Bus Load / Peak Bus Load Discrepancy

[battjb3]

Hello,

I'm working on a project with multiple nodes on a common bus sending nearly identical information, and I'm trying to determine what the bus load is.

I've analytically determined that the maximum bus load (considering bit rate, bit time, worst case Frame size, and number of frames within the busiest minor timeframe) should be ~14%.

The discrepancy comes when I compare that analytical result with the observed Bus load shown by PCAN-Explorer (Ver 6). With the setup described, I get nearly 50% maximum bus load.

I'm struggling to reconcile the two results. How is PCAN Explorer determining Bus Load and Maximum Bus Load?

Thanks

[M.Maidhof]

Hi,

which PCAN hardware do you use? All CAN FD hardware modules are able to measure the busload directly in the CAN core. When you use a CAN 2.0 PCAN hardware, PCAN-Explorer 6 calculates the busload by converting the frame back into the physcial bit stream (with stuff bits and error frames). As a test, please enable the error frames in PE6 and make a trace. Use the trace to recalculate the busload depending on the bitrate.

regards

Michael

[battjb3]

Thanks for the quick response,

I am using the PCAN-USB (IPEH-002021) with PCAN Explorer 6. I have captured a trace with the nodes, but the analysis of the data seems to indicate the same result I see analytically. I have Error Frames enabled, but am not seeing any in the viewer or the logged output and the Bus load peak remains approximately the same.

There seems to be a fundamental difference in my analytical method and how PCAN Explorer is calculating the bus load. With a single node (i.e. one node + PCAN Explorer acting as a node for ACK) sending a single 8 bytes of data at 500 kbps every 500 mSec, I've determined the bus load to be approximately 0.053%. However, PCAN Explorer indicates a Bus Load of 0.4%. I understand that this cannot be linearly extrapolated for the whole of a multi-node system as CAN is not deterministic, but I would expect parity for a single message example.

Perhaps I am fundamentally incorrect in my analytical calculations.


Bus Load (%) = ( [Total # messages] * [Bits per Message] * [Bit Time] * 100) / [Total Time]



Thanks

[M.Maidhof]

Hi,

did you also take care for the stuff bits in your "Bits per Message" definition? This is data specific and could be up to 15% more bits per message.

Please see examples for ID 0x555 with all data 0x55 (nearly no stuff bits) and ID 0x000 with data 0x00 (16 stuff bits):

ID555_Data55.JPG (196.97 KiB) Viewed 4709 times

ID000_Data00.JPG (157.56 KiB) Viewed 4709 times

109 to 124 bits per message depending on the ID and data of the message!

regards

Michael

[battjb3]

Thanks for that suggestion. However, I did take into account worst case scenario with regard to stuff bits. For that reason, I would expect the PCAN Explorer Bus Load output to be less than my calculation. Even if I did not, an additional 15% of total bits in a Frame does not explain the 8X increase measured by PCAN explorer.

[M.Maidhof]

Hi,

please trace the data and send it to us by email, to be able to check the busload of your system.

regards

Michael