MOSFET with integral pulldown resistor?

Thread Starter

wayneh

Joined Sep 9, 2010
17,505
Why aren’t there any? Or maybe I just haven’t seen one.

I get that there are situations where an internally fixed value would be in the way. It seems to me, though, that an integral 10K gate-source resistor would be a big convenience for a large part of the market. It might even ensure that less MOSFETs get destroyed inadvertently before they’re mounted on PCBs. It strikes me as odd that this isn’t offered as an option.
 

WBahn

Joined Mar 31, 2012
30,303
Why aren’t there any? Or maybe I just haven’t seen one.

I get that there are situations where an internally fixed value would be in the way. It seems to me, though, that an integral 10K gate-source resistor would be a big convenience for a large part of the market. It might even ensure that less MOSFETs get destroyed inadvertently before they’re mounted on PCBs. It strikes me as odd that this isn’t offered as an option.
Since I've only designed IC chips, I don't know how directly this would apply to discrete device fabrication, but resistors tend to be very large devices relative to the transistor (again, that's for IC transistors, I would expect a discrete transistor to be considerably larger). Also, it may require additional masks and processing steps that would significantly increase the cost of fabrication.

Also, remember that one of the advantages of MOS transistors is that there is no DC gate current. These devices would have nonzero DC gate current.
 

Thread Starter

wayneh

Joined Sep 9, 2010
17,505
It's hard to believe that if size and cost are primary issues, that the integral resistor wouldn't win as the more efficient option. Otherwise the MOSFET manufacturer is just pushing the size and cost issue off onto their customer. In any business I was involved in, making things cheaper and easier for your customer was what we strove for. Yes, we wanted to reduce our own costs as well, but not if we couldn't create value for the customer.
 

crutschow

Joined Mar 14, 2008
34,845
Many MOSFET applications don't need a gate pull-down resistor as they are driven by a push-pull output.
And some applications, such as using the MOSFET as a simple timer, you don't want any added gate resistance to ground.
 

Thread Starter

wayneh

Joined Sep 9, 2010
17,505
Many MOSFET applications don't need a gate pull-down resistor as they are driven by a push-pull output.
And some applications, such as using the MOSFET as a simple timer, you don't want any added gate resistance to ground.
I get all that, and I suppose the manufacturers have a better idea of how their customers use their products. But there are an awful lot of MOSFETs with external pulldown resistors out there and it seems like an opportunity.
 

crutschow

Joined Mar 14, 2008
34,845
But there are an awful lot of MOSFETs with external pulldown resistors
Depends upon how much is "an awful lot".
They are often included on hobby designs to avoid the MOSFET turning on if the gate is accidentally opened.
But I doubt that may commercial boards add those, since few gate drive circuits require a pull-down.

I think if there were an actual significant market for such a device, someone would build it.
One problem is that the manufacturer would likely have to build two versions of the device, one with and one without the resistor, which drives up costs. It's that cost versus the cost of the added resistor on the PCB for the few designs that require it.
 

Thread Starter

wayneh

Joined Sep 9, 2010
17,505
I think if there were an actual significant market for such a device, someone would build it.
Presumably that's the case. Sometimes manufacturers miss opportunities because they focus too much on their internal manufacturing concerns and pay little attention to customer or market needs.
 

WBahn

Joined Mar 31, 2012
30,303
Its all about what process is used for doing MOSFETs and optimizing die size.
Great ratio accuracy can be achieved, and now thru laser trim same for absolute
value.

Here is some info on widely used R's in IC fabrication -

http://ims.unipv.it/Courses/download/DIC/Presentation02.pdf


Regards, Dana.
But those are resistors used in IC fabrication. Discrete device fabrication is significantly different and has a lot fewer options. I don't know what on-die resistor options would even be available to them without changing the fabrication process itself, particularly for high-valued resistors. In an IC those are often made using either polysilicon that has not had something like a salicide implant or a well resistor. I don't think those are options on most discrete transistor fab processes.
 

dendad

Joined Feb 20, 2016
4,503
I prefer to add a pull down resistor as then the value can e selected. If it was internal, then you are stuck with it.Resistors are cheap and small so I don't really see any advantage of having one internally.
 

Thread Starter

wayneh

Joined Sep 9, 2010
17,505
But those are resistors used in IC fabrication. Discrete device fabrication is significantly different and has a lot fewer options. I don't know what on-die resistor options would even be available to them without changing the fabrication process itself, particularly for high-valued resistors. In an IC those are often made using either polysilicon that has not had something like a salicide implant or a well resistor. I don't think those are options on most discrete transistor fab processes.
Interesting. I was assuming it was a simple thing to do and the lack of the resistor was just a marketing decision. If there's a 'significant' manufacturing cost (ie. more than a cent), it makes sense to leave it out.
 

WBahn

Joined Mar 31, 2012
30,303
I prefer to add a pull down resistor as then the value can e selected. If it was internal, then you are stuck with it.Resistors are cheap and small so I don't really see any advantage of having one internally.
There are two very disparate use cases. For the hobbyist or low-volume guy, the flexibility of deciding whether to use a resistor and, if so, what value to use might be more attractive than reducing component count or board size. For the consumer-volume guy reducing component count or board size enough to shave a cent or two per unit might be a huge deal. But, of course, if the component that lets you reduce the component count and board size costs enough more to negate those savings, that's equally as huge a deal.

Not surprisingly, device fabricators tailor to the needs of volume costumers and everyone else figures out how to best deal with the resulting decisions in their own slice of the pie.
 

danadak

Joined Mar 10, 2018
4,057
There are two cases in the semi biz for primary cost structure (this will be a horrible oversimplification) -

1) Die size, where package costs are subsumed by large die/process/test costs.

2) Package cost, where die is essentially of no value. Many analog devices and simple logic
fit this. Many occasions I heard mangers complain what they were shipping was metal
and plastic, not much else, in the analog group.

Semi manufacturing, the reason price curves generally have declined, are because, as stated
earlier, the largest customer drives the business, the creation of the part, and its ultimate
costs. A good example some years ago the 10/100 phy. One US customer drove this demand
and price curve, the rest of the customer base fell in behind when the standard was accepted.

There used to be a lot of variation in part offerings, but that declined over the years to technical
advancements, inventory management, and wasted time managing small deviations in a part.
You can still get some manufacturers to screen for special conditions, but volume must be very
large. Putting a R on a MOSFET, so what value, what precision, what power capability.....so
many variations. To buy and place on a PCB a R is < 1 cent these days (my price dated). So
why build a large assortment of parts with all these variations.......you will be 6' under before
you make any $ on that.

Power MOSFETS for sure have very optimized processes. But in the low end of that
are intelligent devices, with lots of internal Rs, in the range of single digit amperes.
Some of Infineons intelligent power stuff comes to mind. And there is activity on
hybrids, multiple die solutions, but fairly narrow in total market.


Regards, Dana.
 
Last edited:
There is one more reason why this is not really a "protection" option.

The high voltages which may appear on the human body have a electrostatic energy behaviour; lets say to be more easy to understand, if you touch the gate of a transistor and you are chaarged at a high voltage say 2-5kV, there is nothing a resistor can do against. The pulse of energy has a high amperage even if the time of discharge is of nanoseconds. At lower voltages and energies, a zener diode will be effective, but not a resistor.

Still, at even lower voltages (50-500V of static charge on human body), more efect of protection has the internal gate capacitance of such a transistor, which in normal situations is fairly enough to protect the device; the human is touching the gate lets say before touching the source or drain pins, but the energy transfered will only charge the gate with a small voltage against the other pins, because the capacity of the transistor body in free air to ground is insignificant, probably in the order of 0.1-0.2 pF. The capacitances apear in series, Gate-Source being of nanofarads, and body of transistor less than 1 pF to ground...how much voltage will apear on the gate against the body? Maybe 0.05-0.5V, even if the human is charged at 50-500V. Simple calculating the distribution of voltages across a capacitive divider...

Situation is worse if transistor is inserted in a board, without zener diode already mounted, or other components to absorb the energy, and the board has a leakage to ground by any means; for example being connected through an osciloscope.. in that case, the energy is distributed on a capacitive + resistive divider between human body, the transistor gate-source capacitance, the source/drain to board capacitance or even resistance if it was already soldered on these terminals, and the capacitance/resistance to ground of the oscilloscope. And the voltage on the gate in this situation may be high enugh to damage the gate... I always take care when insert low power transistors in boards to solder them knowing what happen and how to touch them, if I'm lasy enough to not use antistatic bracelet; sometimes I use a corcodile clip on engagement
ring, which I connect it by this means with the soldering iron, and so the "situation is discharged". :D
 

vanderghast

Joined Jun 14, 2018
67
... an integral 10K gate-source resistor would be ...
Even with a built in capacitance as low as 5 pF, that gives an RC of 50ns, which is relatively long in comparison with the actual setting down time of 20ns for the 2N7000 as example. (And that 50ns is just ONE RC, not 3 or 5RC).

Would almost be preferable to connect the Gate and the Source of an NMOS not with a resistor, but with a PMOS (or other) so the PMOS is off when the NMOS is desired on, and which turns on, with its low R_DSON, when you want to supply a path to unload the gate capacitor at the source of the NMOS. But again, that is generally unnecessary since whatever "command" the voltage at the source is generally already able to "sink" V_GS(TH) by itself, implying "zero" Ohm. No?
 

Involute

Joined Mar 23, 2008
106
Here's one I use in a low-volume product. SOT-23, protection up the whazoo, and "only" $.56, qty. 1. Works like a charm. I'd inadvertently omitted a gate pulldown in my initial design and noticed occasional random power-on triggering of the solenoid my previous MOSFET was driving (controlled from an AVR IO pin). Came across this one (Diodes makes several like this) looking for just this capability.
 

Thread Starter

wayneh

Joined Sep 9, 2010
17,505
Here's one I use in a low-volume product. SOT-23, protection up the whazoo, and "only" $.56, qty. 1. Works like a charm. I'd inadvertently omitted a gate pulldown in my initial design and noticed occasional random power-on triggering of the solenoid my previous MOSFET was driving (controlled from an AVR IO pin). Came across this one (Diodes makes several like this) looking for just this capability.
Nice. That's good part to recommend to noobs.
 
Top