Curious if anyone here has done a comparison between using Kepware OPC server vs native WinCC OA driver for comms to Allen Bradley PLCs? the old system we are upgrading currently uses Kepware for PLC comms and we are contemplating whether to keep using it or switch over to WinCC OA native drivers. There's roughly 10k IO points in the system and if we can easily handle that with the native driver I wouldn't mind ditching Kepware (not because I don't like, but because it's another layer which might not be required).
Regards,
Tom
Kepware OPC vs native WinCC OA Allen Bradley driver
- tpjctrl
- Posts:145
- Joined: Tue May 08, 2018 10:30 am
Kepware OPC vs native WinCC OA Allen Bradley driver
- mcmullan
- Posts:16
- Joined: Mon Jan 31, 2011 10:18 am
Re: Kepware OPC vs native WinCC OA Allen Bradley driver
You haven't quited provide enough information here.
We can handle just about any number of IO points. The question is, what’s the polling-rate.
Also, it matters hugely if these are all separate named tags, or if a significant part of these are in arrays.
In 3.16, the throughput compares well to Kepware, for named tags, especially if you use the new multiple-connection setting.
For arrays, it can out-perform Kepware by a lot.
Using a ControlLogix 1756-L72 + 1756-EN2T our throughput spec for named tags was over 9100 values per second. For array tags, it was over 100,000 values per second.
In some cases the throughput was significantly higher than this.
There are a lot of factors though, including the size of the symbol-table in the PLC, and the length of the tag-names.
We can handle just about any number of IO points. The question is, what’s the polling-rate.
Also, it matters hugely if these are all separate named tags, or if a significant part of these are in arrays.
In 3.16, the throughput compares well to Kepware, for named tags, especially if you use the new multiple-connection setting.
For arrays, it can out-perform Kepware by a lot.
Using a ControlLogix 1756-L72 + 1756-EN2T our throughput spec for named tags was over 9100 values per second. For array tags, it was over 100,000 values per second.
In some cases the throughput was significantly higher than this.
There are a lot of factors though, including the size of the symbol-table in the PLC, and the length of the tag-names.
- tpjctrl
- Posts:145
- Joined: Tue May 08, 2018 10:30 am
Re: Kepware OPC vs native WinCC OA Allen Bradley driver
Huge thanks for taking the time to reply Rick!
I assumed the native drive would be able to cope with any number of tags if you give it enough time, but I agree, I should've specified the poll rate as otherwise the conversation is a bit meaningless. We need to pull new data every 0.5s, so the driver needs to be able to pull around 10k tags from the PLC in 500ms. This would be mostly UDTs with a fair amount of arrays inside them, so fingers crossed it's going to be able to cope. We'll have to set up some tests to see if we can hit that poll time requirement, but I just wanted to do some early reasearch to see if the native drive is a good thing or something that works, but only if you don't pull more than a 100 tags at a time;)
Thanks again,
Tom
I assumed the native drive would be able to cope with any number of tags if you give it enough time, but I agree, I should've specified the poll rate as otherwise the conversation is a bit meaningless. We need to pull new data every 0.5s, so the driver needs to be able to pull around 10k tags from the PLC in 500ms. This would be mostly UDTs with a fair amount of arrays inside them, so fingers crossed it's going to be able to cope. We'll have to set up some tests to see if we can hit that poll time requirement, but I just wanted to do some early reasearch to see if the native drive is a good thing or something that works, but only if you don't pull more than a 100 tags at a time;)
Thanks again,
Tom