Talk:Deadband

Content oversimplified

The content in this article is misleading and not entirely accurate. The examples are oversimplified and incorrect. There is only one source citation for the information here, which is dangerous. Single sources should never be used as gospel about any subject. In addition, the chapter and verse for the source is not detailed. Recommend entire revision of this article.64.162.229.144 (talk) 16:49, 7 July 2010 (UTC)

I will go farther than to say "not entirely accurate." The segment that mentions thermostats is flatly incorrect and it is self-contradictory. I am not good enough in control theory to clearly illuminate this topic with brevity, but I can recognize that this one of the few technical pages in Wikipedia where I have found an outright mistake. — Preceding unsigned comment added by 134.134.137.75 (talk) 17:36, 6 August 2012 (UTC)

As mentioned above, the thermostat section is exactly backwards. Still. Somebody pull out a book and put some real infomation here! — Preceding unsigned comment added by 174.136.133.132 (talk) 17:40, 3 May 2018 (UTC)

Here is the transfer function of a thermostat: a hysteresis loop. The horizontal ("in") axis is temperature, and the vertical ("out") axis is the "heat on/off" control signal. +M is "heat off"; −M is "heat on". +T is the "turn off" trigger point, 22°; −T is the "turn on" point, 18°. To understand the cycle, just follow the arrows around the loop.

I don't see that there is anything wrong with the thermostat section. There are different types of thermostats for different heating systems, but for the traditional type of heating system, especially furnace types with complicated igniting sequences, the thermostat must have some hysteresis [1], [2], [3], [4] (one reason for the confusion is that some consumer websites erroneously call this "deadband") This is to prevent the heating system from being made to cycle too rapidly, which is inefficient and wears out the igniter components.[5] Without hysteresis, the thermostat would turn the furnace on as soon as the temperature dropped a fraction of a degree below setpoint. The furnace would only stay on long enough for the temp to rise above setpoint, maybe 10 seconds. Then it would only stay off 10 seconds until the temp dropped below setpoint again. So the furnace would cycle every 20 seconds, causing excessive wear. With a sensitive thermostat and fewer delays in the control loop, there is no limit to how fast the system would cycle.

To prevent this, the "on" trigger point of thermostats is made several degrees below the "off" point. After the furnace turns off, it takes maybe 10-15 min for the temperature to drop to the "on" point, turning the furnace on again, and another 10 min of heating for the temp to rise to the "off" point. So the furnace only cycles every 20-30 min, which is what furnaces are designed for.

You can see that the type of nonlinearity needed is hysteresis, not deadband, because the output of the thermostat, the binary "heat on/heat off" control signal, is a multivalued function of the input temperature, it depends on more than just temperature. In the example in the article, the "on" point is 18° and the "off" point is 22°. If the temp is 19°, for example, and the furnace is off, you want it to stay off until the temp drops to the "on" point, 18°. However if the furnace is on, you want it to stay on until the temp rises to the "off" point, 22°. This two-valued behavior is the characteristic of hysteresis. --Chetvorno^TALK 05:22, 20 May 2018 (UTC)