Control Logix - RTD cards - high resolution

Just wondering if anyone has come across any RTD cards for Control Logix with higher than 16 bit resolution.

Just curious about why 16 bit resolution isn't enough. Thats 1 part out of 65,000 or +/- 0.0015%. I would be very surprised if the RTDs are that accurate!

We are investigating using an Control Logix solution for an application that was previously done by a Bailey DCS. We will be reading thermocouples and RTD's on a Blast Furnace to calculate heat loss. The cards which are currently used are of a higher precision than 16 bit. Our Instrumentation personnel are worried that 16 bit cards may not give them the accuracy they need.

Spectrum Controls is building a Universal Analog module for ControlLogix that claims 0.1% accuracy across the measurement range, which is comparable on it's face to the regular 1756-IR6I module.

Give Spectrum a ring and ask them about the resolution of the A/D converter they are using.

I did a little browsing for DeviceNet I/O modules (the easiest way I know to get 3rd party stuff into ControlLogix) and didn't find anything that jumped out as more accurate than 16 bits. I doubt your existing I/O platform has a DeviceNet adapter, but it's a thought.

Thanks for the reply Ken. I will check with Spectrum controls. This is also for a new instalation and we will be using DeviceNet to communicate to our Motor Control Centers, so DeviceNet may be an option.

Kim, explain to the technicians that just because you can do something, it doesn't mean you should. You might also make references relative to the futility of sow's ear/silk purse transformations.

A typical specification for an RTD is ±0.5%. For extra money you can get ±0.1% accuracy. That is roughly equivalent to 10 bit resolution (although nobody makes a 10 bit A/D that I know of). This means that whether you are using a 128-bit or a 12-bit I/O card the actual accuracy of the measurement will be the same, no matter how many decimal places you display. Furthermore, I have trouble imagining an application where the difference between 355.000 °F and 355.355 °F is incredibly significant. If you have such an application, you should start at the sensor to get better accuracy, not at the I/O card. After you have eliminated or minimized all the other sources of error, then it is time to see if ±0.0015% is really significant or not relative to the other errors.

One other point is the difference betwen resolution and accuracy. For example, I have used A/D cards from one manufacturer that is 16 bit resolution, but the actual accuracy of the measurement is only ±1%.

Imagine this....

Tom Jenkins said:

Furthermore, I have trouble imagining an application where the difference between 355.000 °F and 355.355 °F is incredibly significant.

Click to expand...

A change of .1% can be very signficant if the change is over a short period of time. The differentiator part of a PID works much better with higher resolutions as will any other application that is measuring the rate of change.

If the RTD is very good, and accuracy is ±0.1%, then when you take a reading that the PLC tells you is 355°F, in reality the "true" temperature is somewhere between 354.645 and 355.355. That is as precise as you can be. If you display or show that the reading is 354.99999999, all you really STILL know is that the temperature is somewhere between 354.645 and 355.355 because that is the limit of accuracy of the RTD. In that case, what does the actual resolution increase gain for you in terms of the physical accuracy? If your RTD is measuring off to the low side, and your temperature is controlled to 354.64511111111 instead of 354.645, have you increased the process efficiency or improved the flavor of the pickles you are cooking?

Us old coots what used to use slide rules understand that there are assumptions and approximations built into every process or measurement. If you calculate a guess to 13 decimal places, it is still a guess!

Thanks for all the replys. It has certainly got me thinking. It sounds like 16 bit resolution will be plenty good enough considering that a high resolution RTD only has an accuracy of +/1 0.1 %.