Minimizing Process Variable noise will give you a much more stable and reliable PID system. Learn how to use three techniques to minimize that noise to give you the best possible PID performance.
Check out this playlist to see all of the videos in this series: https://www.youtube.com/playlist?list=PLPdypWXY_ROpXvgmlUuKDKViOFFW8YnoV
Download resources used in these videos here: https://library.automationdirect.com/weg-vfd-pid-resource-page/
Check out all of our videos at https://www.AutomationDirect.com/Videos
Or go to the Automation Video Cookbook at: https://www.AutomationDirect.com/Cookbook
To subscribe: https://www.youtube.com/user/automationdirect?sub_confirmation=1
**Prices were valid at the time the video was released and are subject to change.
Let’s look at three ways you can improve your Process Variable signal and ultimately your PID system performance. And let’s start with Number 3 - process variable filtering - because it’s the one everybody turns to first when it should be the last your resort. One of the cool things about the WEG drives is you can send a copy of the process variable to an analog output. Let’s do that so we can see the actual filtered process variable that PID uses. You need to tell the analog output what it is doing and what kind of signal you want it to output. You can also modify the output signal with these if you want to. I connected an oscilloscope here to look at the raw un-filtered analog input and here to view the analog output which is showing us the filtered process variable that PID actually uses. I’m using the exact same hardware as we used in all of the videos in this series. You can learn about that in this video. The red trace is the raw analog input and the yellow trace is the filtered process variable. Ignore all of these noise spikes – we’re going to talk about those in the next section. Parameter 235 is the analog input filter time. It’s currently at zero, so I’ll open … and close a valve and we see the yellow filtered process variable tracks the red analog input pretty good. If you look close you can actually see the digital steps where the data was sampled. Let’s change the filter time to 1 second and open and close a valve. Yuck. Now the process variable is just a rough representation of the raw analog input. Just for fun, let’s double the filter time … and open and close a valve. Now the filtered process variable doesn’t look at all like the original raw analog input. OK, lets go the other way and make the filter time half a second ... and open and close the valve. OK, the process variable is looking more like the original analog input. Let’s cut the filter time in half again to a quarter second. And open and close a valve. Now the filtered process variable is looking like a pretty good copy of the original analog input. Let’s enter the .15 second filter time the PID application example in the user manual recommends. And open and close a valve. That looks really good. Here are all the tests we just ran. Looks like a filter time of more than a quarter second or so starts to have pretty serious effect on the process variable which in turn slows down the PID response. If your system doesn’t mind a slower response, then that’s fine. But if you want the best possible PID performance, then use the analog input filter built into the WEG drive with the shortest time you can get away with. Before you do that though, there are two other things you should do first. Number 2. Electrical wiring. Did you notice all the nasty noise spikes on my oscilloscope traces? That’s not coming from the sensor. It’s coming from my really bad wiring job on this quick and dirty test fixture I threw together. The best way to fix that is to do everything the drive user manual tells you to do. You can pause the video to read this list or go look at it in the manual – it’s all very self-explanatory, but very few people seem to take it to heart - my test station included. To get the best possible PID performance - and drive performance in general - make sure you go down this list and check everything off. [3.3.1 on pg 19] I do want to expand on the motor wiring just a bit. AutomationDirect offers cable specifically designed to reduce noise generated by the drive to motor wiring. So, if you are serious about getting rid of noise be sure to use VFD specific cable. There are even versions with the control wiring built in so you get both at the same time! Finally, the Number 1 thing you can do to reduce process variable noise is to design your system mechanically to minimize the noise. For example, did you notice that in all of our videos we were using this sensor at the bottom of this stub? Why did we do that? To reduce noise! It’s easy to imagine water rushing down this pipe smashing into this sensor creating pressure spikes and noise. Where this sensor is far away from the rushing water, so it is going to see a much calmer pressure variations. I connected both of these sensors to my oscilloscope and look at that! The only difference between these two sensors is where they are located in my system! The rule of thumb is, keep your sensor 5 pipe diameters away from any disturbance. So, this is one pipe diameter, and this sensor is this many pipe diameters away from the rushing water. Makes a big difference, doesn’t it? And if you have to put the sensor inline like this, the 5-pipe diameter rule applies here too ... keep this 5 pipe diameters away from any bends or other disturbances in the pipe flow. These sections of pipe are clear just so I can show you that effect in action. I’m going to zoom in on those and backlight them to make it easier to see the turbulence. Both of these are empty right now. When I turn the motor on you will see how the stub gets filled and then you will be able to see the turbulence in each pipe. Here we go. Switch to run mode – You see the stub filling up and now you can see the air bubbles dancing around but only up here where the turbulence is. I have a small valve on the inlet pipe so I can inject air into the system to make this easier to see. I’ll open that valve a little bit and we can see how fast the air bubbles are moving through here and smashing into this sensor. But down here we can see the turbulence is contained in the upper part of the stub pipe. If you look close you can see how slowly the small particulates are moving near the sensor at the bottom of the stub because there isn’t any turbulence. I love that demo because you can actually see what is causing the mechanical noise in the sensors and why the 5-pipe diameter rule works so well. Cool. And this applies to any system. If you are doing temperature, don’t put the sensor right next to the heating element or at least put a baffle between the sensor and the heating element. If you are measuring distance, don’t mount the sensor on a moving, shaking or vibrating platform. If you are measuring fluid level, put the sensor inside a pipe to eliminate surface turbulence. If you are measuring flow, then the same 5-pipe diameter rule applies. The point is, think about where you mount your sensor. It makes a big difference. OK, so that’s it. You now have multiple ways you can reduce process variable noise in your system. And you now know why abusing the analog input filtering built into the drive can be a bad thing. Click here to learn more about WEG Variable Frequency Drives. Click here to subscribe to our YouTube Channel so you will be notified when we publish more videos like this and click here to learn about AutomationDirect’s free award-winning tech support options.