MOSFET Switch

Semiconductor Discrete Semiconductor Products
,
1

Pardon the incoming stupidity. I’m a physics student, so something as electronically involved as this is a little outside of my wheelhouse, at least in this stage of my education.

I recently got a MOSFET, and the intention was to use it as a “signal amplifier” of sorts. I have an Arduino to put out a PWM signal, and the hope is that I can use a MOSFET to behave as a switch that turns on and off with the PWM signal. The MOSFET switch would then be connected to a 120V power supply, which runs a motor (really high voltage rating for a motor, I know, but it’s correct). That way, I would basically have a 120V PWM signal instead of the 5V PWM signal, and I could change the duty cycle to control the motor’s speed. Easy peasy. A MOSFET is just a device that I found other people using in this way online. Hypothetically (and idealistically), all I need is for the MOSFET to behave like a switch.

The problem is that MOSFET’s clearly don’t work the way I thought they do, because my setup isn’t working like I expected it to. I have a schematic below:
MOSFET Confusion

This circuit is somehow wrong, and I don’t know why.

  1. There is a diode there inside of the MOSFET I have according to the spec sheet. I’m not sure why it’s there, and I don’t think it affects anything. If I had to take a guess, it’s that building up charge while the MOSFET is off with the voltages reversed would somehow hurt the semiconductor inside, but that’s just a guess without any physical knowledge of its inner workings behind it. First question: why is the diode there?

  2. I see everywhere I look that there’s supposed to be a resistor between source and gate. Second question: Why is that? My guess is that it’s a means of discharging the gate, but I don’t know if that’s right or if that’s a good thing for the MOSFET, anyway.

  3. Third question: why doesn’t this setup work? When the Arduino is unattached, there’s no voltage across the gate, so the motor should not be powered. I measure 0V on my multimeter when I do that. So far so good.

The problem comes when I attach a steady 5V from the Arduino to the gate. This does NOT turn it on (again, measured with a multimeter and not the actual motor), even though I thought it should. Why doesn’t this setup turn on?

NOTE: Yes, I know I should have a flywheel diode in there somewhere. That’s another question itself; I don’t know where I should put it, yet. However, that’s why I’ve been using a multimeter this whole time instead of watching if the motor started spinning.

2

Greetings,

In response to your questions:

  1. The body diode in a FET is pretty much an inevitable artifact of how they’re made, though one which in various circumstances can be put to good use. This resource, while perhaps a bit chewy is a rather good summary of various FET properties and their significance.

  2. While common practice, it’s not necessary in the strictest sense. Putting it there does indeed discharge the gate in the absence of a drive signal, creating a well-defined “off” state under such conditions. Leaving the gate “floating” would leave the FET vulnerable to being turned on by all manner of stray signals, not the least of which is the 50/60Hz floating around everywhere thanks to the electrical grid. That would tend to end badly in cases such as this where some substantial amount of power is being switched.

  3. “cuz ya did something incorrectly” is a correct answer more often than not, with “'cuz ya broke something” coming in second. Neglecting to connect your arduino’s ground to the FET’s source would be a likely example of the first case, and the FET being damaged due to static electric discharge or some other factor an example of the second. FETs tend to fail short circuit at first until they explode or become crispy and burnt, at which point they become open circuits. You didn’t describe that, so my bet is on the first case.

Inductors oppose changes in current flow, similar to the way that an object’s mass opposes changes in its motion. As a physics student the equation f=ma should have some familiarity, and you should also recognize that acceleration (a) is shorthand for the derivative of velocity with respect to time, dv/dt, allowing the equation to be written as f=mdv/dt. Notice that this is very, very similar to the equation for voltage across an inductor, which can be written as v=L*dI/dt.

The point of a “freewheeling diode” here would be to provide a path for current flow through the motor when the FET is turned off, so that a large dI/dt isn’t imposed across the motor’s L, and doesn’t generate a large V which exceeds the FET’s prescribed limits as a result. Think about it for a moment and the proper placement of such a diode should present itself.

Something else to keep in mind is that internal to the FET, there’s perhaps a few nanometers of insulating material between the gate and the semiconductor material that’s fussing around with a 120V supply. If something goes amok and that insulation breaks down, you have the potential of ending up with that 120V supply being connected to your arduino’s I/O pin, which is not likely to end well. The use of isolation devices is a common precaution in such situations.

2 Likes
3

This was a remarkable follow-up! Thanks for all the clarifications!

I do have some follow-up questions, though.

Why is this necessary? Isn’t the FET’s source already grounded by being connected to the other end of the power supply? What does grounding the FET’s source to the Arduino do here that isn’t being done already, and why would that cause my FET to not turn on when the gate is getting its 5V?

The freewheel diode is only something I’ve seen other people use in their own diagrams for projects similar to mine own. Back emf makes sense, and I understand the physics of it (I’m a junior undergrad for reference. The physics is familiar by now, just not the circuit common-sense). My question now is what makes a freewheel diode any different from a normal diode, or is a freewheel diode a normal diode that’s just used on an inductive load in that way?

Furthermore, if the motor already has a ground to it (it has five wires, actually: two for field, two for armature, one ground) that is, in fact, being grounded in the larger circuit I’m constructing, then does that cover me for the back emf and bypass the need for a freewheel diode?

Lastly, thanks for the heads up, but what’s the difference between an isolator and a circuit breaker? Can’t I simply use a circuit breaker, instead, or is that not going to work because circuit breakers are sensitive to current and not voltage? Is a circuit breaker maybe one kind of isolator? All I’m seeing in the link you sent is a bunch of 6 or 8 pin doodads that I don’t know how I would use.

UPDATE: I just got to my lab and tried the circuit with the Arduino ground to FET source connection, and the circuit actually functions like a switch now. Your answer was, in fact, the solution to what I was missing.

4

The arduino’s outputs are measured relative to its own “ground” and the FET’s drain-source resistance is a function of the voltage present at its gate, relative to it’s source; the two need to have a shared reference point in order to have any meaningful communication. Presented differently, the gate-source junction behaves like a capacitor; some current has to flow in order to deliver charge to the capacitor to increase the voltage across it. Current flow requires a complete circuit, which one doesn’t get by connecting an I/O output alone.

Mostly the latter; the term “freewheel diode” refers to the application or intended use more so than to the device itself.

“Ground” is probably the most widely misused term in the electronic realm. Electronics folk often use the term to describe the arbitrary point in a circuit designated as the reference point for other measurements, regardless of what else that circuit node is or isn’t connected to. Electric utility folks typically take “ground” to mean an connection to the giant dirtball beneath our feet, because their power plants are connected to said dirtball.

In all likelihood, your motor’s “ground” terminal has no electrical connection except to the conductive chassis of the device, and is irrelevant to the business of making things spin. That’s done as a safety precaution; in the event a rogue wire connected to the power grid should contact the chassis, fault current can flow to ground through the ground lead, rather than through a person who might happen to be touching the thing at the moment.

“Back EMF” is a subtly different concept than that which a freewheeling diode would address, but that’s another can of worms. An FR207 costs less than a gumball, and as such is some pretty cheap insurance.

Isolators permit the transfer of information between circuits that have no DC connection to each other; circuit breakers interrupt overcurrent conditions with the goal of keeping failed devices from setting the building on fire. Proper use of an isolator will greatly reduce the risk of shock or damage to your arduino and any connected equipment in the event of FET failure.

The basic pattern is a device with a diode on one side and a photo-detector on the other, separated by an optically transparent insulator. Drive the diode on one side from your arduino and your FET is driven from the other side, with insulation good for several kV separating the two rather than just the gate insulation in the FET, which typically breaks down at 30V or less. They’re not that complicated, and good insurance in circumstances such as you describe.

It’s a common enough type of problem, and one of “the basics” a person should start with when troubleshooting electronic stuffs. Right up there with making sure that the power is turned on…

1 Like
5

One last thing still confuses me, and that’s the isolators.

This is one of the devices in the link you sent me: ADUM3221ARZ-RL7 Analog Devices Inc. | Isolators | DigiKey

It’s got eight pins. How is that a diode with a photodetector? The datasheet says the pinouts are two grounds, two isolator supply voltages, and two sets of logic I/O.

If this isn’t what you’re talking about, can you send me an example of a product that is? I’m still quite confused.

6

There’s often more than one route to the same destination; the device you mention is based on a magnetic coupling technology rather than an optical one. Power’s needed on both sides of its isolation barrier, but in return a person gets some performance benefits over other approaches.

A TLP152 would be an example of a diode+detector based device. It does require an isolated power supply on the output side, but the higher-performance optical gate driver devices typically do. A subtle variation on the concept is exemplified by a TLP190B; these products are basically just an LED and miniature solar panel in a package and do not require an isolated power supply, but they’re brutally slow by comparison.

1 Like
7

I think I’m beginning to understand how the isolators work, but now I have some more issues, mostly based on the premise of these devices in the first place.

  1. I think, at least, that the TLP152 now makes sense to me, after further poring through its datasheet. However, if an isolating gate driver works by allowing current to pass on the output side when current is passing on the input side, then why don’t I simply use one of these instead of the FET in the first place? Can’t these isolators also operate as switches? Why is it common practice to use these devices with MOSFET’s, rather than instead of MOSFET’s?

  2. How would I supply power to the output side? The FET’s I have need 5V on their gates to switch, so I obviously need a 5V power source. I can’t use the Arduino, though, because that’s what I’m trying to protect here.

  3. Why can’t I just use a diode here instead of one of these isolators? Wouldn’t that ensure that the path from the Arduino to the FET is unidirectional?

  4. Is the TLP152 suitable for my needs? There’s a warning in the datasheet that says that it’s not reliable when used for long periods of time, but this circuit I’m designing is something that’s meant to stay on for days at a time.

8

Devices such as the H11F1 are available which do exactly this, but their output capacity is quite limited. External transistors add zeros to the voltage and/or current that can be switched.

That’s always the $47 question. There are devices such as the ADUM5230 that build in a power isolator to address the issue. A resistor & zener diode could be used to derive the drive supply from the 120V source, but one’s going to waste quite a bit of power that way; a cheap DC-DC converter would be more efficient. Options and tradeoffs about; ya pay your money and take your choice.

The gate drive voltage is reduced by the diode’s Vf in that case, and turn-off will tend to be slower since the discharge would occur only through the gate-source resistor. Neither is desirable from a performance standpoint, and the safety folks wouldn’t like the idea that there’s no actual isolation. But within the confines of one’s own lab it’d be an option should one care to choose it.

Seems like a valid option so far, but you’ve not mentioned what FET you’re using, desired switching frequency, etc. etc. Note that you’d probably want to move to a 12-ish volt gate drive supply, since the output of the device is only specified to operate down to 10. Check specs on your FET regarding Vgs(max).

Are you referring to the below? Don’t worry about it; it’s speaking on time scales of years (not days) and in context of harsh environmental and electrical operating conditions that you’re unlikely to be imposing.

image
image785×87 5.49 KB

1 Like
9

Yeah, Vgs for the FET has a max of 10V. Here’s the spec sheet: HRISD017-6-66.pdf (widen.net)

Can you recommend an alternative to TLP152?

Also, we’ll may just use an external 12V power supply if said alternative needs it. I need eight MOSFET’s for the final setup, so it’ll be a hassle and I’ll have to split a wire eight times, but that’s manageable.

Another important question regarding the TLP152: While considering a similar product with different voltage ratings, how would you connect a 12V power supply? Obviously, the positive terminal is the supply voltage, but that’s an in without an out, unless the ground 12V supply is also what would be the ground for the TLP152 and the FET source and the 120V supply’s ground.

Good to hear. I don’t think it should be an issue then, but there’s a chance it could. I’ll reply again if it is, but this motor is to raise the platform in a Bridgman furnace, which can get up to anywhere between 800-1100 degrees, depending what we set it to. The motor is external, so I highly doubt that this is a problem, but I’ll double check to make sure there’s never going to be any contact with any of the exposed parts of the furnace that get scaldingly hot, and even if it does, we’ll probably make something to hold it and keep it safe, anyway.

10

IS480P should work down to about a 5V supply.

Interesting choice; did somebody find a stash of the things when they were cleaning out the storeroom? Might as well smoke 'em if ya got 'em, but I’d choose differently if the intent was to pick some up at the quick-e-mart on the way home.

For one motor? Seems like there might be an easier way…

Sounds about right.

Wear mechanisms in electronic devices tend to follow an Arrhenius relationship, doubling in rate for every 10°C increase in temperature. You’re much more likely to have problems for other reasons before opto degradation becomes an issue.

1 Like
11

Nice find. Thanks. You’ve been incredibly helpful!

The choice was deliberate when I asked my postdoc supervisor to order some. It would’ve been fine if the Arduino were connected to the FET directly, since the Arduino doesn’t put out more than that. I don’t really know why that’s deserving of disdain.

Well, I was omitting details… There’s two furnaces, two motors each, with an armature and a field circuit for each motor. That’s eight different circuits in the whole setup.

Other than the above, this is likely it, so thank you a ton for your help!

12

Availability mostly; like OK soda, they’re not a thing one would expect to find at the average corner store these days. Beyond that, tech has advanced a bit in the last 30 years. But function is function, and you’re free to choose whatever device you want.

That would indeed be one of those mundane details that alters the scope of things just a tiny wee bit… I sense potential for simplification/improvement even so, but it’s your project, your choice, and your learning experience. It appears you’re using a separately-excited DC machine; as an undergraduate good documentation for such would likely appear as a daunting mass of charts and tables. Be thankful if that’s the case, because everything in a datasheet that one doesn’t understand is a cue that there might be something that one really ought to be taking into account.

1 Like