How to Ensure FactoryTalk Historian Records All Analog Tag Values Including Zeros

If you set the value of `compdevpercent` to 0.05, then you should adjust `compmin` to 0 and `compmax` to 28800. Click to expand for more details. I was unable to locate these fields in the point builder tool.

• 25-03-2025
• rleon

rleon commented: "I was unable to locate these fields in the point builder. Click to expand... If compdev is set to 0.05, that's acceptable. You have the option to choose either the amplitude or the percentage, with lower values resulting in a 'compressed' effect. It really depends on the data type..."

• 26-03-2025
• dmargineau

To clarify, it appears that the configuration issue lies within TrendX rather than PI. To adjust this, right-click on your TrendX and select the "Pens" tab (if I remember correctly). Each pen has an option for either Analog or Digital settings. When set to Analog, FactoryTalk automatically smooths the line, which does not represent an accurate trend. However, switching to Digital will display the authentic value of the PI point. I personally set all my pens to Digital by default, allowing operators to observe the real-time data instead of relying on a smoothed representation. Additionally, as a troubleshooting method, access the Current Values within PI System Management Tools (SMT). When the motor is not running, if the current values read zero, you can be assured that the system is capturing the correct value as expected.

• 26-03-2025
• Dood

Dood mentioned: "If I'm interpreting this correctly, it appears that the settings are located within TrendX, rather than in the PI configuration. By right-clicking on your TrendX and navigating to the 'Pens' tab, you can find an option to set each pen to either Analog or Digital. When set to Analog, FactoryTalk automatically applies smoothing to the trend line, which means the trend may not accurately represent true data fluctuations. The challenge arises when I am logging trends to a database, such as a dBase or SQL Server. In these cases, there's no need to switch the analog settings to digital mode. For instance, if a PIDE set point changes from 50 to 100 suddenly, the TrendX will display it as a step in SQL Server without requiring the pen configuration to be set to digital. However, if I am working with data sourced from PI, I must utilize the digital setting." This clarified explanation underscores the differences between analog and digital trending in TrendX and its implications for various data logging systems, enhancing SEO through relevant keywords.

• 27-03-2025
• rleon

It seems you need to utilize Excel's Datalink feature to retrieve the relevant data. Datalink is designed to automatically interpolate the "empty" time intervals for you. In essence, you require a client that comprehends how a process database stores information. This is not solely a challenge related to PI; all process databases function similarly by discarding data that remains unchanged. Your alternative option is to configure a maximum compression time, but this will lead to significant data volume on your historical server. Upon reassessment, I realize your compression was indeed disabled, which means it's only filtering based on deviation. I'm curious about the appearance of the raw values and timestamps. Essentially, the trend is performing as expected unless you employ a polynomial curve fitting method, correct? I may be misunderstanding some aspects of the situation. If you're looking to enhance your data analysis skills with tools like Excel, understanding these concepts is crucial for effective data management in process databases.

• 27-03-2025
• harryting

If you find that you don't have many data points to monitor and the value remains stable over time, it may be beneficial to consider disabling compression for those specific points. This approach can enhance your data management strategy without compromising accuracy.

• 27-03-2025
• chopin

Harryting explained: In other words, it’s essential to have a client that comprehends how a process database manages its data storage. This isn’t solely a PI (Process Interface) issue; all process databases function similarly by discarding data that remains unchanged. Your only alternative is to establish a maximum compression time, but this can significantly increase the volume of data on your historian server. Upon reviewing the details, I noticed that your compression feature was disabled, meaning it only filters based on deviations. I’m interested to know what the raw value and timestamp look like. The trend appears to be performing as expected unless you apply a polynomial curve fit, though I may be misunderstanding something. To clarify my requirements, here are some examples: The first image illustrates the current motor transitioning from a stationary state to an operational state, with the X-axis reflecting the time at 16:50:33. The trend indicates the period when the motor was at a standstill (current reading = 0). This is how I can visualize the data using SQL Server in my data logs. The second image presents the same curve generated by PI. Instead of showing a zero line during the motor's inactivity, it interpolates the curve from the initial zero until the current becomes non-zero, failing to log the zero values while the motor is stopped. According to this curve, at 16:26:33, the current reading indicates 40, while in reality, the motor's current is indeed 0, as it has stopped. I need the Historian (PI) to record values in a manner similar to SQL Server. Compression is not a necessity for me, and I am not concerned about disk space limitations since I have 2 TB available. I appreciate all your insights!

• 27-03-2025
• rleon

If adjusting the Step parameter to "ON" does not yield the expected outcome, it can be inferred that the trending package is not properly interacting with the archive to utilize this display hint. It's important to note that PI ProcessBook will indeed utilize this display suggestion. (The Step ON/OFF setting affects only the visual representation in client applications and does not alter the archival data itself.) To achieve optimal results, you can deactivate exception processing by setting the exception deviation to zero. You can either keep data compression turned off or enable it while ensuring the compression deviation is also set to zero. When compression is enabled with a zero compression deviation, this will prevent repetitive zero values from appearing when the motor is off (provided that the interface consistently updates with zero values during each sample while the motor is inactive). When working with PI Datalink in Excel, users can access all archived values through the "Get Compressed Data" feature (which can be somewhat misleading), or by using "Get Sampled Data," where PI interpolates and fills in gaps based on the designated interval. Essentially, the Step parameter informs the trend client whether to access Compressed or Sampled data from the archive. If the trend client does not support Step requests, you may need to configure your stored data to resemble a square wave by setting the exception deviation (execdev) to zero and minimizing the compression deviation (compdev), or alternatively, keeping compression disabled.

• 27-03-2025
• Mispeld

Mispeld mentioned that to attain the desired outcome, you can successfully disable exception processing by adjusting the exception deviation setting to zero. Click here to learn more. Thank you! I’ll give that a try. Much appreciated!

• 27-03-2025
• rleon