WEBVTT Kind: captions Language: en 00:00:00.200 --> 00:00:01.854 This is my stovetop. 00:00:01.854 --> 00:00:06.040 It’s a glass-top radiant electric stove and if you’ve ever used one of these, 00:00:06.040 --> 00:00:09.234 you might have noticed something interesting about its behavior. 00:00:09.234 --> 00:00:16.496 The heating elements quickly begin to glow once switched on  and you can feel the intense heat coming from them right away. 00:00:16.496 --> 00:00:21.281 but when you turn the control knob down to, say, medium power - 00:00:21.281 --> 00:00:23.883 the element simply goes out. 00:00:23.883 --> 00:00:27.671 It’s not still running but at half-power, it’s just off. 00:00:27.671 --> 00:00:31.068 But then, before too long, it comes back on… 00:00:31.068 --> 00:00:33.647 then switches back off. 00:00:33.647 --> 00:00:36.663 This pulsing behavior endlessly repeats. 00:00:36.663 --> 00:00:38.650 What’s causing that? 00:00:38.650 --> 00:00:42.417 The answer is this funky component called an infinite switch, 00:00:42.417 --> 00:00:46.515 also known by the much, MUCH better name Simmerstat. 00:00:46.515 --> 00:00:51.827 It may not look like  much, but each heating element in the cooktop is wired to one of these things, 00:00:51.827 --> 00:00:54.974 and the simmerstats are responsible for that pulsing. 00:00:54.974 --> 00:00:56.645 Let me show you - 00:00:56.645 --> 00:01:00.324 through the magic of buying two of them, and some other stuff, 00:01:00.324 --> 00:01:04.892 I’ve built this little box so I can control anything with a simmerstat. 00:01:04.892 --> 00:01:11.076 I’ll plug in a lamp to make what it’s doing obvious in addition to a mystery load on the other outlet 00:01:11.076 --> 00:01:13.136 which isn’t important right now. 00:01:13.136 --> 00:01:15.142 Let me just stick on a control knob. 00:01:15.142 --> 00:01:16.496 That’s better. 00:01:16.496 --> 00:01:21.999 Set to high, the simmerstat doesn’t interrupt power flowing through it at all, 00:01:21.999 --> 00:01:29.103 but when you move it off of the highest power setting, eventually power cuts out. 00:01:29.103 --> 00:01:31.906 However, the interruption is pretty brief. 00:01:31.906 --> 00:01:34.435 Eventually, power returns. 00:01:34.435 --> 00:01:41.186 These periodic interruptions will repeat indefinitely, but as you turn the control knob further clockwise, 00:01:41.186 --> 00:01:43.965 each interruption increases in length. 00:01:43.965 --> 00:01:51.650 Once you reach the medium setting, it will eventually settle into a point where it spends roughly equal time on and off. 00:01:51.650 --> 00:01:59.462 And then the trend will continue - if you go further, it only operates in brief pulses and those pulses get farther and farther apart. 00:01:59.560 --> 00:02:03.323 The reason the simmerstat behaves like this is because… 00:02:03.323 --> 00:02:06.121 This is kind of the only option. 00:02:06.121 --> 00:02:11.129 Electrically, the heating elements in a conventional electric stove are merely resistors, 00:02:11.129 --> 00:02:12.888 a very simple component. 00:02:12.888 --> 00:02:16.003 But those resistors are huge! 00:02:16.003 --> 00:02:23.329 Each one in this stovetop is capable of outputting at least 1,200 watts and the larger ones pump out three kilowatts. 00:02:23.329 --> 00:02:31.483 That’s great for boiling water, but way too much power for more gentile cooking tasks like, oh, what’s a good example, 00:02:31.483 --> 00:02:33.076 Oh! A simmer. 00:02:33.076 --> 00:02:37.744 To allow for that, we have to tame them and reduce their power output. 00:02:37.744 --> 00:02:39.708 But how would you do that? 00:02:39.708 --> 00:02:46.050 You might think of adding a second resistor in series with the cooktop element, maybe even a variable resistor. 00:02:46.050 --> 00:02:50.933 but at these power levels, that theoretical resistor would have to be gigantic 00:02:50.933 --> 00:03:00.151 and as it restricted current flow it would generate quite a lot of heat of its own which is both wasteful energy-wise and would require cooling. 00:03:00.151 --> 00:03:07.091 So instead you might want to use a variable transformer such as a variac to produce a range of voltages to choose from - 00:03:07.091 --> 00:03:11.671 lower the voltage and you lower the power consumed by those heating elements. 00:03:11.671 --> 00:03:13.720 But there we have a similar problem: 00:03:13.720 --> 00:03:20.430 to handle this much power, that variable transformer would have to  be quite large and thus quite expensive, 00:03:20.430 --> 00:03:23.135 and remember you’ll need four. 00:03:23.135 --> 00:03:30.257 So, to allow for fine control over the power output of the heating elements with minimal energy losses and component costs, 00:03:30.520 --> 00:03:32.406 the simmerstat was born. 00:03:32.406 --> 00:03:40.016 This device produces any arbitrary power level by repeatedly switching its load on and off for varying periods of time, 00:03:40.016 --> 00:03:43.198 a method known as duty-cycle control. 00:03:43.198 --> 00:03:50.598 Before I explain what’s going on inside here, though, I want to point out that these things aren’t by any means a recent development. 00:03:50.598 --> 00:03:53.417 In fact, the tech is quite old. 00:03:53.417 --> 00:04:00.028 Coil-top stoves going back to the 1940’s use these same exact controls to modulate their power output. 00:04:00.028 --> 00:04:03.713 As a matter of fact, the tech just had its 100th birthday. 00:04:03.713 --> 00:04:12.435 The first place I found it described is in this 1924 patent for Chester I Hall’s invention assigned to General Electric. 00:04:12.435 --> 00:04:17.001 Coil-style heating elements like this are also just giant resistors, 00:04:17.001 --> 00:04:25.089 so just as with the glass-top stoves which came later, the  simmerstat was the most cost-effective way to regulate their output. 00:04:25.089 --> 00:04:29.868 That means coil-top stoves are exhibiting this pulsing behavior, too. 00:04:29.868 --> 00:04:36.560 However you generally aren’t aware of this happening with a coil-top stove because you can’t see it. 00:04:36.560 --> 00:04:45.405 The heat produced by these coils is generated by a thin wire element embedded in the center of the hollow metal tube which actually forms the coil. 00:04:45.405 --> 00:04:52.308 But the gap between the wire that produce heat and the walls of the tube is filled with a sand-like material. 00:04:52.308 --> 00:04:58.275 The sand fills the tube so it can be bent into different shapes with the wire inside staying centered, 00:04:58.275 --> 00:05:00.180 which keeps you from getting electric shocks. 00:05:00.180 --> 00:05:01.513 Which is pretty nice. 00:05:01.513 --> 00:05:06.894 But to get the heat produced by the wire element out of the tube and into cookware, 00:05:06.894 --> 00:05:14.910 it must make its way to the outer surface,  meaning it also has to warm up all the sand in the way which is a lot of thermal mass. 00:05:14.910 --> 00:05:20.571 The upshot is that it takes a long time for these elements to get hot enough to visibly glow, 00:05:20.571 --> 00:05:25.068 so the pulsing behavior of the simmerstat is visibly obscured. 00:05:25.068 --> 00:05:30.134 The elements below a glass-top stove are very, very different though. 00:05:30.941 --> 00:05:33.538 If I explained them would that be a tangent? 00:05:34.594 --> 00:05:38.017 Well it’s not directly related to simmerstats so I suppose it would be, 00:05:38.017 --> 00:05:41.124 but too bad - I bought this so we gotta discuss it 00:05:41.124 --> 00:05:42.703 (I expensed it). 00:05:42.703 --> 00:05:48.959 This heating element assembly is what’s actually under the glass of many glass-top radiant electric stoves. 00:05:48.959 --> 00:05:58.025 It’s incredibly simple, just a flat disc of heat-resistant support material  hosting a very long zig-zaggy piece of nichrome wire. 00:05:58.025 --> 00:06:03.679 Actually, two - this is a selectable size burner just like this one on my stove at home, 00:06:03.679 --> 00:06:10.920 composed of a six-inch inner-section and an outer ring to fill it out to the whole 9 inch diameter for larger cookware. 00:06:10.920 --> 00:06:16.181 They’re not all made exactly like this, sometimes the heating elements are structured a little differently, 00:06:16.181 --> 00:06:20.952 but they all work the same way: when voltage is applied across the nichrome wires, 00:06:20.952 --> 00:06:26.084 they get very hot very quickly to the point of glowing brightly. 00:06:26.084 --> 00:06:33.404 That speed is the main functional advantage of a glass-top stove: near instant heat output from cold. 00:06:33.404 --> 00:06:40.354 But you’ll notice that in open-air this looks very bright and orangey like the heating elements of a toaster. 00:06:40.354 --> 00:06:44.512 That’s because, well, that’s pretty much what these are! 00:06:44.512 --> 00:06:50.073 But, you say, under the glass of a stove these appear a deep cherry red when operating. 00:06:50.073 --> 00:06:51.490 Why is that? 00:06:51.490 --> 00:06:58.406 Well, this glass is in fact a special ceramic material  with some very peculiar characteristics. 00:06:58.406 --> 00:07:03.104 It’s deliberately very bad at conducting heat energy through itself 00:07:03.104 --> 00:07:10.700 which keeps other areas of the stovetop cool to the touch even when parts right next to it are hot enough to melt lead. 00:07:10.700 --> 00:07:17.144 But the glass lets infrared radiation pass right through it almost completely unimpeded. 00:07:17.144 --> 00:07:22.983 That’s what allows the radiant heat produced by the heating elements under the glass to make it into your cookware, 00:07:22.983 --> 00:07:28.769 and why you can feel intense heat coming from them right away once they're switched on. 00:07:28.769 --> 00:07:33.255 The deep red color seen through the glass is the result of filtering: 00:07:33.255 --> 00:07:39.413 the glass is opaque to almost all wavelengths of visible light, making it appear black to our eyes. 00:07:39.413 --> 00:07:42.788 But since it passes infrared light just fine, 00:07:42.788 --> 00:07:50.035 the near-infrared frequencies at the very edge of the visible spectrum will escape that filtering and you can see them. 00:07:50.035 --> 00:07:51.908 Pretty wild, right? 00:07:52.000 --> 00:07:57.820 And if you’ve ever noticed the probe thing going across the center of one of these burners and wondered what it’s for, 00:07:57.820 --> 00:07:59.537 well two things, actually: 00:07:59.537 --> 00:08:05.584 this is a sensing probe for two thermostatic switches in this little remote enclosure. 00:08:05.584 --> 00:08:11.311 The first of those switches is very sensitive and closes its contacts in the presence of minimal heat 00:08:11.311 --> 00:08:17.371 to illuminate the hot surface indicator (or indicators) to warn you that the cooktop is still hot. 00:08:17.371 --> 00:08:20.599 Count your blessings if you get a separate light for each burner, 00:08:20.599 --> 00:08:26.098 more often than not they’re all wired in parallel so any one of them can light up just a single warning light. 00:08:26.098 --> 00:08:28.613 Those cost-cutting cost-cutters. 00:08:28.613 --> 00:08:33.353 But the other switch in here is wired in series with the heating element itself. 00:08:33.353 --> 00:08:37.932 Remember that this flat disc of searing heat is trapped below glass, 00:08:37.932 --> 00:08:46.635 and while that glass is transparent to infrared,  it’s not perfectly transparent so it will absorb some heat energy and get quite hot. 00:08:46.635 --> 00:08:50.462 Plus, even if it were perfectly transparent to infrared, 00:08:50.462 --> 00:08:57.575 there’s gonna be a piece of cookware on top of the glass which reflects some of the heat energy right back down and into the element. 00:08:57.575 --> 00:09:06.974 Therefore, tremendous heat builds up in the tiny little  sliver of air space between the glass and the bottom of this infernal frisbee. 00:09:06.974 --> 00:09:09.532 To protect the glass from getting too hot, 00:09:09.532 --> 00:09:16.992 the temperature probe will open a safety switch to remove power from the heating element and keep things from getting all melty. 00:09:16.992 --> 00:09:19.629 It usually resets in a matter of a few seconds, 00:09:19.629 --> 00:09:25.305 but will kill power again if it needs to in order to enforce a high temperature safety limit. 00:09:25.305 --> 00:09:30.924 This usually only occurs if the burner has been in continuous use at full power for several minutes, 00:09:30.924 --> 00:09:33.455 such as when bringing water to a boil. 00:09:33.455 --> 00:09:36.950 Anyway, this video is supposed to be about the simmerstat. 00:09:36.950 --> 00:09:39.923 But, before we talk ab - no I’m kidding. 00:09:39.923 --> 00:09:46.255 Earlier I said that these modulate the power output of the cooktop burners using duty cycle control. 00:09:46.255 --> 00:09:53.950 If that doesn’t mean anything to you,  that’s just a way to say “the percentage of time spent powered over an average.” 00:09:53.950 --> 00:10:02.940 Imagine you have a 1,200 watt burner unit but you only need 600 watts of output for whatever particular cooking task you’re doing. 00:10:02.940 --> 00:10:06.737 Well, If you run that burner with a 50% duty cycle, 00:10:06.737 --> 00:10:11.001 perhaps by running it for 5 seconds of every 10 second period, 00:10:11.001 --> 00:10:16.751 then although the actual heat output will be alternating between zero and 1,200W, 00:10:16.751 --> 00:10:22.391 over time the effective heat output is the full power multiplied by the duty cycle - 00:10:22.391 --> 00:10:27.589 in this case, 1,200 X .5, or 600 watts. 00:10:27.589 --> 00:10:33.550 Those of you who are more digitally minded might be thinking about pulse-width modulation just now. 00:10:33.550 --> 00:10:43.369 Duty cycle control and PWM are quite similar when it comes to their net effect of producing a lower average power output by cycling a… 00:10:43.369 --> 00:10:45.840 whatever on and off repeatedly, 00:10:45.840 --> 00:10:49.577 but the two terms technically describe different things. 00:10:49.577 --> 00:10:56.905 Some might choose to argue with me on that, but PWM isn’t necessarily trying to modulate the power output of anything. 00:10:56.905 --> 00:11:01.565 In fact it can be used as a signaling protocol for fairly complex tasks. 00:11:01.565 --> 00:11:06.419 Servo motors, for instance, are sometimes controlled through PWM signaling. 00:11:06.419 --> 00:11:11.151 The length of the pulses they receive encodes a position for it to assume. 00:11:11.151 --> 00:11:15.097 When you are using PWM simply to modulate power output, 00:11:15.097 --> 00:11:20.957 say when you use a PWM dimmer to reduce the intensity of DC-powered LEDs, 00:11:20.957 --> 00:11:24.890 strictly speaking that is still duty cycle control: 00:11:24.890 --> 00:11:31.157 the apparent brightness of the LEDs is determined by what percentage of time they spend on vs. off, 00:11:31.157 --> 00:11:35.058 which is literally just another way to say their duty cycle. 00:11:35.058 --> 00:11:42.143 Pulse width modulation in that context is simply a modern means to the end of attaining duty cycle control, 00:11:42.143 --> 00:11:47.540 with the added benefit of high switching frequencies  providing it with some functional advantages. 00:11:47.540 --> 00:11:51.580 But since the simmerstat is literally 100-year-old tech, 00:11:51.580 --> 00:11:54.009 well there ain’t anything modern in here. 00:11:54.009 --> 00:11:59.240 Yet it absolutely is a fully-functional,  fully-capable duty-cycle controller. 00:12:00.160 --> 00:12:02.061 How on Earth could that be? 00:12:02.061 --> 00:12:05.754 Well, let’s look inside the simmerstat to see what’s going on. 00:12:05.754 --> 00:12:10.456 That, uh... that is less clear than you’d think. 00:12:10.456 --> 00:12:13.581 So first, because we’re- I'm gonna turn this off now. 00:12:13.581 --> 00:12:17.324 Because we're dealing with the US split-phase electrical system here, 00:12:17.324 --> 00:12:23.655 in the 240V circuit this is used to control there are two hot wires and no neutral. 00:12:23.655 --> 00:12:30.781 The heating elements of a cooktop are wired across line one and line two just like any other 240V device. 00:12:30.781 --> 00:12:34.026 You can learn more about that in this video if you'd like. 00:12:34.026 --> 00:12:37.612 Because of this fact, this device has two switch contacts in it, 00:12:37.612 --> 00:12:39.302 here and here, 00:12:39.302 --> 00:12:46.020 so that it can open both sides of the circuit when  switched off and completely isolate the heating element from voltage. 00:12:46.020 --> 00:12:51.162 That makes the simmerstat at its core a double-pole single-throw switch. 00:12:51.162 --> 00:12:57.325 Now the left side of the simmerstat doesn’t actually matter at all for its power modulation purpose. 00:12:57.325 --> 00:13:00.724 That’s really just an extra isolation point breaking line 1, 00:13:00.724 --> 00:13:05.152 and it’s only ever open when the control knob is in the off position. 00:13:05.152 --> 00:13:07.800 In all other positions it’s closed. 00:13:07.800 --> 00:13:16.233 Oh, and this extra copper piece here is used to send power out to a pilot light to indicate that a cooktop burner is switched on. 00:13:16.233 --> 00:13:22.407 That’s why the terminal on the back is labeled P and why  they didn’t bother giving it a proper contact surface - 00:13:22.407 --> 00:13:27.644 it’s just sending a tiny little amount of current to a neon indicator like this one. 00:13:27.644 --> 00:13:34.255 You’ll notice, though, that the right side of  the simmerstat features a much more robust pair of switch contacts, 00:13:34.255 --> 00:13:37.804 one of which is attached to a wide copper bar. 00:13:37.804 --> 00:13:43.897 This is the contact that regularly opens and closes to modulate power output. 00:13:43.897 --> 00:13:46.078 Uh, to hopefully avoid confusion, 00:13:46.078 --> 00:13:51.020 remember that you only have to break one side of a circuit to kill current flow. 00:13:51.020 --> 00:13:58.998 When the other switch is closed but this one is  open the heating element will still have 120V potential on it from line 1 00:13:58.998 --> 00:14:05.145 but there isn't a complete circuit to line 2 for any power to actually flow through it. 00:14:05.145 --> 00:14:11.807 Except the pilot light - well the pilot light is only operating at 120V and the other side of it is connected to neutral 00:14:11.807 --> 00:14:15.167 so that stays on no matter what cooktop burner is doing. 00:14:15.478 --> 00:14:17.799 I’m very sorry our power system is weird. 00:14:17.799 --> 00:14:21.108 Just… ignore everything on the left. It doesn’t matter. 00:14:21.108 --> 00:14:21.957 Moving on, 00:14:21.957 --> 00:14:26.349 if we look at the backside of the copper bar hosting our main switch contact 00:14:26.349 --> 00:14:31.170 we’ll see something that looks an awful lot like a bimetallic strip. 00:14:31.170 --> 00:14:32.913 Because it is. 00:14:32.913 --> 00:14:35.860 These little heroes show up in the darndest places. 00:14:35.860 --> 00:14:37.651 Why’s it in here? 00:14:37.651 --> 00:14:45.914 Well, this copper bar carries the current flowing through to the cooktop heating element when the switch contact is closed and power is flowing. 00:14:45.914 --> 00:14:49.803 And there’s a small amount of electrical  resistance across the copper bar - 00:14:49.803 --> 00:14:51.927 it’s too small to measure with a multimeter 00:14:51.927 --> 00:14:58.564 but it’s enough to produce a bit of heat when the 10 or 11 amps drawn by the cooktop burner flows through it. 00:14:58.564 --> 00:15:02.396 Bimetallic strips deform when they change in temperature, 00:15:02.396 --> 00:15:08.509 and since there's one attached to this copper bar which we’ve just established gets warm when current flows through it, 00:15:08.509 --> 00:15:12.096 the bimetallic strip will start to bend as that happens. 00:15:12.096 --> 00:15:16.183 Now, unfortunately I can’t really demonstrate this well in-circuit. 00:15:16.183 --> 00:15:18.193 You’ll see why in just a moment. 00:15:18.193 --> 00:15:24.579 What I can do, though, is use this little heat gun and show you what happens when the support bar warms up. 00:15:24.579 --> 00:15:27.410 It’s pretty subtle, so watch closely. 00:15:27.410 --> 00:15:34.341 When hot, the copper bar bends such that the bottom switch  contact moves deeper into the simmerstat body 00:15:34.440 --> 00:15:37.240 and farther away from its partner above. 00:15:37.240 --> 00:15:40.092 Now, think about what that means. 00:15:40.092 --> 00:15:44.794 If the copper bar warms up whenever the switch is  closed and power is flowing, 00:15:44.794 --> 00:15:50.252 and that warmth causes the bar to bend such that the switch will open, 00:15:50.252 --> 00:15:54.134 then the switch does not want to stay closed. 00:15:54.134 --> 00:15:58.920 Any time it is closed, it heats up and the bimetallic strip then tries to open it. 00:15:59.720 --> 00:16:02.089 In practice, it looks like this. 00:16:02.089 --> 00:16:09.345 From below we can’t see the switch contacts but we can see the copper bar moving back and forth ever so slightly. 00:16:09.345 --> 00:16:16.516 Every time the switch is closed and power flows to the load, the copper bar begins to warm up because of internal resistance. 00:16:16.516 --> 00:16:19.732 We can see this quite clearly with the thermal camera. 00:16:19.732 --> 00:16:28.302 As that happens, the bimetallic strip bends the bar such that  the lower contact begins moving closer to the camera and away from its partner. 00:16:28.302 --> 00:16:33.876 Eventually the strip bends far enough to break the circuit, so current stops flowing. 00:16:33.876 --> 00:16:37.744 Once it does, though, the bar rapidly begins to cool down. 00:16:37.744 --> 00:16:42.244 That causes it to reverse course and start  moving closer to the other switch contact, 00:16:42.244 --> 00:16:45.089 then they actually touch, current can flow again, 00:16:45.089 --> 00:16:48.401 the bar warms up, and the cycle repeats. 00:16:48.401 --> 00:16:52.622 And now, let’s see if you make a Technology Connection. 00:16:52.622 --> 00:16:55.877 I'm using a lamp right now plugged into the simmerstat. 00:16:55.877 --> 00:17:00.663 And that lamp, because of the simmerstat, is flashing. 00:17:00.663 --> 00:17:06.131 I’ve covered a certain other piece of tech which we use to make lamps flash in the past. 00:17:06.131 --> 00:17:10.548 That piece of tech is put in-series with the lamps it’s meant to flash. 00:17:10.548 --> 00:17:16.061 Current flowing through it would cause a switch inside to briefly open then shut repeatedly. 00:17:16.061 --> 00:17:19.870 I’m speaking, of course, about the turn signal flasher. 00:17:19.870 --> 00:17:24.574 These old-school thermal flashers also work thanks to a bimetallic strip. 00:17:24.574 --> 00:17:30.097 That strip will deform when heated and open or close a switch (depending on the design particulars) 00:17:30.097 --> 00:17:38.347 which repeatedly applies and removes power to the incandescent lamps which form the turn signals to make them flash and thus more noticeable. 00:17:38.347 --> 00:17:44.704 Now that behavior is certainly a great deal faster than the pulsing the simmerstat is doing right now 00:17:44.704 --> 00:17:48.018 but it’s not really any different, is it? 00:17:48.018 --> 00:17:51.032 It’s the same thing, just sped up. 00:17:51.032 --> 00:18:03.726 And at its core, this simmerstat really is just an overgrown turn signal flasher capable of handling up to 11 amps of current at 240V, or 2600W. 00:18:03.726 --> 00:18:07.036 But the simmerstat is a flasher with a twist - 00:18:07.036 --> 00:18:09.759 uh, literally. 00:18:09.759 --> 00:18:16.929 The turn signal flasher has a fixed duty cycle and behavior, at least when controlling the same load at the same voltage. 00:18:16.929 --> 00:18:22.084 But the simmerstat can adjust its duty cycle based on the position of the control knob. 00:18:22.084 --> 00:18:24.629 Why? And how? 00:18:24.629 --> 00:18:27.950 Y'know, in Britain they call cooktop burners hobs. 00:18:27.950 --> 00:18:30.398 Does that mean that this is a hob knob? 00:18:30.398 --> 00:18:33.123 Anyway, when we were looking at this from above, 00:18:33.123 --> 00:18:37.403 you might have noticed that the switch contacts were nowhere near each other. 00:18:37.403 --> 00:18:41.276 They’re clearly sprung such that  they default to the open position, 00:18:41.276 --> 00:18:44.977 so what closes those switches in the first place? 00:18:44.977 --> 00:18:47.983 To find out, let’s switch to the Cam Cam. ♫ Offenbach plays out of nowhere ♫ 00:18:47.983 --> 00:18:50.000 It’s this cam! 00:18:50.601 --> 00:18:52.844 I don’t know how I live with me either. 00:18:52.844 --> 00:19:00.606 The control knob of the simmerstat is attached via a shaft to this plastic cam covered in a copious amount of grease. 00:19:00.606 --> 00:19:09.292 When assembled, the cam presses down on these protrusions once  rotated out of the off position, and that is what closes the switch contacts. 00:19:09.292 --> 00:19:12.904 You’ll notice the cam has two sections with different profiles - 00:19:12.904 --> 00:19:18.684 the inner section presses on and closes the switch  on the left of the simmerstat which remember, doesn't matter. 00:19:18.684 --> 00:19:20.421 It’s just a safety switch. 00:19:20.421 --> 00:19:27.014 But the outer section, which engages with bimetallic switch, has a very subtle ramp built into it. 00:19:27.014 --> 00:19:28.392 Can you see that? 00:19:28.392 --> 00:19:31.894 This is the off position which doesn’t press on the switch at all. 00:19:31.894 --> 00:19:37.221 But just to the left of that position, the cam profile gets very tall. 00:19:37.221 --> 00:19:40.926 That means it presses the switch down quite far. 00:19:40.926 --> 00:19:49.701 But then there’s a fairly steep dropoff before the cam very subtly gets thinner and thinner all the way back to the off position. 00:19:49.701 --> 00:19:51.800 Can you figure out why that might be? 00:19:51.800 --> 00:19:53.737 Vote now on your phones. 00:19:53.737 --> 00:20:04.049 What that varying cam profile actually does as you turn the knob is change the resting position of the two switch contacts in the bimetallic switch. 00:20:04.049 --> 00:20:11.298 You can see from below that as I turn it, the copper bar and bimetallic strip are moving up and down very slightly. 00:20:11.298 --> 00:20:13.681 This might seem pretty inconsequential, 00:20:13.681 --> 00:20:18.518 but that right there is actually the key to this whole device. 00:20:18.518 --> 00:20:22.485 But explaining why... is complicated. 00:20:22.485 --> 00:20:28.707 I’ve been stuck on this script for a while because  although this component is incredibly simple, 00:20:28.800 --> 00:20:34.440 there are three connected concepts all  working together here to make this what it is,   00:20:34.440 --> 00:20:38.441 and that makes it hard to explain without getting stuck in a loop. 00:20:38.441 --> 00:20:39.916 But I’ll try. 00:20:39.916 --> 00:20:42.159 Let’s revisit that footage from earlier. 00:20:42.159 --> 00:20:46.312 Here, the simmerstat was set to medium and I let it stabilize. 00:20:46.312 --> 00:20:50.975 In this condition, the switch is closed for about 5 seconds before it opens, 00:20:50.975 --> 00:20:54.757 and then it stays open for about 5 seconds, and this repeats. 00:20:54.757 --> 00:20:58.241 That’s how we get a 50% duty cycle. 00:20:58.241 --> 00:21:03.961 But why precisely is the switch opening and closing with such predictability? 00:21:03.961 --> 00:21:06.956 To find out, let’s look at the thermal camera again. 00:21:06.956 --> 00:21:13.789 You’ll notice that at this setting, the contact’s support bar seems to peak right near 100 degrees Celsius, 00:21:13.789 --> 00:21:16.166 then the temperature begins to fall. 00:21:16.166 --> 00:21:23.306 We then see that it consistently bottoms out right near 78 degrees Celsius and it begins to rise again. 00:21:23.306 --> 00:21:32.133 This tells us that the switch is actually opening and closing based upon the temperature of the support bar and its bimetallic strip. 00:21:32.133 --> 00:21:35.193 Simple enough, but why, though? 00:21:35.193 --> 00:21:40.452 Why is the circuit opening and closing at those specific temperatures? 00:21:40.452 --> 00:21:47.039 Well, that’s because of this cam and how far it’s pushing down on the top switch contact. 00:21:47.039 --> 00:21:53.048 Remember that the bottom switch contact  retreats into the body of the simmerstat as it warms up. 00:21:53.048 --> 00:22:00.600 How far that contact actually moves is a function of the temperature of the bimetallic strip in the support bar. 00:22:00.600 --> 00:22:03.922 And through the position of the cam and its profile, 00:22:03.922 --> 00:22:11.255 we have chosen to place the top switch contact at some specific point along that deflection path. 00:22:11.255 --> 00:22:19.303 In this position, the contacts are forced to stay together until the copper bar has reached 100 degrees Celsius. 00:22:19.303 --> 00:22:25.416 At that precise temperature, the deflection of the  bar is sufficient to open the switch. 00:22:25.416 --> 00:22:30.713 So really, the core function of this device... is a thermostat. 00:22:30.713 --> 00:22:32.105 Through turning the knob, 00:22:32.105 --> 00:22:39.059 we’re deciding how hot we want to allow that copper bar to get before it switches off the cooktop burner. 00:22:39.059 --> 00:22:45.364 So, the last piece of the puzzle is how that choice becomes a consistent duty cycle. 00:22:45.364 --> 00:22:52.361 Because remember, the goal of this device is not to maintain a specific temperature like the thermostat in an oven, 00:22:52.361 --> 00:22:58.114 but to maintain a specific duty cycle and thus power output for the cooktop burners. 00:22:58.114 --> 00:23:02.966 Yet somehow we’re doing that with what ostensibly is a thermostat. 00:23:02.966 --> 00:23:05.709 Well, here’s where I hope it all comes together. 00:23:05.709 --> 00:23:12.639 Remember that the copper bar inside here is both a thermostatic switch AND a heater. 00:23:12.639 --> 00:23:20.826 Whenever the switch is closed, it dissipates a consistent amount of power and that generates a consistent amount of heat within the bar. 00:23:20.826 --> 00:23:26.082 And like all heat-producing things, with a consistent power output 00:23:26.082 --> 00:23:36.321 how hot it actually gets and thus how far the bar will deflect is a function of how long that heater runs in a given period. 00:23:36.321 --> 00:23:39.046 And now physics becomes our friend. 00:23:39.046 --> 00:23:43.456 Let’s say I change the knob’s position to that of medium-high. 00:23:43.456 --> 00:23:48.222 What that will actually do is press down farther on the top switch contact 00:23:48.222 --> 00:23:55.265 to force the two contacts to remain together until the lower support bar has reached 130 degrees. 00:23:55.265 --> 00:23:57.936 So, let’s do that. 00:23:57.936 --> 00:24:03.077 The temperature of the bar, since power is now flowing through it, is climbing. 00:24:03.077 --> 00:24:09.826 But you’ll notice that the rate of change in temperature is slowing down as it continues to increase. 00:24:09.826 --> 00:24:17.511 What’s happening here is the result of the fact that the bar is approaching the limit to how hot it can possibly become 00:24:17.511 --> 00:24:25.132 before the heat it gains through its internal resistance  matches the heat that leaves through radiation to the air surrounding it. 00:24:25.132 --> 00:24:30.620 And as we approach that limit, heat gain slows significantly. 00:24:30.620 --> 00:24:36.243 This means that it will take longer to reach the new target temperature of 130 degrees, 00:24:36.243 --> 00:24:43.842 which in turn means that the switch contacts will stay  closed for a longer period of time before they open again. 00:24:43.842 --> 00:24:50.843 And adding to that, the temperature differential between the bar and the air around it is greater when it’s hotter 00:24:50.843 --> 00:24:57.402 which means that once it stops being heated, it’s going to lose heat energy faster than it did before. 00:24:57.402 --> 00:25:02.590 That shortens the time the switch spends open before it closes again, 00:25:02.590 --> 00:25:07.660 though that effect is minor compared to the stretching of the on-time. 00:25:07.660 --> 00:25:09.885 But what about going in the other direction? 00:25:09.885 --> 00:25:13.849 What happens when you turn the knob to, say, medium-low? 00:25:13.849 --> 00:25:21.152 Well, with the knob at the nine o’clock position, the cam is only very slightly pushing on the switch. 00:25:21.152 --> 00:25:28.793 It’s pushing so gently that the bar only needs to hit about 60 degrees Celsius before it bends enough to open the switch. 00:25:28.793 --> 00:25:34.260 Since that’s much closer to ambient temperature than 130 or even just 100 degrees, 00:25:34.260 --> 00:25:41.124 it takes a very short time for it to reach that temperature when being heated by the current passing through. 00:25:41.124 --> 00:25:47.544 And you’ll notice that the switch only closes again when the bar drops to about 40 degrees Celsius - 00:25:47.544 --> 00:25:51.653 but it takes quite a while for the bar to cool to that temperature, 00:25:51.653 --> 00:25:56.154 so the pulses it sends out are both short and infrequent. 00:25:56.154 --> 00:25:58.649 Do you see how this all fits together? 00:25:58.649 --> 00:26:04.875 We are really controlling this circuit based on the temperature of the heater inside of it. 00:26:04.875 --> 00:26:10.131 When calibrated correctly, that can be used as a duty cycle controller. 00:26:10.131 --> 00:26:16.626 Because to get that heater hotter, the heater itself must run with a longer duty cycle, 00:26:16.626 --> 00:26:25.310 so its average temperature becomes an effective proxy for the duty cycle necessary to attain that temperature. 00:26:25.310 --> 00:26:27.882 And that’s incredibly fortuitous. 00:26:27.882 --> 00:26:33.312 Just calibrate the bimetallic bar to dissipate the right amount of heat when the circuit is passing current, 00:26:33.312 --> 00:26:41.124 give the cam pressing on the switch a nice subtle ramp to allow you to break the circuit at a specific temperature of that bar, 00:26:41.124 --> 00:26:47.640 and you can produce any duty cycle  you need with incredibly crude technology. 00:26:47.640 --> 00:26:49.578 But it gets even better! 00:26:49.578 --> 00:26:54.566 Because the switching action occurs based on the temperature of the bimetallic strip, 00:26:54.566 --> 00:26:58.205 there’s a really helpful memory effect happening here. 00:26:58.205 --> 00:27:03.438 The main annoyance of using a conventional electric stove is the reaction time. 00:27:03.438 --> 00:27:07.020 The thermal mass of the materials involved retain heat for a while, 00:27:07.020 --> 00:27:11.678 either the coil of a coil-top stove or the glass of a glass-top. 00:27:11.678 --> 00:27:16.164 That’s useful because it helps smooth out the effect of the pulsing behavior 00:27:16.164 --> 00:27:22.238 but it also means it takes a good while for the cooktop to  react to a change in power level. 00:27:22.549 --> 00:27:28.471 The simmerstat can’t solve that problem, but consider what happens when you change the power level: 00:27:29.080 --> 00:27:34.640 If, say, you were on a medium-low heat  but needed to move to a medium-high heat,   00:27:34.640 --> 00:27:42.166 then your turning of the knob is simply changing how hot the internal heater needs to get before it cuts power. 00:27:42.166 --> 00:27:47.782 And if at the medium-low setting the heater was maintaining, let’s say, 80 degrees on average, 00:27:47.782 --> 00:27:54.762 then when you change the setting it’s gotta get all the way up to 130 degrees before it switches the element off. 00:27:54.762 --> 00:28:01.762 That means it’s going to produce a very long pulse of output and help the burner get to your new target as quickly as possible. 00:28:01.762 --> 00:28:06.964 The same goes when you lower the output - if it’s already quite hot, 00:28:06.964 --> 00:28:13.400 then turning the knob will open the switch and it won’t close again until the heater has fallen down to the new target,   00:28:13.400 --> 00:28:15.700 which is going to take a while. 00:28:15.700 --> 00:28:23.623 I first noticed this behavior of my glass-top stove and assumed that this was being accomplished with logic but nope! 00:28:23.623 --> 00:28:25.554 That’s just how this works! 00:28:25.554 --> 00:28:28.001 Now I’ve left something important out. 00:28:28.001 --> 00:28:34.361 You might wonder how the simmerstat can keep the element on at full-power when it's set to high. 00:28:34.361 --> 00:28:39.214 Well, that’s what that really high point on the cam profile was for. 00:28:39.214 --> 00:28:46.075 That just really jams that switch down so  it’ll never open no matter how hot the bar inside is getting. 00:28:46.075 --> 00:28:47.771 Pretty crude, huh? 00:28:47.771 --> 00:28:50.918 Except… the crudeness doesn’t stop there. 00:28:50.918 --> 00:28:59.510 This particular simmerstat only works correctly with fairly large heating elements that draw the 8.9 to 11 amps it’s rated for. 00:28:59.510 --> 00:29:05.548 That’s because how quickly the copper bar heats up  depends on how much current passes through it. 00:29:05.548 --> 00:29:11.544 If you try and control a load outside of that range,  things get out of whack. 00:29:11.544 --> 00:29:18.856 Through sheer dumb luck, this 1,100 watt hot plate draws about 9 amps at 120V, 00:29:18.856 --> 00:29:22.139 so this simmerstat works correctly with it. 00:29:22.139 --> 00:29:25.428 The hotplate has been my mystery load throughout the video. 00:29:25.428 --> 00:29:30.215 But if I plug in this smaller hotplate which only draws 900 watts, 00:29:30.215 --> 00:29:33.925 the duty cycles this is meant to produce get all out of whack. 00:29:33.925 --> 00:29:38.424 And with a load much smaller than that, it never interrupts power. 00:29:38.424 --> 00:29:44.176 You also might have noticed that the way this switches the load is terrible! 00:29:44.176 --> 00:29:50.411 Switches ideally should have a snap-action to reduce arcing when they break electrical loads. 00:29:50.411 --> 00:29:52.674 This fella just doesn’t. 00:29:52.674 --> 00:29:57.064 The contacts barely move despite switching 10 amps, 00:29:57.064 --> 00:30:00.685 and so some fairly nasty arcing occasionally happens. 00:30:00.685 --> 00:30:05.613 It’s not too too bad because these are just used to switch resistive loads, 00:30:05.613 --> 00:30:14.360 and the overextension of the switch as the cam presses down produces a wiping effect that helps to clean the contacts of carbon buildup and debris. 00:30:14.360 --> 00:30:18.820 Incidentally, that also happens inside the switches of  an electromechanical pinball machine 00:30:18.820 --> 00:30:22.156 and yes part three is coming I haven’t forgotten about it hold your horses! 00:30:22.156 --> 00:30:24.942 I just thought a script like this one would be faster. 00:30:24.942 --> 00:30:26.165 Why did I think that? 00:30:26.165 --> 00:30:28.695 I don’t know. It never works! 00:30:28.695 --> 00:30:37.969 Anyway, before I end this video I want to make sure I say that not all simmerstats are going to function exactly like these ones do. 00:30:37.969 --> 00:30:41.230 For instance, the simmerstats on my stove at home? 00:30:41.230 --> 00:30:50.386 Three of the four can’t possibly use the current flowing through  them to heat the bimetallic strip inside because they have selectable size elements. 00:30:50.386 --> 00:30:52.879 One even has three sizes. 00:30:52.879 --> 00:30:57.876 I suspect the core functionality of these simmerstats is exactly the same as this simple one, 00:30:57.876 --> 00:31:04.529 but there’s probably a dedicated resistor with its own path back to neutral producing the heat for the bimetallic strip, 00:31:04.529 --> 00:31:10.053 that way the duty cycles are consistent regardless of the load they're controlling at any particular time. 00:31:10.053 --> 00:31:14.118 I also think it might be controlling a relay or, another possibility, 00:31:14.118 --> 00:31:21.019 the knob could actually be controlling a variable resistor producing differing amounts of heat to open a limit switch. 00:31:21.019 --> 00:31:27.818 I kinda think that might be the case because these have a definite click as they cycle on and off which this basic model doesn’t. 00:31:27.818 --> 00:31:30.418 That’s just a guess, though - I’m not tearing this apart. 00:31:30.418 --> 00:31:31.479 I need it. 00:31:31.479 --> 00:31:37.529 And by the way, I built this box for this video  but I’ve been wanting to build something like it for quite a while 00:31:37.529 --> 00:31:41.264 specifically because of these cheap hotplates. 00:31:41.264 --> 00:31:46.710 If you’ve ever used one, you might have noticed that they are impossible to control! 00:31:46.710 --> 00:31:52.902 That’s because this knob is not controlling a simmerstat - it’s just a plain ol’ thermostat. 00:31:52.902 --> 00:31:57.816 Notice it only clicks on when you’re well away from the off position. 00:31:58.779 --> 00:32:01.697 You can kind make this work for what you need, 00:32:01.697 --> 00:32:09.443 but it’s incredibly hard because you don’t even know where it’s measuring the temperature and it’s going to change based on whatever cookware you're using 00:32:09.443 --> 00:32:12.420 It’s a bit easier to control this style of hotplate 00:32:12.420 --> 00:32:17.645 where the heating element is embedded in a metal disc and its average temperature kind of means something, 00:32:17.645 --> 00:32:23.142 but this coil style is flat-out impossible to control with a thermostat. 00:32:23.142 --> 00:32:28.955 I’m sure it’s just a lot cheaper than a proper simmerstat - I mean, these things are like $15. 00:32:28.955 --> 00:32:34.527 But it’s incredibly annoying and makes these pretty much only useful for boiling water. 00:32:34.527 --> 00:32:39.275 I thought I was going to end up building, like, an Arduino-based controller or something 00:32:39.275 --> 00:32:43.776 but it turns out you can just stick a simmerstat in a handy box and be done! 00:32:43.776 --> 00:32:46.005 I added the lights because it pleases me. 00:32:46.005 --> 00:32:50.242 The red one is the pilot light and the yellow one indicates power is flowing. 00:32:50.242 --> 00:32:53.182 Uh… I will not show you how I made that work. 00:32:53.182 --> 00:32:58.550 And a final point: it could be argued that we really shouldn’t be using these anymore. 00:32:58.550 --> 00:33:06.360 I mean, if you get an induction stove it’s not going to have them, but don’t get me started on the touch controls those often have - 00:33:06.360 --> 00:33:10.861 hey, note to appliance designers, nobody wants that! 00:33:10.861 --> 00:33:12.630 Just use knobs. 00:33:12.630 --> 00:33:22.035 Touch controls on a stove would be like taking away a car’s turn signal stalks and forcing people to adapt to weird buttons for no good reason at all. 00:33:22.035 --> 00:33:31.078 Anyway, what I mean by that is the simmerstat works great with coil-top stoves since these heating elements retain so much heat, 00:33:31.078 --> 00:33:40.340 but with glass-tops, where radiant heat is transmitted  right through immediately, the switching frequency is arguably too slow. 00:33:40.340 --> 00:33:45.371 I don’t often notice this,  and to be honest I’m not sure it actually matters, 00:33:45.440 --> 00:33:49.316 but when I’m frying up some veggies in a thinner piece of cookware, 00:33:49.316 --> 00:33:52.702 I can tell when the element is running and when it’s not. 00:33:52.702 --> 00:33:59.947 The sizzling gets a little louder whenever it’s on and if there’s any water in the bottom of the pan, it bubbles more vigorously. 00:33:59.947 --> 00:34:03.166 Again, I’m honestly unsure of how much that actually matters, 00:34:03.166 --> 00:34:07.947 but I’m not a good enough cook to pretend my opinion is any good. 00:34:07.947 --> 00:34:11.985 Anyway, with modern high-power solid-state switching components, 00:34:11.985 --> 00:34:17.086 cooktop burners could in theory be controlled with as much finesse as a dimmer switch provides. 00:34:17.086 --> 00:34:22.554 That would cost more, of course, but at this point I’m not sure how much it would. 00:34:22.554 --> 00:34:29.345 I mean, induction stoves have some wild power-switching circuitry in them and they’re not very expensive anymore. 00:34:29.345 --> 00:34:34.286 Now, the main thing I’d be worried about in this hypothetical is actually noise. 00:34:34.286 --> 00:34:38.282 These fellas make a fairly noticeably humming when they’re switched on, 00:34:38.282 --> 00:34:44.865 and adding some high-frequency switching to that mix could turn that hum into a weird ring. 00:34:44.865 --> 00:34:53.152 But hey, even just, like, a 1 Hz switching frequency would be a huge improvement over the simmerstats in my stove at home. 00:34:53.152 --> 00:34:54.823 Something to think about. 00:34:54.823 --> 00:34:57.119 OK. Well, that’s it! 00:34:57.119 --> 00:34:58.399 I think. 00:34:58.399 --> 00:35:01.342 Didn’t imagine this script would get so out of hand. 00:35:01.342 --> 00:35:04.906 I mean it’s literally just a bimetallic strip in a box with a knob. 00:35:04.906 --> 00:35:08.965 But at this point, I don’t know why I’m surprised. 00:35:08.965 --> 00:35:12.591 Here - how ‘bout we gey a taste of November in April? 00:35:12.591 --> 00:35:16.690 I think I’m gonna make something actually simple before the month is out. 00:35:16.690 --> 00:35:18.417 Can I rise to the challenge? 00:35:18.417 --> 00:35:21.448 Find out on the next… whatever. 00:35:21.448 --> 00:35:22.667 Whenever it happens. 00:35:23.972 --> 00:35:24.787 Bye. 00:35:25.530 --> 00:35:28.081 ♫ dutifully smooth jazz ♫ 00:35:29.434 --> 00:35:33.028 I’ll plug in a lamp to make what it’s doing obvious in addition to the 00:35:33.028 --> 00:35:34.900 eahhhh 00:35:34.900 --> 00:35:36.376 The ans- [strange noises] 00:35:36.376 --> 00:35:39.249 However, the interruption is pretty brief. 00:35:39.249 --> 00:35:42.255 Eventually, power returns. 00:35:43.404 --> 00:35:44.745 [eventually stretches on] 00:35:45.056 --> 00:35:46.114 [laughs] 00:35:46.114 --> 00:35:49.489 This is gonna be harder to time than I thought it was going to be 00:35:49.489 --> 00:35:52.448 Once you reach the medium setting, now - 00:35:52.448 --> 00:35:55.223 oh right, that’s what’s supposed to happen 00:35:55.223 --> 00:35:58.699 These periodic interruptions will repeat indefinitely, 00:35:58.699 --> 00:36:03.331 but as you continue to control the turn knob further clockwise [breaks into laughter] 00:36:03.331 --> 00:36:08.395 Earlier I said these modulate the power output of the cooktop burners (loud thud) 00:36:08.395 --> 00:36:10.489 that was loud! 00:36:10.489 --> 00:36:13.826 But to maintain a specissssffsfsfsfssfs 00:36:15.035 --> 00:36:17.528 You've heard of a flash in the pan, 00:36:17.528 --> 00:36:20.276 what about a flash below the pan? 00:36:20.276 --> 00:36:23.884 I think that joke needs a little more time in the oven, don't you? 00:36:23.884 --> 00:36:26.052 Half-baked at best. 00:36:26.052 --> 00:36:28.878 But hey, at least I made a meal out of it. 00:36:28.878 --> 00:36:30.521 Soup's on!