Hantek 1008C CAN L H Waveform

[kr]

Hi, I'm new here so I should introduce myself a bit. I'm from Poland and I'm beginner in automotive diagnosis, troubleshooting etc.
I bought a cheap Hantek 1008C oscilloscope some time ago, but yesterday i was trying to lookup how does my car CAN waveform is looking on this harware.
It was measured at ECU(DDE)
There are no error codes related to CAN BUS so it should look normal, but it's not..
Is this oscilloscope that crap or I am doing something wrong?

[Chriss]

Hi!
I already checked with my Hantek 6022BE scope the CAN bus.
I got the same crappy waveform as yours.
I think it is because the low cost scope.

Can somebody else make a measurement with a Rigol or so?

Thanks in advance.

[Rerouter]

what ground are you using, it makes a difference with automotive stuff, one is body ground and one is canbus sheild,

I cannot speak for the scope in itself, but canbus in cars is only 250Khz, so if it has a sample rate higher than 1MS/s it should be able to capture both without any issues.

[Chriss]

I made the measurement on the bench.
Measured the CAN H and L from a ecu flashing device.
On my measurement the GND is the power supply GND.
But however, Hantek is not a precise scope.
I don't like mine.

[Rerouter]

It has a 2.4MS/s sample rate, if you had all channels turned on, then you would be below nyqist, if it shares the bandwidth between only the active channels and you where only using 2, then it should be more than sufficient.

I still suspect your setup, and not the tool, your needing to extract about 0.5MHz of bandwidth to properly display your signal, so having the ground return loop off all that way i can see as being the issue.

Instead try a local ground, close to where your measuring,

[Chriss]

Here is the pattern what I got out from my CAN bus with the Hantek 6022BE.
I assume it is a usable wave pattern even if I don't like my scope.

My scope setup:
Time/DIV: 20uS
Voltage/DIV: 1V
The measured frequenci of the signal is 166KHz

I have a question, maybe this is for a new thread but maybe not.
If I would check with this method the CAN H and L and I got this type of signal.
I mean the CAN H and L are mirrored same signal, and all the time when I repeat
the signal I get the same pattern.
Could I be sure does teh CAN bus of my device is working correct?
Which in this case is a car diagnostic tool on the bench.

[Rerouter]

That looks like a perfectly valid capture to me, there is a slight difference in the recessive state when the tranceiver is active but that is likely to be the termination resistance being a bit high,

the frequency would be 250Khz, you work off the shortest state time, to me you have measured 011 or 2/3rd the actual frequency.

[Chriss]

Actually I don't have any nodes on the CAN bus.
And I also don't have the 120Ohm termination resistor.
I connected the CH1 to CAN H and CH2 to CAN L direct to the obd II
connector which is on the diagnostic tool.
Than I tried to make an ECU ID reading and captured the signal.
Of course on the end I got also a wakeup error message on my diag tool
because there is no ECU connected but I got a valid signal.

I have another tool for programming the ECU's which has some trouble
when I r/w through CAN bus and I'm searching for some idea how to check
the CAN line on that tool but without the ECU. I won't kill an ECU just to try the think...

If I understand this I can make the measurement also like I did this and check for
the validity of the signal.
A similar pattern should show up on that "bad" tool also if the CAN bus is ok.
Am I correct?
I know, the pattern can be different because the data what is transmitting but the
signal can be recognized as a valid CAN signal. Yes?

Actually I don't have an ECU simulator or so. That is too expensive for me...

Thanks in advance.

[Rerouter]

The tool will likely have a built in 120 ohm resistor, as its expected that what is plugged in to the OBD plug will,

However without the matching one on the ECU or Gateway, its too high, so the voltage shifts a little, having another 120 ohm resistor hooked across the lines should bring it in to spec.

As for your tool causing issues when reading or writing, That can be a number of things, reading should be safe with most tools, however writing may require having a battery charger hooked on to the car and a healthy float voltage, at-least for a number of European brands,

On a hardware level it appears your tool is broadcasting a nice clean signal, and an OBD socket ideally should be a direct connection to an ECU or Gateway, Its possible someone else has hooked in to it, e.g. a toy car logger that has a termination permernantly fitted, Almost every auto tool i have come across has a built in fixed termination of 120 ohms, so if there is another device hooked in with its own 120 ohms, it can pull things out of spec, equally there are a few tools that don't broadcast the silence command before an operation and a lot of Can connected toy products that don't respond accordingly to it.

In general most ECU's have the termination hard wired, so with the ignition off, if you measure anything less than 120 ohms between the can wires with the tool unplugged you may have trouble.

[Chriss]

Thanks for that explanation, I appreciate that really.
I measured another tool right now what I 100% know it is good and working tool and 100% safe would I say.

Here is the ECU wake up signal pattern.
That looks much better than the tool before.
Actually that tool before ( what I posted before ) is the problematic tool.

If I compare that two patterns I can recognize a the wake up signal but the
shape of the signal in my earlier pictures are somehow tooth shapes, in this
good working tool I have a nice and almost clean square wave form.

These differences, could that make that first tool unstable until a CAN communication?

[Rerouter]

The rounding you see is caused by a limited drive strength on the drivers, likely some inline resistors to protect the device against you connecting one to ignition and the other to ground. In my own devices i have used an photofet to terminate the canbus and used a window comparitor so that if the voltage across the resistor was higher than 7V disconnected the termination, while still allowing the system to switch in or out the termination depending on the dropped packet rate.

Try placing a 120 ohm resistor across the wires of your suspicious tool, It should greatly remove the RC time constant effect you see while slightly reducing the margins of your tool. currently they are both driving about the same amplitude but its not quite centered on the reccesive states which is a little weird,

[Chriss]

I already checked the termination resistor on my suspicious device and it is a 60Ohm resistor.
In the good tool it is a 120Ohm resistor.

How much can this affect the communication?
I can remember on the time when I set up Nove Netware with coax cables and there was the BNC terminators.
Some time we had some unexpected issue on the net because a bad BNC terminator or a faulty one.

Is the situation also here the same?

Im not shure does I understand the problem what you mentioned under
"its not quite centered on the reccesive states which is a little weird".

Can you pls give me a link about that so I can see on a pattern maybe?

Thanks.

[Rerouter]

Your suss tool is not built to spec then, it means the vehicle may not be able to reply correctly because its driving something that demands 150% the current to meet the same thresholds,

The termination resistance pulls both lines back to the recessive state which should be equal in the middle,

What i mean about the pattern being offset can be seen by comparing the green line between the 2 devices, on the suss one when it turns on its tranciever the recessive state voltage shifts down by a few hundred mV, which implies something weird is going on about it.

There are a few ways these devices are terminated, personally i am a big fan of the split termination stlye using a chip that outputs a buffered 2.5V supply that both wires are then terminated to with 60 ohm resistors, there are others that literally plonk a 120 ohm resistor across the 2, and it works, but long wiring runs can lead to common mode range issues (e.g. bus engine to instrument cluster), then there are weird ones like theravin termination, I really havent seen many of them since the J1708 days,

The termination equally plays a role of soaking up any ringing or reflections on the data pair, in general not having termination doesn't cause noticeable packet loss until you get to more than 2m of wiring near a noise source, but some trancievers drive way to hard and the signal while being decoded correctly looks like a mess on a scope. Its meant to be at the start and the end of the bus and not intended to be at each device like J1708, however for OBD2 sockets as the tool is the end of the bus they tend to fit it.

[Chriss]

I  think now it is a much better signal from then before.

I changed the CAN bus transceiver ic from 82C250 to TJA1050.

What do you think about this result?

[Rerouter]

now it looks better, did you fix the termination resistor value?

[Chriss]

Hi!
Sorry because I'm late with the answer.
No, I didn't put any extra resistor.
I changed just the CAN bus transceiver ic from 82C250 to TJA1050
and the signal was pretty much better.
I can't explain why.

I checked my OBD cable and I recognizet the H & L wires are terminated with a resistor R=120 Ohm.
But in the tool inside near the TJA1050 is also a resistor R=120 Ohm connected through a relay
between the H & L line.

When the relay is in a not working condition then the resistance between H & L is 60 Ohm.
When the relay is in working state then the resistance between H & L is 120 Ohm.

Here is a schematic.
Sorry for my bad drawing skill. 

So, my question is:
Why is the resistance of the CAN bus in some situation 60 Ohm but in another situation 120 Ohm?