A Great Scott video (Electronic Basics #23: Transistor (MOSFET) as a Switch - YouTube) came up in my feed and I took a look over lunch today and the section at 2:24 caught my eye - he was able to turn on the n-channel MOSFET via electrostatic discharge, adding a 10K pull-down on the gate solves this issue.
Dustin activates the gate by connecting the gate to a battery via his body resistance - showing the low threshold for activating a MOSFET as opposed to Great Scott who uses pure electrostatic voltage to activate the gate
in Dustin’s video, the MOSFET remains on, even after the gate voltage is removed, whereas Great Scott’s turns off.
The solution to both cases is to add a 10k pull-down to the gate.
I have been unable to reconcile the two effects shown in the videos. I intuitively get Dustin’s video - he is able to get enough charge into the gate to open the channel and as there is no path for the gate capacitance to discharge it remains open.
In Great Scott i can appreciate the ES discharge is able to activate the gate, though I would expect the ESD to destroy the insulation on the gate (but it didn’t for some reason) - and with no discharge path, I’d expect the MOSFET to stay on
The pull down as a way to ensure with no input the MOSFET is off is fine, not sure how it’s solving the ES issue in Great Scott though
I’m in the Australian outback, and ambient relative humidity can fall to around 5%, so electrostatic discharge or electrostatic fields are something I experience regularly.
I’m able to turn on the open gate of several devices (2N7000, BUK455, MTP3055E) by waving my hand above the gate.
Electrostatic discharge does not always cause damage or misbehaviour. That’s one of the frustrating aspects of managing it. It can be hard to reproduce the source of a chain of faults.
The other frustrating aspect is that the damage can occur yet function of the device continue for some months, enough to make it dead on arrival or fail during early or regulated warranty period. This has yield impacts on manufacturing, though it occurs as returns or contract issues.
Consider the dynamic system. With a pull-down, in both cases the voltage rise from the input is substantially reduced, and so the gate does not reach the level of activation, or does not remain activated. A pull-up resistor will also function to prevent an otherwise open gate from being dragged to ground by an accidental input, provided the designer expects the gate to be driven low instead of driven high by the circuit.
The key message to me is that gates should always be driven somewhere, either by a totem-pole output earlier in the circuit, or a switching element and a pull-up/down resistor.
aussie, aussie, aussie! (im an expat aussie in europe these days, currently experiencing 95% humidity and nearly freezing temperatures, quite the contrast)
i must admit, ESD in detail is a bit hand-wavy for me; I use “best practices” and rely on others expertise, so an area I’d like to dig into a little more and understand better. But I guess there are a lot of complex factors as to when damage would occur so mitigation of the worst case scenario is preferred - prepare for the worst and hope for the best
my assumption here is that you are capacitively coupling to the gate? so, the charge in your hand is enough to cause a separation of charge in the gate, leading to a potential difference high enough relative to the (grounded) source that the channel opens to allow conduction
considering this circuit (which I think replicates the basic idea from both videos)
the 2N7000 has a Max Vgs (continuous of +/-20v; pulsed of +/-40v) and a Vgs(th) between 1v and 2v, so a logic level N-Channel MOSFET. No problems driving from a GPIO pin
(the below may be widely inaccurate, but is my current understanding - please correct me where I am wrong!) → this may also be a case of rubber-duckying
we have three cases (1) Great Scott - direct contact to the gate, no power supply; (2) Dustin - power supply connected to the gate via human (3) Quozi - indirect contact with the gate
the gate to source is high-impedance, but capacitive on the order of 50pF
ESD is on the order of 4KV (though could be substantially higher) and producing several amps of current (though again could be higher) but for extremely short periods of time, and a low total amount of charge transfer
given the above
For some reason in Great Scott’s case, the ESD conditions (voltage/current/charge transfer) is not sufficient to damage (visibly and immediately at least) the MOSFET; he may also be using a FET with higher tolerances than the 2n7000 here. Though not explicitly stated it appears he is using an IRFZ44N, which has a max Vgs +/-20V.
in all three cases though, there is a voltage applied to the gate (Vesd) - either directly (Great Scott/Dustin) or capacitive coupled (Quozi).
due to the high impedance of the gate, assuming it doesn’t breakdown, no current will flow, that ESD voltage allows enough charge to flow into the gate to charge the gate capacitance enough such that the FET turns on. so far so so good! The lack of conduction path though would lead to the FET remaining on, as the gate capacitance can’t discharge anywhere (i.e.: the result seen in Dustin’s video; the implication from Great Scott is that there is a path somewhere for the capacitor to discharge to allow the FET to switch off (leakage?))
the solution becomes
taking the same conditions, if the gate impedance is sufficiently high, the current will flow through R2 to ground, for a very brief time (the time it takes for the ESD to discharge), momentarily raising the gate equal to Vesd, which should cause the MOSFET to very briefly turn on, though perhaps it is just too fast to see in a video. Once the Vesd has discharged to 0, there is no longer any charge and so the FET is off and pulled down to ground via the pull-down
in Great Scott’s video, as long as he is holding the gate, the MOSFET is on; indicating the Vesd he is supplying is not sufficient to destroy the MOSFET, but is supplying sufficient charge to activate the gate
in Great Scott’s video, he repeatedly touches the gate activating the MOSFET multiple times; when he removes his finger the MOSFET switches off (with no pull down), which I find a little surprising - the MOSFET gate must be charging when he touches the gate, and discharging (somehow) when he removes his finger. I am thinking he must be coupling in some energy that is not just ES (like walking on a carpet), rather he is coupling in some other lower voltage source as it is continuously present and not high enough to be harmful.
in Dustin’s case though, we have something like this
(google seems to suggest that the hand-to-hand resistance is about 1K)
in a normal MOSFET circuit we might add 100ohm series resistance to the gate to control in-rush current, so this is an order of magnitude higher, but would have a similar effect; combined with the parasitic capacitance we would have an RC circuit which would slow the rise time (and so also reduce any potential ringing/oscillation) - but more or less we have a voltage divider here from the gate’s perspective - assuming he was using a 3V coin cell battery, with that setup we’d see around 2.7v at the gate, which is above Vgs(th) for that FET.
now, in reality and practicality, what we want is for the MOSFET gate to be at a known value until we want to change it - for example:
thus the pull down is serving two purposes
forcing the gate off in any unexpected states where the gate might be floating or high-impedance (GPIO as input at reset for example)
giving the gate a path to discharge (only of concern if switching speed is a concern)
the pull down should be sized appropriately for the priorities of the design.
in terms of ESD or other coupled noise, there are better ways if these are of concern; a shunt capacitor, TVS etc. The take away from the videos should be that it is incredibly easy for stray voltages to influence a MOSFET gate if left floating (ESD is one way that this could happen, but coupled noise from other traces, RF/EMI noise etc are others - ESD would have it’s own problems in a worst case scenario of course)… but basically don’t leave a MOSFET gate floating
Well, yes, but it depends on what you want as circuit designer. Some touch sensors rely on this property. A sufficiently large value pull-down, such as 10 megohm, may be used, along with protective diodes to supply and ground. Body diodes on microcontrollers. RF can certainly be an interesting trigger; if the gate is in proximity to a low powered transmitter, or within cooee of a high powered (amateur) transmitter. Perhaps Great Scott had a HackRF One transmitting nearby, or a bluetooth mouse, or anything. Just can’t know.
1) sorry! damn fonts! 2) I am rather fond of abusing SED in this way myself! must be an aussie thing… or a life time spent as a sys admin…
but then, if I understand you correctly you don’t see anything wrong with my reasoning or understanding of MOSFET circuits and design principles - the basics seem ok (I know enough to be dangerous at least!); but yes, some application specific cases (that I am yet to come across) would require some more in depth re-neducation. Honestly when I saw those videos I questioned how well I understood the topic, which is when i stop and spend days reading!
I don’t see anything wrong with your description of the videos or your later reasoning and understanding about MOSFETs. No significant conflict. There are uses for electrostatic fields though, so it isn’t quite right for me to say “always pull-up or pull-down”.
no indeed, and that is an error on my side; the context of this is really, MOSFETs as switches
and as i’ve learned the hard way, in electronics (and many fields) there are no absolutes - at best these are rules of thumb; as you say, with a push-pull/totem-pole output you can get away without it. or the old favourite practice of decoupling vs muntzing (it’s fine until it doesn’t work!)
Is it really ESD? Ambient AC coupled to his body? It only takes a few V of AC to turn the gate on/off - without a scope trace, we don’t really know what the on state looked like. The “tell” on this, I think, is that if you watch the video at 1/8 speed, the LED sometimes turn on, sometimes turn off as he touches and releases the gate, but when his finger is on the gate, the LED appears to be about half as bright as the static on state when the gate is at some voltage.
Also, unless his body and the circuit shares the same ground, the FET circuit is floating (-ish). So being at “2000V” doesn’t necessarily mean the FET sees that 2000V. Ask any bird sitting on a powerline…
@ToyBuilder thanks! the fact that you thought to (and bothered to) slow the video down to 1/8 is dedication to the cause!
I have looked at it myself now at 0.25x and I see what you mean. there are occasions where he is clearly touching the gate, but nothing happens. Cases where, as you say, he touches the gate and it’s half power before he releases his finger and it’s full power - before touching it again and it drops to half power and oscillating like that as he touches/removes. It actually looks remarkably like Dustin’s video. Now of course there are cuts and edits here. But actually at 0.25x its not as strange as I thought.
You make a good point on different grounds as well.
But I can agree, that in the video, it’s very hard to tell exactly what is going on, but I will then say, I don’t like the blanket “even electrostatic voltages of my body can turn on the loads”
I feel that statement is ripe for misunderstanding. A better wording might be preferable as his point is basically, don’t leave the gate floating, anything could happen and it doesn’t take much to cause unintended effects. In any case, I’m fairly happy to have consolidated my understanding on the topic
An aside, this reminds me of my uncle who was a system controller at one of the two inbound grid substations in Sydney, where the long distance transmission lines arrive from power stations to be transformed to lower voltages.
When taking lunch, watching the pelicans, they’d happily sit on the bus bars, clear of the ground and the pests thereon, but they soon learned not to peck each other from separate phases. They learned to line up along one phase at a time.
(im an expat aussie in europe these days, currently experiencing 95% humidity and nearly freezing temperatures, quite the contrast)
considering this circuit (which I think replicates the basic idea from both videos)
Perhaps Great Scott had a HackRF One transmitting nearby, or a bluetooth mouse, or anything. Just can’t know.
1) sorry! damn fonts! 2) I am rather fond of abusing SED in this way myself! must be an aussie thing… or a life time spent as a sys admin… 
(it’s fine until it doesn’t work!)
- I had heard about cows having unfortunate incidents due to SWER, but hadn’t considered birds on different phases interacting!