CAN FD Eye Diagram Mask Testing using InfiniiVision X Series Oscilloscopes


Keysight’s InfiniiVision X-Series Oscilloscopes
offer a broad range of application specific licensed options to help you debug your automotive
serial buses more efficiently, such as, CAN, CAN FD, LIN, and FlexRay. In this short video, I’ll give the demonstration
of performing an eye-diagram mask test on the CAN FD bus. Eye-diagram mask testing is an important test
for many of today’s higher speed buses such as the new CAN FD bus. So lets get started: here we show an InfiniiVision
X-Series Oscilloscope triggering on and decoding a differential CAN FD bus. This bus has a base rate of 500 kilobits per
second, and an FD rate of 10 megabits per second. Now we can clearly see the FD fields here
where you see the speed really kick up. Now we can zoom in on one particular frame
and see it closer. Or we can even zoom in closer to see just
the empty bits where they begin to take off. Now this right here is the bit rate switch
bit and all the bits following it until you get to the CRC field are the FD bits. In this view of the waveform we can see our
amplitudes look pretty solid – not a lot of noise or ringing. But it’s much more difficult to get information
about the timing: are these edges in the right location? And that’s where a CAN FD eye diagram is really
going to help us out. So let’s go ahead and perform an eye diagram
mask test. First, I’ll insert my USB memory device that
has stored copies of CAN FD mask files that I downloaded from Keysight’s website. Then I’ll recall the appropriate mask file. There are a variety of mask files, some specifically
for dominant bit low probing, some for dominant bit high probing, and various FD baud rates. I’m probing dominant bit low and I have a
10 megabit per second FD rate, so I’m going to select that file. When you recall the file the scope will automatically
set itself up vertically, horizontally and triggering, and overlay all dominant and recessive
bits in what looks like an eye. This is called an eye diagram. Now, CAN FD eye diagram mask testing tests
just the first ten bits in the FD data phase, but it tested from all CAN FD frames from
all nodes. Ten bits covers the worst-case re-synchronization
time which can happen if two consecutive stop bits occur. This will test for worst-case clock stability,
and jitter over ten bit periods. Overlaying just the first ten bits of the
FD data phase of all frames will show waveform characteristics of all FD bits from all nodes
in the system. Let’s now take a closer look at the mask itself. The mask defines a fail zone, which is shaded
in gray here. This is the area on the scope’s display where
signals should not enter for this pass-fail eye test. The mask is based on a six point polygon,
which is common for many of today’s serial bus standards. The upper boundary is set at -.5 V. The lower
boundary at -.9V. Now this is for if you’re probing to view
dominant bit low. If I was probing to view dominant bit high
and I recalled that particular mask file, it would be +.5 to +.9. The left point on the mask is set at 30% of
a bit time and the right point on the mask is set at 80% of the bit time. A bit time in this case, since it’s 10 megabits
per second, is 100 nanoseconds. Now, the sample point on most FD buses is
typically somewhere above 50%. Where my mouse is pointing now is approximately
60%. Sometimes it’s as high as 75%, which is at
the apex of the polygon mask in that corner right there. So that is why the mask is shifted to the
right rather than right in the center of the screen. Let’s now zoom back out and interpret what
we’re seeing here. In this particular CAN FD eye diagram mask
test we can see that one of the system’s nodes generates a recessive bit which is high in
this case, and it rings and it dips down into our mask and violates it. So you can see where it’s highlighted in red
indicating failures. This test reveals that there may be a risk
at this dip in the differential signal from that particular node could be interpreted
in other CAN FD transceivers as a dominant bit. We can also see that this CAN FD system generates
approximately 20-25 nanoseconds of timing uncertainty or jitter. And if we look at the statistics here we have
a failure rate of about 1.7% after testing over 750,000 CAN FD bits. Now watch what happens if we add noise to
the signal, which might be more typical of a real automotive system. You can see the eye is beginning to close,
which means that we have a smaller valid data window, and it even begins with this ringing
and dip. It begins to look like the eye is winking
at us. Now to exit a CAN FD eye diagram mask test,
all you have to do is press “clear mask” and it goes back to the sequential timing mode
that we were looking at before. To learn more about Keysight’s IInfiniiVision
X-Series oscilloscopes and how they might be able to help you test and debug your automotive
designs, contact a Keysight authorized distributor, and ask for a demonstration, thank you.

One comment

Leave a Reply

Your email address will not be published. Required fields are marked *