Is this a recent installation? Have you investigated the origin of the noise? It's best to consider filtering as a final option. "They are both exhibiting identical behavior." Take a closer look at cable routing and grounding practices.
In accordance with guidelines from Temposonics, it is recommended to use low-resistance twisted pair and shielded cables for optimal connection. Ensure the shield is grounded externally through the controller equipment. Keep control and signal leads separate from power cables and away from interference sources such as motor cables and frequency inverters. Utilize connectors with metal housing and connect the shielding to the connector housing. Maintain a large connection surface at both shielding ends and keep non-shielded leads as short as possible. Ground connections should be short with a large cross section to avoid ground loops. To prevent compensating currents, use potential compensating leads with a large cross section or cables with separate double shielding, connecting only one end of the shield. Additionally, it is advised to use stabilized power supplies that meet specified connecting values.
If you own an oscilloscope, it's a good idea to start by checking the power supply. Other devices connected can cause interference on the power supply. Next, try disconnecting the long cable and analyzing the signal at the pins on the tempo sonic rod. If there's no interference, the issue may be with the cable itself. Be sure to meet the necessary requirements for the cable to avoid any problems. In most cases, filtering in software shouldn't be necessary.
- 16-12-2024
- Peter Nachtwey
Tinine inquired if the installed system was new and advised to investigate the source of the noise before resorting to filtering. Both devices are exhibiting similar behavior, so check the cable routing and grounding. Both devices have proper grounding and are on the same card but are different types. These Tempos have their own software, so a call will be made to gather more information. Normally, they have to be sent in for further inspection. I will provide updates once I have more details.
OkiePC shared a valuable tip I picked up from Mickey in a discussion forum some time ago: implementing a first-order filter technique. This involves setting the filter to run at specific intervals using a one-shot method, while incorporating a portion of the difference between the original and filtered values in each cycle. Typically, I apply this method to the mA measurement before converting it to engineering units, although it can be used interchangeably. In the image below, you can see how I integrated this approach into a scaling AOI. Just remember to keep the filter constant (FC) value above zero or equal to 1.0 to ensure optimal performance. If set to 1.0, the filter will be deactivated. Click to see more details.
OkiePC shared a valuable tip about using a first-order filter learned from Mickey on a popular forum. This filter is set to run at regular intervals using a one-shot method, and incorporates a percentage of the difference between the raw and filtered values in each cycle. Typically, this method is applied to the mA value before converting it to engineering units. However, either approach works effectively. I have included a screenshot of this process integrated into a scaling AOI. It is important to ensure that the filter constant (FC) is set between zero and 1.0 to enable proper filtering. If FC is set at 1.0, the filtering function will be disabled. Additionally, I have discovered that the stable North and South Tempo is a SSI. I have attempted to implement this logic, but am unsure if I have done so correctly.
OkiePC shared a valuable tip on this forum years ago about using a first order filter for data processing. This filter, which I often apply to the mA value before converting it to engineering units, involves setting it to run at a fixed interval on a one-shot and incorporating a percentage of the difference between the raw and filtered values in each cycle. I have also included a screenshot of how I integrated this filter into a scaling AOI. It is crucial to ensure that the filter constant (FC) is set to a value greater than zero but less than or equal to 1.0 to enable effective filtering. If the FC is set to 1.0, the filtering function will be disabled. If you are experiencing any issues with this setup, feel free to ask for further clarification.
If you could provide a more detailed explanation, I think there may be an error. To review the issue further, please refer to the PDF document attached in post #3 of this discussion. Click to expand for more details.
If you have set the FC (Filter Constant) to 1.0, it may be disabled. Try adjusting it to 0.1 or 0.05 to see the difference. I suggest creating a trend chart with both the real-time value and the filtered value on a unified scale. This will allow you to observe how the filter affects noise and actual process signal variations.
Is it considered "engineering" by some to simply mask underlying issues? For example, if you were to report a strange knocking noise coming from your car engine, and the solution provided was to just increase the radio volume to drown it out, would that be considered a valid fix in your eyes?
Karl362 inquired about stabilizing the North and South Tempo in a SSI setup and sought clarification on the filtering logic. It is important to determine the tag value that needs stabilization, especially when analyzing code and filter functions. By replacing the tag name in the CPT instruction, the desired filtered value can be achieved. Additionally, addressing any sources of noise, whether it be process noise or measurement noise, is crucial to obtaining accurate results. If noise persists, considering alternative filtering methods such as periodic or moving averages may be more effective. Understanding why the fluctuations are occurring and addressing any issues with perception or control loops is essential for optimizing system performance.
I possess ample experience working with Temposonic rods, which undergo testing by Delta Motion before being made available to the public. Personally, I am not a fan of analog Temposonic or Balluff rods due to various reasons, but excessive noise levels should not be present, as mentioned by OP. I concur with Tinine that filtering merely masks the root cause instead of addressing it properly. The lack of specific details about the noise is concerning – could it be attributed to a 60 Hz hum or other sources like drives? Does the noise subside when other devices are switched off? It is baffling that such questions still need to be raised despite the passage of time and numerous discussions on the topic. To mitigate noise issues, I often recommend providing individual power sources to analog sensors and avoiding connections to switching type loads. Additionally, it is disappointing that even after over 25 years on this platform, no standardized method has been established for specifying coefficients for a low pass filter within a certain bandwidth. Two pole Butterworth filters offer a more efficient noise reduction solution with minimal extra effort. Will this discussion be yet another unresolved thread that lingers on endlessly?
- 20-12-2024
- Peter Nachtwey
Noise issues often stem from inadequate screening and interference from external sources such as drives, which can be particularly problematic when using remote I/O. I once encountered this issue in a system where the screens were connected at both ends of the remote I/O, along with a separate earth cable and bonding to the machine using RIO from the main panel. By simply disconnecting the screen from the RIO, the problem was resolved. It is important to note that simply adjusting the analogue value may not always be the most effective solution to address noise problems.
This post is dedicated solely to closed loop filters. Open-loop filters are deemed uninteresting as they are often viewed as a question of confusion rather than a mathematical one. As Peter Nachtwey pointed out, there seems to be a lack of understanding when it comes to specifying coefficients for a low pass filter to achieve a certain bandwidth after many years of discussion. The question remains: How can one determine the necessary filter bandwidth for a real process with actual noise data, especially for use in a closed loop system?
Nachtwey also highlighted the effectiveness of two pole Butterworth filters in noise reduction with minimal additional effort. The Butterworth filter's superiority in closed loop applications, supported by a video on its calculation, is worth considering. If delving into this topic, it's crucial to analyze the Magnitude and Phase responses of the Butterworth filter for a comprehensive understanding.
Nachtwey posed the intriguing question of whether this discussion will continue endlessly without a resolution. The complexity of the 1st order low pass filter's formula, as mentioned by another user, adds to the depth of the conversation. Let's ponder on the possibility of finding a resolution to this ongoing discourse.
While I'm not recommending an LPF as the ideal solution for the original poster, it is worth considering the CLX's built-in LPF function block as an alternative to using timers and ladder logic. I have successfully implemented this on agitated level control systems.
Before installing a software filter, it is important to address any potential sources of electrical noise. Ensure that the grounding tab at the rear of the Temposonic sensor is properly grounded and that shielded cables are used. Once these steps are taken, then consider implementing a software filter. In my experience, I encountered noise issues with the Temposonic Ethernet/IP version, but after sending it back for a firmware upgrade, all noise problems were effectively resolved.
- 20-12-2024
- APS_Programming
MaxK discussed the importance of closed loop filters over open loop filters in the context of determining filter bandwidth for real processes with noise. The key is to find the right balance between filtering out noise without impacting the motion frequency. For industrial applications with motion frequencies between 5-10 Hz, a filter at around 100 Hz is effective in reducing noise while maintaining motion integrity. It is also advisable to filter the output rather than the feedback in closed loop systems.
When it comes to Butterworth filters, they offer a sharper response at the corner frequency and faster drop-off compared to single or two pole filters. The aim is to limit higher frequencies from passing through, making Butterworth filters ideal for closed loop applications. To better understand the effectiveness of Butterworth filters, it is recommended to analyze the Magnitude and Phase response.
Regarding the resolution in filtering, it is crucial to consider the specific application requirements and utilize appropriate filtering techniques. Additionally, the choice between analog MDT rods and SSI rods can impact the overall resolution and performance of the system. For precise measurements, SSI rods with resolutions as fine as 0.1 microns are preferred over analog MDT rods.
- 20-12-2024
- Peter Nachtwey
According to OP, proper grounding seems to have been addressed, but other potential sources of noise such as power supply and electromagnetic interference (EMR) still need to be considered. APS_Programming shared that after a firmware upgrade for the temposonic Ethernet/IP version, all noise issues were resolved. This could be the solution, although it's possible that the "firmware upgrade" simply added a filter to a noise-prone system. The reason behind the need to reduce erratic movement is still unclear.
Peter Nachtwey wondered if this discussion would become another endless thread with no resolution. It seems likely. Has anyone taken into account the possibility of a mechanical problem? Indeed, it is a common issue that can occur.
In response to message #23, I made a conscious effort to remain composed. However, your expertise once again shone through. I have a question for you - why would you share material that could potentially result in students being disqualified from exams? Furthermore, why are you directing me to content that seems to be on a basic level?
I had a simple request for you to explain the Magnitude and Phase responses, with a particular emphasis on the Phase response. I assumed that a knowledgeable individual, upon analyzing the curves, would naturally inquire about the phase shift, especially at the cutoff frequency, and consider how this shift would impact the behavior of the closed circuit.
After taking a closer look at the curves from your communication, I noticed that the cutoff frequency is at 5. This raises the question of what the phase shift is at this frequency and how it will influence the closed loop system. While you may find this task straightforward, for me, it requires a meticulous approach and thorough investigation into each specific scenario.
MaxK expressed his frustration with the lack of understanding in response to a question asked by another user. He questioned why students would risk being kicked out of exams due to their publishing choices. MaxK also criticized the quality of links being shared and emphasized the importance of demonstrating both Magnitude and Phase responses. He highlighted the significance of considering phase shift at the cutoff frequency, particularly in closed circuits. The discussion touched on the use of filters in Delta Motion RMCs, affirming the preference for Luenberger Observers for motion control and alpha-beta-gamma filters for noise reduction in feed chains. This underscores the need for careful research and consideration in addressing specific problems in the field.
- 20-12-2024
- Peter Nachtwey
In my experience, I encountered a similar issue with a ControlLogix system that had 2 FAST analog cards. Despite using a calibrator to monitor the mA signal and using it as the power source, I noticed that the readings were fluctuating rapidly, jumping from around 1000 to the expected value every half a second. This erratic behavior occurred on all channels, whether only one was in use or all of them on a card. I tried adjusting the card settings and grounding out unused inputs, but to no avail. Even dampening the value resulted in lower than accurate readings. Moving the analog cards to different slots did not solve the issue either. I found that recording the highest value in the past 90 seconds was not ideal as it would overlook any decrease in the signal during that time. As I needed immediate and accurate readings for the PIDE and HMI display, I decided to replace the FAST modules with regular modules. Surprisingly, this simple swap resolved the problem without the need for any further adjustments to settings, grounding, or dampening techniques.
Peter Nachtwey exclaimed "WTF?" It's so true! Your comments are spot on for many of your posts! If you don't find my questions worthy of your responses, then I reserve the right to reciprocate. Stay focused on enjoying yourself.
For those interested or seeking useful insights, I have included a picture illustrating my point. The top chart shows the "Closed-loop" system with a PI-controller feedback, while the lower chart depicts the system with an added filter. The introduction of the filter has drastically altered the system's behavior due to phase shifting effects on the signal. The red curve represents the system's output, while the blue curve represents the filtered value. This shift in values impacts the input to the PI controller, resulting in inefficient control.
Please make sure to respond without filtering out any feedback, as I explicitly mentioned before. Resist the urge to censor.
- 20-12-2024
- Peter Nachtwey
Peter Nachtwey emphasized that using a filter is causing the response you are experiencing, which should be avoided. For more insights on this topic, consider exploring the Control Theory subreddit on Reddit for assistance, as it is more suitable for these inquiries.
- 20-12-2024
- Peter Nachtwey