. CAN bus is useless on 1Mbps.
With or without isolation, that's my point.
 

Offline AndersJ

  • Frequent Contributor
  • **
  • Posts: 415
  • Country: se
Re: Isolated CAN ground connection
« Reply #19 on: September 06, 2019, 02:53:53 am »
1. If your distant house/building is more than 50m away. CAN bus is useless on 1Mbps.
With or without isolation, that's my point.

Your point is taken.
But,
this thread is about isolated CAN and whether signal ground is needed or not.
Bit rate is irrelevant.
Isolation is needed for other reasons than speed.
"It should work"
R.N.Naidoo
 

Offline Jeroen3

  • Super Contributor
  • ***
  • Posts: 4209
  • Country: nl
  • Embedded Engineer
    • jeroen3.nl
Re: Isolated CAN ground connection
« Reply #20 on: September 06, 2019, 05:43:49 am »
I'm a bit rusty on the bit-timings of CAN bus and I never gave it much thought, but does the delay in the isolator reduce the maximum bus length?
 

Offline Dave

  • Super Contributor
  • ***
  • Posts: 1356
  • Country: si
  • I like to measure things.
Re: Isolated CAN ground connection
« Reply #21 on: September 06, 2019, 06:16:12 am »
I'm a bit rusty on the bit-timings of CAN bus and I never gave it much thought, but does the delay in the isolator reduce the maximum bus length?
CiA 301 (page 23) states that bus lengths are based on an estimate of 5ns/m propagation delay. So yes, the response time of the isolator should reduce the maximum usable bus length.

The isolators supposedly have a 9ns delay in one direction (Si841x/2x - the ones I used in my design), so that should reduce the maximum bus length by just over 7m (communication between two isolated devices).
« Last Edit: September 06, 2019, 06:17:50 am by Dave »
<fellbuendel> it's arduino, you're not supposed to know anything about what you're doing
<fellbuendel> if you knew, you wouldn't be using it
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf