Diagnosis of Canbus with Hantek 2D82 Auto (Oscilloscope)

Brokecanadian

2005 Cargo 2500 SHC NA
Yes, what I was hoping for was an obvious interruption so I could plug and unplug modules to find the culprit

Now, both rear sensors are pulled so how relevant this is I dunno. Yes scope will test sensors. Guess I'm pulling the jack out and spinning some wheels after I hook them up
 

Brokecanadian

2005 Cargo 2500 SHC NA
IF (and it's a big if) I'm using the scope correctly, my issue has more to do with wheel sensor output than a bad module. Good thing I didn't buy them all to throw parts at it

Going to try and get wheel sensor output at the ABS module and see what it's seeing

Since its a DMM too, maybe I should get the resistance of the modules just to rule it out and test that feature.
 

NBB

Well-known member
lol - GFL with that toy.

The Google suggests that scope is usable up through 30 or so MHz before attenuation, much higher if you accept the anti-aliasing filters - which will also remove anything of real interest. CAN is glacial in comparison - 100's of kHz - so that part shouldn't be an issue.

The Google also suggests this thing has no memory beyond what you see on the screen - so if you zoom in far enough to capture a transient (fault) - you'll likely be looking at such a small sample you won't likely see a fault.

An adult level, non-toy, for-realz scope is going to have a very deep memory buffer and sophisticated triggering to scan that buffer and would be capable of capturing faults - for like at least another $20-100k in hardware. With less - IMO you're wasting your time.

That said - nice tool for $100 or so? Just not for this.
 

Brokecanadian

2005 Cargo 2500 SHC NA
lol - GFL with that toy.

The Google suggests that scope is usable up through 30 or so MHz before attenuation, much higher if you accept the anti-aliasing filters - which will also remove anything of real interest. CAN is glacial in comparison - 100's of kHz - so that part shouldn't be an issue.

The Google also suggests this thing has no memory beyond what you see on the screen - so if you zoom in far enough to capture a transient (fault) - you'll likely be looking at such a small sample you won't likely see a fault.

An adult level, non-toy, for-realz scope is going to have a very deep memory buffer and sophisticated triggering to scan that buffer and would be capable of capturing faults - for like at least another $20-100k in hardware. With less - IMO you're wasting your time.

That said - nice tool for $100 or so? Just not for this.
It has a memory. Not sure where or how to use it. This is a preliminary thread to introduce the tool.

Also has a USB connector, whether for updating or output I dunno. NVM I see it takes an SD card. Going to see if I can download the output for analysis on a laptop :bounce:

Ok. Correction: USB connector not present and SD reader looks blank. Gonna see if a card just falls inside the case..
 
Last edited:

Midwestdrifter

Engineer In Residence
That output looks pretty good at a glance. Obviously the issue is intermittent. Have you confirmed the low and high sides between each other and ground?
 

NBB

Well-known member
It has a memory. Not sure where or how to use it.
Dude - no. What I described is what makes a real scope expensive. Your scope has a tiny buffer with trivial triggering. If you actually capture a CAN fault with that thing, you'll have won the lottery.
 

Brokecanadian

2005 Cargo 2500 SHC NA
Dude - no. What I described is what makes a real scope expensive. Your scope has a tiny buffer with trivial triggering. If you actually capture a CAN fault with that thing, you'll have won the lottery.
By that logic, a Picoscope should suck too. But very handy on vehicles.

That being said - still looking for memory on it lol
 

Midwestdrifter

Engineer In Residence
Logging continuously can be very helpful, but its a pain to dig through eye watering pages. Especially when you move past the basics, like, is there frames, and are the voltages right. The first step is trying to create the fault while your watching. If you can get more details on its nature, thats makes the investigations much easier.

In your case you just want to see if teh WSS signals are good. Then maybe you can move on to watching the canbus itself. I have successfully identified CAN modules with bad transceivers using just a scope. Its a simple system, so issues tend to show up in obvious ways. Voltages all over the map, etc.
 

Brokecanadian

2005 Cargo 2500 SHC NA
Ok. Specifications attached. I have no idea whether 6k samples is good or bad, but I found the memory. On training videos (for picoscope and similar laptop scopes), a bad module was clearly seen in real-time. Someone can tell me if this is a useless cheap Chinese tool (I happen to love cheap Chinese tools, they save me money)

Also worth mentioning: I did not calibrate scope. I watched a review of this against a 10x as expensive scope; when used with quality accessories there was a negligible difference. Included probes are cheap. Still I got a waveform. I don't care about accuracy, I care about the ability to see things like a module dragging canbus down
 

Attachments

Last edited:

Brokecanadian

2005 Cargo 2500 SHC NA
https://www.circuitspecialists.com/blog/what-is-oscilloscope-memory-depth/

Ok. So, explains what NBB meant.

Is this sufficient as a cheap scope to diagnose canbus issues? I'll try and do some research tonight. Barely getting my head around the relationship between memory speed and sample size/bandwidth.

Basically, my premise was: a broken module would stay broken, and as my issue repeats EVERY drive, it would show up if canbus related.

I'm starting to believe it's a failure mode of a module lacking some vital info due to my specific broken sprinter, and not a broken module. I'll add here when I discover anything new. What I see is normal can communication, at least so far.

Is this a useful tool? Dunno. I do know scopes similar can be as cheap as $50 usd...don't buy this without understanding it. Worth it to me so far tho
 

Brokecanadian

2005 Cargo 2500 SHC NA
Watching the wheel speed sensors directly can also have some value with a good scope. An intermittent failure in the ESP modules input circuit could pull a sensor signal down to zero randomly.
Hmm...actual error reported is wheel speed implausible, then a default value is used by module. Shifting stops working at same instant

Looks like I'll be tapping the WSS outputs next.
 

Midwestdrifter

Engineer In Residence
Make sure you do this with them connected to the ESP module. It could be the sensor or the module which is causing the issue.
 

NBB

Well-known member
So here's the deal - the CAN signal looks fine. If it didn't, you wouldn't know what to do anyway. You're not an R+D engineer to get anywhere beyond the fact the module is spitting out a somewhat valid looking signal - you're DONE. It's likely the problem is elsewhere. I'm no T1N expert, but I would guess corroded ground issues given various posts to this forum.

However, before you leave, this is a diff signal and I'm not sure where your ground reference in to the scope is. Might want to probe around a few other bus's that are working to see that the voltages are similar with respect to each other and ground. If not - this might be your grounding mentioned above.

EDIT:
To make my suggestion more clear - probe the bus's, zoom in on the waveform, and pay more attention to the voltage with respect to ground and each other. If you can, with the scope, reference a ground pin on each module - check both sides of the link. Mess with ground connections nearby and see if something changes on the scope.
 
Last edited:

Brokecanadian

2005 Cargo 2500 SHC NA
So here's the deal - the CAN signal looks fine. If it didn't, you wouldn't know what to do anyway. You're not an R+D engineer to get anywhere beyond the fact the module is spitting out a somewhat valid looking signal - you're DONE. It's likely the problem is elsewhere. I'm no T1N expert, but I would guess corroded ground issues given various posts to this forum.

However, before you leave, this is a diff signal and I'm not sure where your ground reference in to the scope is. Might want to probe around a few other bus's that are working to see that the voltages are similar with respect to each other and ground. If not - this might be your grounding mentioned above.

EDIT:
To make my suggestion more clear - probe the bus's, zoom in on the waveform, and pay more attention to the voltage with respect to ground and each other. If you can, with the scope, reference a ground pin on each module - check both sides of the link. Mess with ground connections nearby and see if something changes on the scope.
The "mechanic" method is to identify a module causing interference with the signal on the can bus; unplug them and test, plug back in until the signal "looks" like what normal communication "should"

A "mechanic" or "backyard mechanic" would then simply toss the module and replace it. This method sometimes works, and I've seen it correct faults from the ABS module on other vehicles like BMW which use a similar BOSCH module. The BMW had a Christmas tree effect on the dash, all fixed with replacing the module. Whatever breaks on those screws up communication between the other modules on the can bus

What is broken in the module or why, isn't my concern. Just that it's broken. Yes I'm done with that.

On to the wheel sensors, and other tests as you mentioned. Unfortunately my problem wasn't as simple as an obvious fault that fills the screen and wipes out the "normal looking" waveform (as seen in several videos I've watched, the effect seems fairly obvious)

Hantek - not for professionals. I get it. But oh so cheap...and no "throwing parts" at the van in an effort to fix by trial and error

Saga continues
 

NBB

Well-known member
While you're thinking deeply and screwing around - Google "logic threshold" - as that's actually what I'm suggesting you look for. It's not trial and error.

You're not going to find an actual fault at the bit level - there's no way. If the thing is spitting out as clean a bit pattern as you see - it's not your problem. Even with the best equipment - YOU won't find the problem without incredibly detailed and deep knowledge of the underlying data structure. In addition to a better scope, the big boys are decoding the data as well.

In general, it's a common trap to overlook the basic (a ground) for something WAY more complicated - that's why modules blindly get replaced "trial and error".
 

Top Bottom