Devicenet Speed

Hello all, I don't have much experience using devicenet and was wondering how the I/O update speed compares to hard wired digital I/O. My application would have up to 30 nodes but they are all local with short drop lengths. The nodes are servo drives that need to start and stop at precise time on a high speed rotary turret assembly machine. We have used digital I/O in that past and it worked fine but it was too much wiring.

Thanks.

The maximum 'Bit Rate' for DeviceNet is 500kbps, so the speed depends on the size of the data-packets.

For controlling servo drives in a coordinated fashion, I'd look more to Sercos, ControlNet, or (DEDICATED NETWORK) Ethernet, in that order.

I'm in the process of setting up my first Devicenet project. 1747-SDN scanner with 35 nodes including 12 Ultra3000 servo drives and a SmartGuard600 with 9 safety modules. I had to downgrade to 250kbps because I was getting transmission errors at 500. I think it's safe to say that my network is fully loaded--I'm using all of the discrete I/O in the scanner module plus about half of the M-file area. I'm also using change-of-state messaging for most components so I don't have a fixed I/O update rate but I'd say it's in the neighborhood of 20-30 ms (compared to 5 ms for rack I/O in an empty SLC). The other issue is that accessing the M-file area in the Devicenet scanner adversely affects the scan time of the PLC. So yes, it's definitely slower but your application requirements will dictate whether that is acceptable. If you are using the PLC to send precise start and stop signals to a servo drive, then you may have trouble. In my case, I'm just triggering pre-programmed motion profiles so it's no problem. I had a few cases where the PLC was not able to pick up really short sensor pulses, but I solved that by adding a pulse stretcher to those devices (Turck SmartPlug).

I ran some experiments recently on a system containing a single DeviceNet drop of Beckhoff remote I/O. The PLC was set for 4 ms fixed scan time and the DeviceNet scanner was set for 4 ms poll cycle synchronized with the PLC scan. I wired a pushbutton to two inputs (a hard-wired one and a remote one) using blocking diodes and traced the response.

I found that the lag between a hard-wired input and a DeviceNet input varies between 8 and 16 ms; about 12 ms on average.

So a good rule of thumb for us when deciding should a particular input has to be assigned to a hard-wired or a remote set is about 20 ms. Anything that needs to be detected faster has to be hard-wired.

I also agree that DeviceNet by definition is too slow and non-determenistic to be used as a motion bus.