1
\$\begingroup\$

I have a scope connected to my canH line and ground and noticed there is some distortion on the waveform in my setup.

waveform

Setup:

Setup details

I have spliced the CAN cable to connect the PCAN-USB device via the breakout box. I measure 60 ohms across the canL and canH pins (pins 2 and 7 of the breakout, respectively).

The PCAN device has two pins marked as GND (3 and 7). I connected these to the power supply ground...is that OK? The PCAN device itself is connected via USB to my laptop so I wasn't sure where to connect this.

Any idea what can cause the distortion ? Do I need to tie the cable shielding to my power supply ground as well?

\$\endgroup\$
2
  • \$\begingroup\$ Likely a problem with the measurement itself and poor grounding of the scope probe. \$\endgroup\$
    – Lundin
    Aug 26 '20 at 7:59
  • \$\begingroup\$ Not necessarily poor grounding or measurement. The CANH is just open-drain signal so it will rise sharply but fall slowly. \$\endgroup\$
    – Justme
    Aug 26 '20 at 20:20
1
\$\begingroup\$

Your signal doesn't look distorted, and is typical of what you might expect with an ad-hoc connection of an oscilloscope probe to a digital signal.

You can probably improve on the noise with a better earth connection, if your scope has a second BNC type input then solidly ground that to system ground using short thick cable (braid is good),

Also try to minimise the area of the loop that is formed between the probe tip and the earth wire on the probe.

\$\endgroup\$
2
  • \$\begingroup\$ Actually this is not the reason it looks like that. CAN has dominant and recessive states, dominant has low impedance and recessive has high impedance. The CANH is either driven high, or left floating, so that only the 60 ohm termination is responsible for pulling CANH to CANL level. The user Maple has a correct answer. \$\endgroup\$
    – Justme
    Aug 26 '20 at 20:18
  • \$\begingroup\$ @Justme I was really addressing the poor fidelity of the signal, essentially the "noise", The rounded trailing edge is due to cable capacitance and higher impedance and measuring only half of a differential signal, it's really part of the "signal". I've upvoted Maple's answer BTW. \$\endgroup\$
    – BobT
    Aug 28 '20 at 2:39
2
\$\begingroup\$

I've seen something similar before in a setup of mine - it turned out to be a measurement error. Since CAN-Communication is differential you should use a differential probe to measure your system precisely (one probe to CAN-L and the other one to CAN-H) By using a single-ended probe your reference changes and seemingly introduces errors in the waveform. I wouldn't worry much about the shielding of the breakout-box; your impedance seems alright and your cable length is not very long.

\$\endgroup\$
2
  • \$\begingroup\$ Well, it is sometimes desirable to play around with slope control of the transceiver, in order to reduce noise. You have to measure the actual CANH and CANL then. \$\endgroup\$
    – Lundin
    Aug 26 '20 at 8:01
  • \$\begingroup\$ @lundin he does seem to have some overshoot on the edges, so a slower risetime would clean up (or he has some stray inductance on his scope probes) \$\endgroup\$
    – BobT
    Aug 28 '20 at 2:47
2
\$\begingroup\$

I believe the main point missing from other answers is that what you call a "distortion" is perfectly normal. While CAN lines are actively pulled apart for dominant bits, they passively pulled back together by resistors fighting line inductance. For this reason the slope of a trailing edge is always different than the one of the leading edge. The rest is most likely the measurement noise as @BobT pointed out.

\$\endgroup\$
1
\$\begingroup\$

Just viewed some old messurements I made with CAN... looks normal to me in first place. some little overshot is not a problem as the sampling point of can is variable (normaly) and in the middle or back section of a bit.

Can you make a new (better) osciloscope shot with the signal scaled to full range and post the aquisition setup? like bandwidth of the scope/probe, sampling rate, etc...

your display looks familar to me like the old tektronics desktop scopes...

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.