CanBus on Transit Custom

If you are not a PicoScope user and want advice with a diagnostic problem then post here. Forum members may be able to help.
bullstar
OneWave
OneWave
Posts: 29
Joined: Wed Nov 21, 2012 9:48 pm

CanBus on Transit Custom

Post by bullstar »

transit 2014 canbus.psdata
(657.62 KiB) Downloaded 893 times
Transit custom 2014 model 2.2 tdci came in to and diagnostic tool came up with 11 faults and most were "U" Codes no coms to Nodes so looked like a can bus fault, I set up the canbox and looked at can H and L , These were the results, My question is does this look correct or to much noise and the peaks seen to go over 3.5 on H and 1.5v on Low.
I found a poorly repaired can bus wire and and there seemed to be corrosion on the line, we have ordered a new harness so I traced the old wire back and found corrosion about 3.5ft back along this wire before the copper looked as it should.
Van was in limp mode, speedo not working, abs fault, airbag fault, traction body module, instrument module.

User avatar
STC
Banned
Posts: 790
Joined: Sun Sep 13, 2015 2:10 am

Re: CanBus on Transit Custom

Post by STC »

My question is does this look correct or to much noise and the peaks seen to go over 3.5 on H and 1.5v on Low.
The noise at the top and the bottom will not be your issue. The ECU's cannot see it. It only ever looks for a 2.0v voltage difference between the signals, not the signals themselves individually.
Can Transit.jpg
As you can see above in Frame 27 of your capture I have added a math channel, and Guidelines 2.0v apart, the Can Frames to the right do not cross that Magical 2.0 Volt threshold. It does happen in other parts of your capture randomly.

That is quite likely the reason for the fault not the noise.

bullstar
OneWave
OneWave
Posts: 29
Joined: Wed Nov 21, 2012 9:48 pm

Re: CanBus on Transit Custom

Post by bullstar »

Hi STC
thanks for your explanation on this, So when looking at can signals apart from seeing lines shorted etc, I should be looking for the 2.0v difference as in your drawing, So in frame 27 where the voltage is lower than 2.0v is this indicate a fauty node or connector.
I did say in my post that I thought the broken wire on the harness was canbus but its not its the wiring going to the rear abs sensor as Ive just been able to check Autodata ! (this confirms no data being received at abs ecu on live readings on that wheel)
Would this cause all failure on the networks not being able to communicate with each other ?
thanks

User avatar
STC
Banned
Posts: 790
Joined: Sun Sep 13, 2015 2:10 am

Re: CanBus on Transit Custom

Post by STC »

I should be looking for the 2.0v difference as in your drawing,
The Math Channel is just A-B, So at any given time the voltage difference between Can Hi & Can Lo is Either 0v or 2v (In an ideal world.)

When Hi & Low are dormant they will both be at ~2.5v. (2.5v - 2.5v = 0v)
When Hi & Low are recessive Hi will be at ~3.5v. Lo will be at ~1.5v. (3.5v - 1.5v = 2v)
So in frame 27 where the voltage is lower than 2.0v is this indicate a faulty node or connector.
It could be node, connector, resistance or even a short somewhere.

As with all Can faults start with measuring the resistance across the network, short to plus and minus and then start pulling nodes off the network (be aware of where the two 120 ohm resistors are located).
the broken wire on the harness ................... going to the rear abs sensor ..........

Would this cause all failure on the networks not being able to communicate with each other ?
Theoretically it would/should not but fix that first !!! and look for other wiring damage in that vicinity !!!

Dcunning35
TwoWaves
TwoWaves
Posts: 238
Joined: Thu Jan 28, 2016 12:33 pm

Re: CanBus on Transit Custom

Post by Dcunning35 »

Check the bcm and abs connectors they love a swim ,especially on these early models . Bulkhead drains water off at each side.
If you post your dtcs I will try and help , have you done a network test which modules dont respond .
Danny

User avatar
FioranoCars
TwoWaves
TwoWaves
Posts: 386
Joined: Thu Dec 13, 2012 11:26 am
Location: London, UK

Re: CanBus on Transit Custom

Post by FioranoCars »

STC wrote:I have added a math channel, and Guidelines 2.0v apart, the Can Frames to the right do not cross that Magical 2.0 Volt threshold.
The 2.0v is not the real threshold, I've seen good cars 1.8v to 2.2v and your measurement location and grounding (ecu ground verse chassis etc) can be confusing as to your measurements verse the ECU's real measurement.

Another case of needing to ground any 5v or CAN/Data signal from the ECU ground NOT chassis ground, or you'll get very poor results especially if doing precise tests!

With CAN it's about the symmetry, the signals should mirror, which for a 2.5v idle state they should be circa 1v increase for Can High and circa 1v drop for CAN Low, and if you math channel the 2 (High-Low) hopefully the result will reduce the noise and give a circa 2v Zero based signal - this is what the CAN transceiver are looking at, they then fall back to single signals only if this fails to give whatever threshold was set by the chipset designer ... 1v 1.5v 1.8v but never 2v as the threshold, it's not ever going to happen at the full ideal spec, it'll be slightly lower, always.

Some CAN or quasi CAN might have different voltages, and K-Line can be single 12v or 0v with 5v or 12v signal amplitude, if you suspect the amplitude, then you need to know the spec for your vehicle/specific BUS but looking at your signals, they can appear to be fine, notwithstanding you need to replace the corroded wires, the signals appear fine and I'm sure if you turn on the PICO decoder you'll be able to get the signal decoded to prove it.

Certainly the wires don't appear shorted at the measurement point you are in at, more on this in a second...

CAN bus ohms is another dangerous test, while the anti reflection original specification was to use 2x 120ohm terminators, much like old ethernet coax cables, giving a 60ohm value on the bus (when powered off, of course!), this has been tweaked, and now there are many ECU's that have complex resistances, so MED9.9.1 that we see, use an internal 240ohm in both (dual ecu car), other ECU's have no termination, one in that car has 120ohm, and overall the residence is close to the 60ohm spec, but it's not 60ohms. Many use external resistors. An open or short on the can bus are giveaways of an issue, don't get hung up on that while you have clearly a viable data set. Ford I know have some strange resistances built in to some ECU (65kohms I think was 1 from memory, which or what don't recall, sorry)

Noise is another red herring, it might be an issue, but often you'll chase your tail.

Next up is a CAN Topology, to understand where on the car a fault might lie, use the Topology to work out which ECU is talking on which segments and which ECU are past a break/poor wiring and hence giving no or intermittent comms.

Many ECU have within the parameter data/status some CAN flags about what's talking ok or not, sometimes they may be flagged here as "not OK" but no error is set!

Use this information with your Topology diagram and then start measuring at the ECU's using the farthest ECU as a base point on the first 2 scope channels as your benchmark, and then the other 2 channels move ECU to ECU along the bus, where there is deviation you are the other side most likely to the wiring fault, of course some faults are in the CAN transmission/receiver boards of the actual ECU or the external resistors. Wiggle tests and the usual tricks all help to prove and trace wires while viewing the datastream. Oh, did I mention use the ECU earth as your ground NOT chassis!!!

Unless you have need to, wait for your new harness to arrive and test once it's hooked up, but still accessible, seems pointless other than to learn and be curious to test until you fix a known issue, while I'm curious, I've learnt that a sore head can sometimes follow swiftly from playing at this stage!! :D

So CAN bus is not simple, but not complicated, don't try to read too much into minor things, but bear them in mind, it might be the issue, but could be a dead end. Get as much tech data as possible, certainly a Topology, and the pairs verse single wires and their voltages/speeds. Focus from the ECU with common errors or use the topology to decide where an error might be and test either side of the suspect point of failure. Look at the Net (maths channel) voltage, and see if you have a good 1.5v+

Of course good old visual loom and connector inspection are a great start, with contact cleaner spray being an essential treatment to anything you separate or disconnect!

Let us know where you took the captures (ECU pins or otherwise) and if you have other captures showing extreme locations on the Topology ... and of course where you took earths!

Best Richard

User avatar
STC
Banned
Posts: 790
Joined: Sun Sep 13, 2015 2:10 am

Re: CanBus on Transit Custom

Post by STC »

If we did not over complicate Can - it would be easy - because it is and was designed that way. Reliable and robust !!!
The 2.0v is not the real threshold

Really ?
Last edited by STC on Sun Mar 05, 2017 5:16 am, edited 1 time in total.

User avatar
FioranoCars
TwoWaves
TwoWaves
Posts: 386
Joined: Thu Dec 13, 2012 11:26 am
Location: London, UK

Re: CanBus on Transit Custom

Post by FioranoCars »

STC wrote:Really, Prove it ???
Simple, take any sensor or other ECU using CAN to communicate over a specific can-bus and is the ECU able to display the parameter data that it should see, as plausible data ... seriously though, with a specification saying 2.0v is ideal, the true threshold and redundancy/noise and other things going on, I'd "suspect" that even a 1.0001v "nett" signal would be readable by the ECU, maybe not, but certainly 1.6-1.7v would be, and 1.98v is not going to give an ecu any issues, plus it would revert to single wire mode if it could not, possibly generating a different error in the process, but still allowing the ecu to understand the signal ... so look at the Error codes, the status flags and the data the ECU can see, then you'll know if the ECU is indeed successful in decoding the signals ...

There is by design a huge amount of leeway built in and failover / fallback modes to ensure that many common wiring defects can be corrected without affecting the vehicle operation (as in dead, but maybe imposing a limp home mode).

Similarly, the Termination for anti-reflection is not about the 60ohms, it's about anti-reflection of the signals, that's why the VM's and component OE's have started to be creative with the resistances, to best suppress the ghost/reflected voltages, and using just 2 end points, was at best a big compromise, so multipoint termination, especially on a star topology is essential, and as speeds increase and number of ECU's do, this will become more complex if you want multi point failure robustness.

I don't have a VM figure for minimum voltage to be acceptable for can to "operate" and suspect it's rarely discussed or published outside the core chipset developers (ie TI or the like, rather than Bosch or Delphi) but if anyone can enlighten, I'm all ears :roll: :roll:

More to the point, knowing exactly where you are measuring what, and ground is key, will be the critical element of this, if not using ECU ground, then any measurement of Potential Difference are completely wasted ... it's not what the ECU is seeing, so is pretty irrelevant ... Amplitude and noise both...

As always Just IMHO, but based on a huge amount of research in this instance!! :D

Always willing and want to learn more, so anything factual, bring it on ...

Richard

User avatar
Fat Freddy
TwoWaves
TwoWaves
Posts: 335
Joined: Fri May 04, 2012 4:52 am

Re: CanBus on Transit Custom

Post by Fat Freddy »

Here's an example of a bad CAN bus. Astra DOA but one of three scan tools was still able to read live data. A good example of how robust CAN can be and with the right software and hardware it's still valid.
Disconnection of engine ECU cleared up the mess and IIRC I was able to access all the data on the PCM.

Applied maths looks good.


Cheers
FF
Attachments
BMT 67X Astra H 9-08 CAN bus at OBD connector. Engine ECU connected DecSSreading live data.psdata
(3.48 MiB) Downloaded 762 times

User avatar
FioranoCars
TwoWaves
TwoWaves
Posts: 386
Joined: Thu Dec 13, 2012 11:26 am
Location: London, UK

Re: CanBus on Transit Custom

Post by FioranoCars »

Hi FF

Amazing how some tools/chipsets can read from a single channel, that itself seems to be beyond usable!
Fat Freddy wrote:Applied maths looks good.
Could not see any maths (or undisplayed Maths channels) ?!

Separately do you have a capture of the same when fixed, curious as to voltage levels

Thanks

Richard

Post Reply