Controllers
and Controlling Electric Power

   The purpose of a controller is to regulate the temperature of a heating device either at a fixed point or as a controlled ramping change from one temperature to another over minutes or hours. Most glass studios have several controllers for the heat supplied to annealers, color kilns, and furnaces and these may be combined in one electronic unit or be separated. A controller may include safety features required by either the electrical code (NEC) or one of several fire codes and thus cost more.
  Most controllers these days use solid state electronics in both the box that sets the temperature and in the device that actually throttles the electricity, which are connected by wires.  With solid state control, small amounts of electrons injected into the device permit large amounts of power to flow and this can be very finely tuned. Solid state boxes allow programming ramps up and down in sequence for special tasks.
   Alternatives to solid state controllers include carbon rheostats which add resistance to the circuit, variable transformers which change the output voltage and saturable reactors which interfere with current flow (?).  The first wastes power and generates a lot of heat while the second is limited in power and is costly in larger sizes and both would need a motor drive to control remotely.
   The pieces usually found in a system include the controller, a thermocouple to measure temperature, a heat source (gas or electric elements), and the interface to take the weak signals from the controller and apply them to the heat source. Schematic  It is my philosophy to match the capacity of the interface to the demands of the heat source and locate it near the source and to locate the control function in a convenient location.  This means short heavy wires carry power to the heat source and longer light weight wires carry the thermocouple signal and control signal from the heated item to the controller.  The most common alternatives to this are to put all the functions in one box located near the heated item - done with ceramic kilns - and to mount several heavy duty interfaces near the main power supply for the building with controllers grouped elsewhere - this being especially done with mercury contactors requiring rigid positioning.  Controllers and SSR's are sensitive to overheating and must be located with that in mind. 2005-12-16
   Controllers are now made with advanced features in a small box for under $150, or with a single display and key pad for 5 devices, or with computer control with history and plan on the computer screen.  Either electric heating or gas burning can be controlled, the latter involving considerably more complicated and costly safety features. Gas Control & Ignition
  Most of the discussion on this page is related to heating, annealing, sagging, and fusing glass.  Melting glass for blowing using controllers is beyond my experience and requires special elements and special power supplies.

Among features that may be important for particular purposes are

Choice of action at the end of the ramp - shut down or hold
Choice of what happens if temp is not reached at end of ramp, wait until reached or quit at achieved temp
Multiple choices of sequences of ramps (programs) stored  to avoid reentry
Stitching program sequences  to make longer sequences
Controlling a sequence like ramp over 3 hours to 900F, hold 30 min. to soak, rise quickly to 1300F, soak 10 minutes, cool quickly to 900F, soak 1 hr, ramp down over 8 hours to 650F, shut down.

Choices (and history) in Temperature Controllers for Glass By Mike Firth

Rev. 3/13/97, 2/18/98, 2000-8-20

Controlling temperature for melting glass and annealing it has used a number of devices historically and has seen great improvement in the last few years. Electric melt and annealing control is relatively easy, but limited in quantity and maximum temperature; while adding control to gas furnaces jumps the cost. One ongoing problem is the need to accurately control the fall of temperature in the annealer over many hours.

Temperature control can take two forms, open loop and closed loop. In an open loop, a device controls the heater, thus raising or lowering the temperature, but the temperature does not affect the device. In a closed loop, the measurement of the temperature modifies the behavior of the device in controlling the heat. A resistance control, dimmer or Variac (variable transformer) is open loop; modern controllers are closed loop.

To quickly summarize the following, open loop devices were first used, then as microcomputer technology began in the 70's, fairly expensive controllers that used closed loops were available. Then the cost started coming down and a series of multi-step keypad programmed controllers, such as Digitry, came on the market. Most recently, controllers that "learn" the characteristics of the furnace or annealer and adapt for better control have come on the market. Some of these have computer ports so that control can be maintained on a computer screen. And now we are seeing multiple ramping units at relatively low cost (under $200.)

In glass working, the chief open loop devices have been the food cooking oven controller, variations on dimmers, and the variable transformer (Variac, a brand name.)  In each case, if a roughly constant temperature is wanted, the human adjusts a knob until it is about right. Most retail dimmers do not have anything like enough capacity to control an annealer; but a dimmer can control a heavier triac or it is not difficult to build a 25 amp dimmer (or bigger.)*  Oven controls used are variable resistors, like on home ovens, which are easily available and have the capacity but waste energy and are not very sensitive.

The Variac has a venerable history in modern glass blowing because at the time the modern art movement started, there were a lot of large capacity Variacs available as military surplus and triacs and electronic dimmers did not yet exist. A Variac is two coils of wire, looking somewhat like a motor with a knob on the shaft, which is adjusted to change the voltage from 0 to maximum available. They are efficient. They are also heavy and if not available surplus they are also expensive and have a limited capacity. Variacs are important because early in the history of modern glassblowing, they allowed annealing to take place over many hours when hand control would be impossible because they were built into a open loop automatic control as follows.

The Variac was mounted on a board with a 24 hour timer whose purpose was to provide a very slow motor. On the shafts of the timer and transformer were mounted disks or segments around which string was run from one to the other. As the timer dial turned, the Variac dial was dragged along, slowly lowering the voltage. Some people used cam shaped disks, so the Variac moved more slowly during the first hours and faster during the later ones. Some of these are still in use, I saw one at John Littleton and Kate Vogel's studio in 1995. As the voltage fell, the power fell as the square of the change and the temperature fell accordingly. But the temperature curve was only monitored by a human and did not affect the timing of the process. (Image shows Harvey Littleton's Variac controller from his book.)

Recently, exactly the same thing can and has been done for a few dollars and it is one better than Variac. Peet Robison in New Mexico built a controller in which comparator chip compared the amplified voltage from a thermocouple to the output from an digital to analog converter (DAC) turning on a triac controlling the heating element when the thermo voltage was lower than the DAC. The DAC setting was controlled by a small counter that changed every few minutes. The total parts cost is about $8 for building this. Note that it is closed loop, because the temperature reading influences the comparator, but what is being controlled is some arbitrary number not a precise temperature. Also the temperature curve is straight line, which is an improvement, but not the best. Today, for $12.50, a chip is available which corrects for oddities of a K-type thermocouple, so the output is (within a small percentage) a voltage the same as the temperature (6.55 volts = 655.C) and thus finer control is possible.

DIGITRY - The most widely used controller in glass melting and annealing is probably the Digitry GB4 [lower in image] which has now been succeeded by the GB5 in 2005.  These control up to 5 units which costs $1465 for the basic unit plus about $55 per kiln or annealer for thermocouple and additional solid state relay or contactor. The GB1 controls one device and costs about $550 + kiln add-ons. The Digitry is convenient to use, providing specific keys on a keypad to select the device and numeric keys to enter device numbers, temperatures and hours. It allows multi-step programs, so an annealer can be set to hold a temperature for 2 hours, then take an hour to drop 50 degrees, then take an hour to drop 150 degrees, and so forth. The cost per unit controlled is not as high as a dedicated commercial controller (which may run $1500-2000), but is still irritatingly high, especially if less than 5 units are controlled. And there is the problem that if a single controller goes down, the shop is down. Similar units are made by other companies, including Paragon
Exact features of the GB5 for comparison include: Scanning display through 5 devices, large LED display, 15 set points per device, programming to 500 hours per step (as in annealing castings), 4 extra profiles can be stored, retains profiles during power failure, set delayed schedule for off peak rates, audio alarms, skip steps if needed (in fusing usually), optional power sharing between units, PC connectivity, and 4-20 ma control.

Two new technologies are available for control. One is a self contained unit that will provide a single or multiple ramp and learn the characteristics to the controlled unit. [These can look like the two upper units in the image.] The other is using a computer to control the controller. Actually, the first units can be had with an option that allows computer control of changing the ramp and then letting the unit run.

Several companies have come to market with small (2" x 2" x 4") controllers for about $200 that avoid a complication that has plagued controllers for years. PID (Proportional-Integral-Derivative) is a method that allows programming controllers so they do not overshoot their goal or over-control, wasting energy. Unfortunately, PID requires careful selection of parameters to match the furnace/annealer to the heat source. This is eliminated in the new units because they "learn" the characteristics of the equipment, adapting to different building temperatures, etc. I have an annealer with a low wattage light bulb across the elements that goes on and off with the elements. It is easy and fun to watch as the controller turns on the element and, after the temperature has risen, turns it off to see how much coasting goes on. After testing, the unit finally runs the temperature right up to the setting.

A $300 solution involves buying a CN76000 Auto-Tune Controller (dc pulse output with alarm option, CN76120, $195 or adding remote setpoint option, CN76120-SP, $234, page P-95) from Omega (1-888-TC- OMEGA, ask for the Temperature Handbook, www.tcomega.com ), along with a K-type thermocouple, DH-1-8-K-12, $19, (page A-13), 25 feet of K type thermocouple wire, PR-K-24, $15, male and female K-type miniature plug connectors (SMP-K-MF $4, page G-16) and a Solid State Relay SSR240DC25 (DC control voltage, 25 amps, up to 280 Volt) $26 + Heat Sink FHS-2, $17. There are only 3 pairs of wires to hook up - AC power, thermocouple, output. [Controller Comparison]

AD595 - There is a set of Analog Devices chips that will take a K-type thermocouple and produce a 10 mv reading per degree C or produce a setpoint controller. With a fairly simple switch, a single chip can do either. I paid $12.50 a chip (buying 2 to meet $25 minimum from Newark Electronics, current suggested retail for quantity 100 is $7.28. Data Sheet PDF on 2012-10-18

The two chips shown above are designed to have a K-type connected to the pins at one end (pins 1 and 14/8). For the reasons given in the discussion of thermocouple connections, AD recommends making heavy copper connections to these two pins showing wide foil connections on their PCB layout. I used a wire wrap socket and bent the pins over to make the solder connection rather than use smaller wire. All this is to keep the connections as close to the chip temperature (which is measured by ICE POINT COMP) as possible. I solder a short length of thermocouple connection wire to the PCB and put a mini-connector on the end of that.

Circuit to be tested and posted soon. (he said on 9/20/2000)

This is the circuit for using the AD595. As shown in the diagram, a DPDT switch allows choosing between measuring temperature and checking the set point. A digital volt meter connected to TP (Test Point) and ground (GND) will show 0.001 volts per degree Centigrade (1 mv/C) I use a multi-turn potentiometer. A limitation of the chip is that it will only drive 5 ma. There are opto-isolators that will work with that drive level, but only just barely. So gates are added and the easiest to get are inverting. Although hard to see, the three gates are Schmidt Trigger inverting gates (3 of 6 in a 74H14) because the signal changes slowly and CMOS usually doesn't like slowly changing signals. [2002-10-26 I had kind of let this circuit lay around for a while, but got bugged and today realized I might need a pull-up resistor.  When I began looking for specs, I was dismayed to find that the 74HCT14 from Radio Shack are very limited on voltage range because they are to work with TTL. So my whole reason for choosing CMOS - voltage variability - is voided because I didn't know/look at the specs.  Now to find one cheap/quick.] 

2002-11-04 I got the thing working, after redrawing the schematic from the bottom and tracing all the wires. There was not a good connection at the measuring point.  The core problem was that the measuring points were not connected reliably.  I tore off the wires and removed the Molex.  Finding nothing smaller, I used two terminals of a European style terminal strip.  Once firmly connected, it seems to work well at light bulb temps.  Will test with one of my boxes when it isn't raining and I am awake.

Power control - Relays

The three basic ways of controlling a bunch of power (30-100 amps, 1 or 3 phase power) are a contactor, a MDR or an SSR. All of these are relays and a relay is a device for switching on and off power in one circuit with another circuit, which may be of a different voltage, for convenience.

A mechanical relay consists of a coil (a winding of copper wire) that creates a magnetic field in a iron core that attracts the iron plate that carries the contacts that switch the power. While relays may take dozens of forms, contactors are a special form that has only normally open contacts that are pulled shut with the magnetic field. The disadvantages of contactors are that they are normally noisy, clanking with each open/shut cycle, may produce electrical noise from the arc, and they cannot be switched more often than every couple of seconds. A major advantage of contactors is that power is completely off when they are open, no leakage, so they are used to kill circuits ahead of SSR's, etc.. A contact may throw an arc on closing or opening. Contactor

"Mercury Displacement Relay (MDR) - An electromechanical switching device having one or more poles that contain metallic mercury and a plunger which, when energized by a magnetic field, moves into a pool of mercury, displacing the mercury sufficiently to create a closed electrical circuit." For this definition and safety steps go to Disposal of Mercury Displacement Relays - Watlow Electric Manufacturing Company

The main advantage of an MDR over a contactor is silence although it can also be cycled faster than a contactor. A big disadvantage is disposal and risk if mercury escapes, which is unlikely but critical because mercury vapor is poisonous and the spill must be treated as hazardous.  A nuisance is that the MDR must be mounted in a specific position to work, which may complicate a portable rig (which must then not be operated on its back, only upright.)

SSR is a solid state relay, discussed more below. The advantage of an SSR is that is can be cycled very rapidly (60-120 times per second if needed). The disadvantages are that a small amount of current leaks through the SSR to the element, even when it is off, and when it fails, it is more likely to fail on (short) than off (open.) Also, an SSR, SCR, or Triac must be provided with a serious heat sink and good air flow. It should not be enclosed inside a box.

From: <JWalsh2000@aol.com>
Subject: Your SSR and Control web page
Date: Monday, October 15, 2001 1:18 PM
A few additional items:
You can use a mechanical, mercury or solid state relay (SSR) to activate the heater in your kiln. A mechanical relay has a recommended cycle time (on time plus off time) of more than 60 seconds and usually about 1 million mechanical cycles. A mercury contactor has a recommendation of 10 second cycle time and 3-8 million cycles (a couple of years). A SSR has a cycle time of less than one second and a virtually unlimited life. The long cycle times required in the mechanical or mercury devices means that your heater will experience thermal stress and will fail sooner than when used with a SSR product. That failure is frequently around the heater's connection due to the thermal expansion and contraction that occurs when the heater in "on" for many seconds and then "off" for many seconds. The shorter that you can keep these "on" bursts and the "off" burst, the longer the heater will last.

Therefore a SSR is an ideal device to control electric heaters. The biggest concern with SSRs is that they generate a little heat that will destroy the SSR if you do not remove that heat. You can view the math at: http://www.ciicontrols.com/library/ssr/heatcalculations.htm
As long as you use a quality SSR and mount it on a heatsink, it should have an incredibly long life.

Regards;
John Walsh, General Manager
Continental Industries International (We make temperature controls and SSRs.)

Phase Control

: I'm not an electronic engineer or even close, but my understanding is that
: SCR's are the most sophisticated switch available. A mercury switch turns
: on and off every few seconds. An SCR turns on or off 120 times a second.
: This allows for proportional control. (which lessens the overshoot factor)
: It can also be adjusted relative to bias and gain, which can limit the
: amount of current it draws. For instance I have an element that, due to
: it's being shortened, draws 2 amps more than the breaker will allow. On
: full gain it will pop the breaker, with the gain turned back it doesn't. I
: don't understand the difference between SSR's etc, but I think the SSR is
: less sophisticated and cheaper. It may indeed do a very good job of
: switching a kiln.
: Bert

"An SCR (Silicon Control Rectifier) is (without going into great detail) an electronic switch. It is either off or on, but you can delay when it is turned on in relation to the alternating current's cycle. If you turn it on early in the cycle, you get near full power. If you turn it on near the end of the cycle, you get very little power. Once turned on, it will stay on for the rest of the cycle. You can turn it on with vary little power, i.e. a few milliamps to the gate (control element) will allow you to "switch on" many amps (usually the full 15Amp circuit).
With this technique you can build a fairly well regulated power supply that will support regulated power to heaters, lights, etc. The only down side to a controller like this is that it creates a lot of electrical noise, and may drive radios in the area nuts."
samg (Sam Gaylord)

Since I have built the things from scratch (but not melted the silicon) I suppose I am qualified to offer a few definitions.

An SCR (Silicon Control Rectifier) is a device that rectifies current (changes it from AC to DC) which can be turned on with a small current. If used alone with AC, it produces pulses of 1/2 the AC power when on. Therefore, when SCR's are used, they are almost always used in pairs to pass both halves of the AC power.

A triac is effectively two SCR's built on one small piece of silicon with one wire controlling them. The two can't used separately. It is a triac that is used in dimmers and small motor speed controls.

An SSR (Solid State Relay) is a combination of a Zero Crossing Switch and a Triac/SCR. One of the problems of phase control it that it can produce electronic noise and some devices will overheat or otherwise be damaged by the different shape of the power (from a smooth sine wave.) If the SCR turn-on signal is applied at the time of the zero power crossing, then the output will be an AC half-cycle.  Then instead of phase controlling each cycle, power is on or off for complete half cycles and power may be controlled to the nearest 120th of a second.

Why then use SCR's instead of Triacs or SSR's? Well, for higher voltage, higher current, and 3 phase power.   Triacs will break down at higher voltages, have limited amperage available and don't offer the needed control. And for motors a curious thing happens: a motor shifts the phase relationship of current and voltage so the Triac thinks the power has never shut off, so it stays on. SCR's allow specific circuitry to deal with 3 phase power and the phase shifts.