TEMPERATURE CONTROL 



body rises in the 'on' part of the cycle and falls in the 'off' part : this would 

 certainly be true in the impossible condition of the T.S.E. being in contact 

 with the heater; in fact there will always be a time-lag betv/een the two. The 

 thermal capacity of the heater might itself smooth out the intermittent 

 nature of its power supplies, but especially in the most refined systems, where 

 forced circulation assists the distribution of heat in a fluid, it is all too easy for 

 the time lag to be of the same order as half a cycle of the hunting head. 

 When this happens we have a form of positive feedback, and temperature 

 fluctuations can become violent (Figure 29.12b). A further danger is present, 

 particularly in air-bodies having a defined closed circulation (e.g. a plant 

 breeding chamber); in one apparatus built by the author, where the air 

 circulation and the mark-to-space cycle were both of the order of one minute, 

 the cycles actually beat in and out of phase every two hours, giving a regular 

 and large two-hourly temperature cycle. Whenever there are indications of 

 longer term variations in a controlled system it is essential to discover by 

 sensitive recording machines whether this is in fact cyclic in nature, and to 

 suspect a positive feedback system if this is so. 



Relays for mark-to-space systems — As the working of a mark-to-space 

 system relies on the accurate division of a time interval it is clearly essential 

 that there shall not be any variable time factors in the accessory parts of the 

 mechanism. Thermostats of this kind are universally used with relays to 

 switch the heater current, and while mechanical parts of relays will inevitably 

 have a time constant this must not only be much shorter than the time cycle, 

 it must be an absolutely constant delay, or it will modify the controlling 

 interval. Mercury relays are particularly useful for this purpose. Relays 

 of the hot-wire switch type are notoriously unrehable for three reasons: 

 they are not only slow-acting, but have materially different periods for their 

 opening and their closing movements; the variation in the energization to 

 switching interval is very large ; when operating with a frequency greater than 

 4 per minute the interval between energization and switching also depends 

 on the previous switching experience. 



In general, mark-to-space systems will put a greater strain on the relay 

 system than a more simple type of control, and relays must be chosen 

 carefully; the contact device of the thermostat must be guarded against 

 arcing and literal burning, and there is everything to be said for the use of 

 an electronic relay between the thermostat and the large mechanical relay 

 (see Practical Circuits). 



Non-linear temperature-sensitive elements 



We have assumed in the foregoing that the T.S.E. has changed substantially 

 linearly with temperature — that it has in fact been a typical thermometer 

 element. Although they have not found use in very sensitive work, for a 

 limited range of required temperatures there is an advantage in an element 

 with a very great rate of change over only a small part of the temperature 

 scale; thus a mechanical T.S.E., such as a metal rod, would have to be very 

 long indeed to produce a big change in movement per degree, and even a 

 bimetal spiral is still excessively high in thermal capacity to be at all sensitive. 

 Likewise, one could use a Beckmann-type thermometer with inserted con- 

 tacts, but the volume of liquid would be much too great for use with anything 



401 



