Driving power mosfets like signal mosfets

[zinn]

Hello, to all, I have a few questioms here. I am still new to MOSFETs and stuff, however learning abojt them, and reading the datasheets I have here, I wondered if I can use power mosfets like the irf540 or irf9540 as it is a signal mosfet for switching even if I have to severely derate it, I mean, I know it wasn't what they intended it to be used for, but, the question is, can I? according the datasheets on the typical output characteristics and Vgs(th) it should be possible, also, from what I understood of my "research" gate charge is affected by Vds and lower Vds means lower gate charge, unless I am mistaken and if such is the case by all means please tell me why.

So, here are the conditions: I have an IRF540n with souce connected to ground, drain connected to 5v though a 10W 12V car incandescent lamp and gate is to be switched between 5v to turn on and ground to turn off. again, I know this is not how it is supposed to be done, but I do not know why people do no do it if, in theory it is possible.

My point is, Vgs is going to be 5v(ok, a bit too close of the max Vgs(th) of 4v) but the graph shows that with 5v Vds it can conduct quite a bit of current, so why have I not found cases of this? I mean, I had to create this account just because the best answers I got was from AI, but AI is dumb, because it was suggesting Vgs of 10V, but Vgs of 10v and Vds of 5V would not make it work in the ohmic region/linear/triode ? and on the plus side would defeat the whole purpose of using a power MOSFET as a signal FET(or whatever that is supposed to mean, because I do not know what other FETs are called, Ihave read "low power FETs, signal FETs, small FETs") I do not expect it to work at crazy high frequencies if I do decide make an oscillator with it(assuming it is possible) all I am asking is:

Is it possible? regardless of the answer, why?

Assuming it is, does it damage the part? why? should I expect excessive heat dissipation even if maximum current in the circuit to less than a tenth of an amper?

Assuming it is, why do people not do it? even as a form of hacking, as in doing what should not be done just because despite it, it works.

As for my motivation, where I live MOSFETS are not the cheapest things around, not expensive but when 20u of irf9540 come cheaper then 10u logic level N channel mosfets, your mind start to wander and you start questioning what is real and what is possible and if you already have what you need but do not know it.

if you got until here, hey man, thanks just for reading.

[Andy Chee]

So, here are the conditions: I have an IRF540n with souce connected to ground, drain connected to 5v though a 10W 12V car incandescent lamp and gate is to be switched between 5v to turn on and ground to turn off. again, I know this is not how it is supposed to be done, but I do not know why people do no do it if, in theory it is possible.

People do this all the time and it works.  What are you talking about?

You will have to post more details and a schematic for proper analysis.  But the general idea is ok.

When you say "not how it is supposed to be", how do you think it is "supposed to be"?  This is the missing detail, otherwise we have nothing to compare and contrast with.

[Berni]

There is nothing special about power MOSFETs other than they are bigger and tougher to handle the power and provide a way to cool them.

The reason people use signal FETs is when they need them to be physically small to use less area on a space constrained PCB. Other reason is that they might have some more demanding requirements for its specs (that power FETs can't provide). Perhaps they need a very low leakage current, maybe they need low gate capacitance, maybe they need to switch it very fast, maybe they need a weird response like a depletion mode FET...etc

The reason that people tell not to use a power MOSFET as an amplifier is that most of them are not rated for linear operation. They still operate just fine in the linear region, however they can't handle the rated power in that region. The reason you pick a power MOSFET is usually because you are dealing with high power. But if you linearly drive a switching MOSFET (optimized for low RdsON) at high power you get unequal current sharing inside the silicon structure, so some parts get hotter, this is made worse by the negative temp co that these FETs typically have, causing hotspots in the silicon die to go into thermal runaway, eventually making the transistor blow up. This is solved by either buying the more expensive linear rated power MOSFETs or by de-rating a switching MOSFET to much less power than what the datasheet says it can handle. Since if it doesn't get very hot then it can't thermal runaway.

[pcprogrammer]

My point is, Vgs is going to be 5v(ok, a bit too close of the max Vgs(th) of 4v)

The Vgs(th) of 4V is the worst case threshold voltage on which some will turn on. The minimum of the spread is 2V. With your 5V on the gate you can rest assure that it will turn on. The maximum Vgs is +/- 20V.

A higher Vgs makes it turn on even more, in the sense that Rds will be on or below what is the specified max. In the datasheet you will see that the test condition for this value is at Vgs = 10V and Id = 17A (Pulsed according to note 4) So the AI you asked is not entirely dumb per se.

Berni already wrote about the linear region, but with 5V on the gate and the source at ground, you are outside the linear region, so not an issue here. The transition between ground and 5V on the gate takes it through the linear region, and can cause problems if this is done to slow. Depends on the drain load of course.

[zinn]

Really? I thought it was some obscure thing, I looked around on the internet, I know it feels a waste to use something like a power mosfet to drive a few mW even a few W at such low voltages, but I could not find examples or materials on it, just peolple on the stackexchange saying it could not be done because Vgs is barely above the maximum rated threshold of 4v. But enough ranting, here is what I am talking about:

the schematic of what I described in my starting post and tge graph of output characteristics and the line circulated in red about how it behaves with Vgs of 5V which seem to me it is plenty enough for conduction, of course, it might not be good enough to drive tens of amps but I think I can get away with driving signals of a few mA perhaps a few hundreds of mA

on about the schematic in question, I wonder, is 5v Vgs good enough for this schenario? or do I need 10v or something close to it? and if yes, would it make a Vgs>Vgs(th) and Vds<Vgs scenario, turning it into linear operation mode?

Edit1: That was my reply to Andy Chee, and while I was typing Berni an pcprogrammer replied

to Berni, I never thought of that, I have read infineon linear fet operation, though I do not intend  on driving these in the linear region, just switch at low voltages, these are HEXFETS(tm) and they do not like that, plus, there is no DC line on the SOA chart, but your clarification is much appreciated and this information cleared up some earlier questions of mine(not mentioned on this post) about linear operation and de-ration as well as why people use irf540 and irf9540 on audio power amps even if they do not have a DC line on the SOA chart but they do with 10 pairs in source follower configuration! thank you very much!

and to pcprogrammer,

Thanks for clarifying this Vgs issue, I mean, the AI said it would not work because with such low Vgs even though it is above the max Vgs(th) it would not work properly and might not turn on and might be on the linear region, but the source was some post on allaboutcircuits forum where a novice was asking about using source follower and it was not working and the explanation was, confusing, it seems that there is a consensus that Vgs bellow 10V will not work at any cost just because it is not optimal for switching and not offering the lowest Rds(on) which is fine, but for my application I am fine with not having the lowest Rds on, as long as it will not damage the mosfet overtime. I have some FPQ rated for DC operation(SOA), but IRFs are so cheap, and I can get P channel for cheap as well different from other no SMD/SMT p channel fets or Logic Level FETs that are too expensive to just "play with" and it is nice to know that de-ration is an option, since none of it was mentioned in my early "research". thank you very much!.

Edit2: typos(I was on the bus) and again, thank you all for your time and attention on this thread. to me this means a lot, feels like a weight has been lifted from my shoulders.

[MrAl]

Hello, to all, I have a few questioms here. I am still new to MOSFETs and stuff, however learning abojt them, and reading the datasheets I have here, I wondered if I can use power mosfets like the irf540 or irf9540 as it is a signal mosfet for switching even if I have to severely derate it, I mean, I know it wasn't what they intended it to be used for, but, the question is, can I? according the datasheets on the typical output characteristics and Vgs(th) it should be possible, also, from what I understood of my "research" gate charge is affected by Vds and lower Vds means lower gate charge, unless I am mistaken and if such is the case by all means please tell me why.

So, here are the conditions: I have an IRF540n with souce connected to ground, drain connected to 5v though a 10W 12V car incandescent lamp and gate is to be switched between 5v to turn on and ground to turn off. again, I know this is not how it is supposed to be done, but I do not know why people do no do it if, in theory it is possible.

My point is, Vgs is going to be 5v(ok, a bit too close of the max Vgs(th) of 4v) but the graph shows that with 5v Vds it can conduct quite a bit of current, so why have I not found cases of this? I mean, I had to create this account just because the best answers I got was from AI, but AI is dumb, because it was suggesting Vgs of 10V, but Vgs of 10v and Vds of 5V would not make it work in the ohmic region/linear/triode ? and on the plus side would defeat the whole purpose of using a power MOSFET as a signal FET(or whatever that is supposed to mean, because I do not know what other FETs are called, Ihave read "low power FETs, signal FETs, small FETs") I do not expect it to work at crazy high frequencies if I do decide make an oscillator with it(assuming it is possible) all I am asking is:

Is it possible? regardless of the answer, why?

Assuming it is, does it damage the part? why? should I expect excessive heat dissipation even if maximum current in the circuit to less than a tenth of an amper?

Assuming it is, why do people not do it? even as a form of hacking, as in doing what should not be done just because despite it, it works.

As for my motivation, where I live MOSFETS are not the cheapest things around, not expensive but when 20u of irf9540 come cheaper then 10u logic level N channel mosfets, your mind start to wander and you start questioning what is real and what is possible and if you already have what you need but do not know it.

if you got until here, hey man, thanks just for reading.

Hi,

The IRF540 is one of the older MOSFET devices, probably one of the first, which came out roughly around 1980 I think.  It is not a logic level device meaning a 5v gate to source drive is not a good idea.  The Rds resistance could be higher than expected and with that the temperature rise higher than expected, and with that the Rds resistance could be even higher.

The 4.5 gate voltage spec usually means it requires 10v to properly switch 'on'.  A 5v gate voltage may not be enough.  A true logic level MOSFET will have a gate threshold much lower, like half of that, which would even allow switching from a microcontroller i/o pin.

It's not worth wasting the time, unless you already have a lot of them and don't care if the Rds is higher, and the DS current is low.  You'll also have to test very carefully, and probably have to make sure the input goes all the way up to 5v and not just 3.5v like some logic families might provide as a max high level voltage.

Good luck with it.

[Berni]

Yes 5V will turn it on, but it is fairly on the edge.

You can see how much difference there is between the 5V and 4.5V curve, this is because they are getting into the linear region. Yes the FET is turned on and will pass a few amps of current, but the RDSon will not be as good as it could be (so it will heat up more under load due to higher voltage drop). Additionally the signal you drive it might not be 5.000V, perhaps your supply voltage droops or the IO pin driving it has a bit of a drop. The curves also move around with temperature. So at -20°C the FET might end up well into the linear region at 5V gate at higher load currents. There is also a lot of manufacturing tolerance so each FET has a slightly different gate threshold voltage.

The FETs that are marketed as "logic level input" tend to have a more sensitive gate that turns on at 2V already, so when you put 5V or even just 3.3V the FET is well above the linear region with a safe margin.

For just hobby use it should be fine enugh, but it is not something you would design into a mass produced product as it doesn't cost much to simply select a more appropriate transistor and not have to worry about it not working to spec in certain edge cases. But for hobby use there is no harm in just trying it and if it doesn't work try some other transistor from the stash of parts you got laying around.

[Terry Bites]

You'll probably find that 5V is enough to bring the MOSFET near to saturation. A good enough result is often ok.
At Vgs=5V@1A its unlikely that an IRF540 will get hot even with a Vds 0.5V or so.

For a guaranteed sub 50mV Vds the simple solution is to use a level shifter that switches the full 12V on to the gate.
About $0.20 will cover the extra cost. Any small signal bipolar transistors will do. The value of the resistors are not critical, anything in the range of 1k-100k will work fine.
It may look a bit complicated bit its just two common emitter switches.
You can reduce the component count if you use a p-channel MOSFET.

[MariuszD]

As for my motivation, where I live MOSFETS are not the cheapest things around, not expensive but when 20u of irf9540 come cheaper then 10u logic level N channel mosfets, your mind start to wander and you start questioning what is real and what is possible and if you already have what you need but do not know it.

Are transistors like 2N7002, 2N7000, BS170, BS107, BS250, BSS138 available? They should be cheap.

[Terry Bites]

You list low power <500mA types, Your load is 1A, no?

How about using MOSFETs that are designed to be diven by 5V or less?
That would solve your problems with no extra parts. Dirt cheap.....
eg AO3400 AO3401 AO3402 AO3404 AO3406 AO3407 AO3415 AO3416 etc
https://www.lcsc.com/product-detail/C347475.html?s_z=n_AO3400

Aliexpress is giving them away!

https://www.aliexpress.com/item/1005008339792809.html?spm=a2g0o.productlist.main.9.2a8141b2laGwPQ&algo_pvid=849d796d-7ecd-4b93-bcae-dff062030d47&algo_exp_id=849d796d-7ecd-4b93-bcae-dff062030d47-58&pdp_ext_f=%7B%22order%22%3A%22-1%22%2C%22eval%22%3A%221%22%2C%22fromPage%22%3A%22search%22%7D&pdp_npi=6%40dis%21GBP%210.81%210.60%21%21%211.04%210.77%21%402101df0e17646066771414498e318d%2112000044654963246%21sea%21UK%213100120370%21X%211%210%21n_tag%3A-29919%3Bd%3A6be96178%3Bm03_new_user%3A-29895&curPageLogUid=znobeUBp00Hp&utparam-url=scene%3Asearch%7Cquery_from%3A%7Cx_object_id%3A1005008339792809%7C_p_origin_prod%3A

[magic]

So, here are the conditions: I have an IRF540n with souce connected to ground, drain connected to 5v though a 10W 12V car incandescent lamp and gate is to be switched between 5v to turn on and ground to turn off. again, I know this is not how it is supposed to be done, but I do not know why people do no do it if, in theory it is possible.

That's 1A of current, according to the plots you found it should work even with 4.5V gate voltage. But that's "typical" and may be a bit off due to chip-to-chip variation.

If you have those FETs, just try it. If not, maybe consider buying something else.

Gate threshold also increases at low temperature, but that's not a problem because in the cold the FET will first heat up to usual temperature and then work as usual.

[Refrigerator]

Hello, to all, I have a few questioms here. I am still new to MOSFETs and stuff, however learning abojt them, and reading the datasheets I have here, I wondered if I can use power mosfets like the irf540 or irf9540 as it is a signal mosfet for switching even if I have to severely derate it, I mean, I know it wasn't what they intended it to be used for, but, the question is, can I? according the datasheets on the typical output characteristics and Vgs(th) it should be possible, also, from what I understood of my "research" gate charge is affected by Vds and lower Vds means lower gate charge, unless I am mistaken and if such is the case by all means please tell me why.

I haven't read through the replies, but i'll add that large MOSFETs have large internal capacitance so they are not suitable for high frequency signals. Otherwise they work the same considering you gave enough VGS to turn them on.

I'll share some relevant experience i have with this exact topic. At work i was given an "unrepairable" welder that someone else had tried to repair. Turns out that a small MOSFET meant more as a signal MOSFET was replaced with a power MOSFET, namely the IRF740.
The result was that the inverter was not switching properly and had terrible efficiency. Replacing the power FET with a smaller signal FET solved the issue entirely and the welder works perfectly.

So, here are the conditions: I have an IRF540n with souce connected to ground, drain connected to 5v though a 10W 12V car incandescent lamp and gate is to be switched between 5v to turn on and ground to turn off. again, I know this is not how it is supposed to be done, but I do not know why people do no do it if, in theory it is possible.

My point is, Vgs is going to be 5v(ok, a bit too close of the max Vgs(th) of 4v) but the graph shows that with 5v Vds it can conduct quite a bit of current, so why have I not found cases of this? I mean, I had to create this account just because the best answers I got was from AI, but AI is dumb, because it was suggesting Vgs of 10V, but Vgs of 10v and Vds of 5V would not make it work in the ohmic region/linear/triode ? and on the plus side would defeat the whole purpose of using a power MOSFET as a signal FET(or whatever that is supposed to mean, because I do not know what other FETs are called, Ihave read "low power FETs, signal FETs, small FETs") I do not expect it to work at crazy high frequencies if I do decide make an oscillator with it(assuming it is possible) all I am asking is:

Incandescent bulbs have a very low impedance when they're cold, so make sure your 5V rail can handle the current spike. Beware that turning the FET off will cause a voltage spike at the drain if enough inductance is present in the circuit. To control current spikes you can add a diode+zener in series between gate and drain, so that any voltage spike would pass through the zener and turn the FET on ever so slightly. The FET then shunts the voltage spike that would otherwise damage the silicon. Or a simple diode from drain to the positive supply would work too.
To be honest i don't see any problem in yout application, just make sure you add a gate resistor simply to protect whatever is driving yout FET in case it fails.

[zinn]

To MrAl

I know, I am just trying to "wing it" with whatever I have on hand, I got my hands on some IRLZ44 last month, but I have a lot of IRFs 630, 640, 540 I think there are some 510 somewhere on some box, I will perform some tests and see how both of them compare, my issue is the P channel ones, they are not so easy to get a hold and I do not have so much disposable income to spend on a hobby for such few parts, but if I find it is absolutely necessary, then I might program my budget for it, despite that, I do not do much stuff with 12v or more but it feels like a waste having these parts lying around doing nothing so I will try and see how they perform as signal transistors if that is even a thing.

To Berni

luckily for me -20 C is not becomming a reality anytime soon, but I will keep in mind, interesting, thought they would become more conductive in the cold I do tend to forget that temperature is a parameter I have to manage or at least consider. And yes it is for hobby, I think no sane person would try and consider such approach for mass production, but I do not plan on using any IC to drive them, I might use them for oscillation that is the current goal.

to Terry Bites

You are correct but that is not the case, you see, I cannot level shift if the ceiling is Vcc that is 5v I am willing to consider a voltage doubler just to drive the gate tough, but I do not think it can be done with P channel fets (it would be very good if I could make -5v bellow ground) and I do not want to use floating power supply because I will be limited to optocoupler speeds which from what I read tops at around 80KHz, not bad for a buck or boost converter but I want to experiment with more speed and keep things simple even if inefficient, I just need it to work and it will be good enough, then I can think of next steps and better parts. mosfet gate driver was something I considered but to make your own is quite complicated from what I have read.


To MariuszD

yes, I have a bunch of 2N7000(to-92) and some 2N7002(smd) I never messed with the bs### and bss### but the idea was born from the fact that I did not found a to-92 p channel fet that os lile the 2n7000 I looked for popular transistors of low power of this kind and it returned the bss84 AO4409 and AO4407 which are pretty good, but the bss84 is smd and I want to experiment with to-92 first and the AO440# are pretty expensive, not expensive, but not the money I want to spend on a test, a whim or something I expect to burn due to me pushing its limits to see what it can do. not now, but in the future these are options and for special projects which warrant a budget, options I will be willing to spend on, do you have a popular, cheap even if it is 50 years old list of p channel fets that work on a logic level? or at least of similar ratings as the 2N7000?

to Terry Bites(2 reply after mine)

That is wonderful, I will consider these part numbers on the future and see if I can get from local suppliers in the near future, but my quest is for p channel and I would like PTH parts, but smd is acceptable however I prefer the larger packages since my soçdering skills are not that good yet. but to be honest since last week I became hell bent on it, I mean, I have the parts so why not? just for fun, nothing serious. but thanks, I have not considered smd options in a serious manner before and they are dirt cheap, if I am willing to use power fets to drive signals, why not chamge from PTH to SMD instead?

to magic

I would expect that, since I read that even of they come from the same wafer, there will be variations, So, the gate heats up before conducting?

to Refrigerator

Yeah, that beefy capacitance, but it can still get to the low, very low MHz, so I want to try on with that. and on your tale, do not worry, I understand what you mean and I intend to use only in standalone projects, after all I know that this large gate capacitance has been know to burn IO pins of ICs just because there was nothing limiting current to/from the gate, I myself lost at least 2 2N2222 and 555 for this exact issue in my youth, thank you for sharing this experience, I will remember it.

about the spike thing, in case of inductive loads, you are talking about the "flyback diode" right? or subber diode?  and on my example, for now it will by my hand and a 10k potentiomenter but I will make sure not to forget the gate resistor.


All and all I got good replies that gave me good ideas, means to properly drive the fet, do as I initially planned(I do not expect best performance, but as long as it works, I will be satisfied) or get parts to do the job, I will move forward with the plan because at this point I want to see what is possible, but I to do the other things I was considering I think that acquiring the proper parts will be the best thing to do for the sake of reliability.
 
Sorry, the text might be weird because I am on the phone so formating is a bit hard and watch out for typos, they will be everywhere.

[MrAl]

To MrAl

I know, I am just trying to "wing it" with whatever I have on hand, I got my hands on some IRLZ44 last month, but I have a lot of IRFs 630, 640, 540 I think there are some 510 somewhere on some box, I will perform some tests and see how both of them compare, my issue is the P channel ones, they are not so easy to get a hold and I do not have so much disposable income to spend on a hobby for such few parts, but if I find it is absolutely necessary, then I might program my budget for it, despite that, I do not do much stuff with 12v or more but it feels like a waste having these parts lying around doing nothing so I will try and see how they perform as signal transistors if that is even a thing.

Hi,

When you say 'signal' transistors, what do you mean by that.  Do you mean very light DS current flows?

If you expect a DS current of only 100ma it may work ok with Vgs set to 4.5v, but you have to be careful about what you drive it with because many logic families do not put out 4.5v as a high signal it may only be 3.5v, so that would require testing.

Luckily, if the MOSFET gets warm the Vgs requirement comes down, so it would actually turn on harder I think.

I'd test each MOSFET you have with loads like 100ma, 200ma and 500ma and 1000ma if needed, and various gate drives like 3v, 3.5v, 4v, 4.5v, and 5v, setting each gate voltage taking care to get it nearly exact.  You can then create a table for each MOSFET, even those that have the same part number: one table per MOSFET.  Should be interesting for you and everyone else here too.
Also note if the MOSFET under test gets warm or hot and include that in the tables.

[magic]

So, the gate heats up before conducting?

What I meant is that threshold voltage decreases with increasing temperature.

If the FET is cold and not turned off "sufficiently" for the current which wants to flow through the lightbulb, FET source-drain voltage will increase and power dissipation in the FET will increase as well. As the FET heats up, its threshold voltage will decrease allowing more current to flow at the same gate voltage.

If 25°C is enough to turn it on, it will heat up from whatever it was to 25°C and then stabilize.
If it's 70°C, it will heat up to 70°C.
If it's 250°C, you will be in trouble
But according to those plots, typical sample should work at 25°C and probably much less.

The important thing here is that in cases where gate voltage is the limiting factor, Vds decreases with temperature. It would be really bad if it were the other way around, and high temperature caused more Vds, more heat and higher temperature still. But that's not the case.

[zinn]

MrAl, hi Again, it seems you are on to something here, may I present you a day's worth of work, I don't know if my testing methods are good enough, but to me, these seem at least sensible I hope it become useful to somebody or at least some odd curiosity to be appreciated

Before I forget, the power supply I used is a 5V capable of spitting out 5A it is a SMPS, gave me some headache with ripple voltage that is why I had to use the caps gotta make or buy a linear adjustable PSU to get a clean signal whenever i get to do this, on this note, for all voltages, please consider +/- 0.2V margin, my multimeters aren't the professional stuff I am a hobbyist.

I'd post a pic of the test rig, but it is so ugly that my schematics and data results will have to do:
Te test Rig for the N-Channel, please pay attention to the voltmeters and ammeter, that is where I collected the results, the thermocouple is a type K cheap stuff but good enough, I used a small clamp to get it close to the FET right on the dain plate, no heatsink was used and on idle the thermocouple shows 27ºC(yes it is in Celsius, if there are any complaints I'll use Kelvin next time Haha)


Figure1: N channel test rig


Figure2: P channel test rig

At the end I included PDF and a spreadsheet file, I used xlsx format, but it was made on libreoffice, I figured somebody may play with it or make charts, not much data, but enough to make sense of things.

Conclusions

Aside from FET #06 that was somewhat not a good performer, but on the datasheet it states that Vgs(th) max is 5V so, that is that... I was actually surprised, I did not tested with different loads, but I did tested with something somewhat worst case scenario for what I intend to do, but I presume that under these conditions even a 500mA load would warrant a heatsink, even small, the lines where  the results are with "NULL" are cases where I deemed the test unnecessary due to previous results. Transistors #08, #10-#13 were the best performers, of course #10 is rated as a linear FET, I bought for a project never came through, #11-#13 are P channel, so long as Vgs is kept at -4.9 to -5V it will be fine, but -4.5 things go sour but FET #08(irf540n) goes fine even with Vgs of 4V which was surprising
 
Bellow is the data of one IRLZ44(Manufactured by Vishay)
Cg = 0.990nF
Rds = 0.11Ohm
Vth = 1.6V
diode voltage=462mV

Vds = 0.175 Id=1.10 T(30s) 33.2ºC @ Vgs 5V
Vds = 0.223 Id=1.09 T(30s) 36.3ºC @ Vgs 3.62V
 

Here it is interesting to see even with this low Vgs a LL FET still maintains "cool" and low Rds,,  too baad I could not find the rest of my IRF540N and I bet there were 3 more IRF95440 somewhere but I can find it later.

with this I can say  with confidence that it is possible to use power MOSFETS on 5V Vgs, but only on light loads and maybe medium  with a beefy heatsink and like many above said,  if it is for repair, it is not worth replacing a signal with power FET.  and if it is design, remember to account for the large gate charge  and gate capacitor or you  might just burn the IO pins on  your IC.

To magic:

I see, I wish more manufacturers would put this on the datasheets I remember seeing one with Vgs(th) lines for -55/25/175 ºC  but it's been some time, don't even remember what it was. but that is interesting, reminds me of the tubes,  but self starting!

Edit: sorry, had to delete and repost because I am still figuring out attachments in here.

[Berni]

The reason manufacturers don't put this datasheets is because nobody is supposed to use a FET for power switching in such conditions, they instead show you how much Vgs you need to really fully turn it on so that the FET can reach its rated low Rdson in all the edge cases it might find itself operating in. For temperature they usually do put some figures in the tables for the very edges of operation, stay away from those numbers and you are fine.

Similar to how you are not supposed to use power switching FETs for linear operation, hence they don't have a DC area in the SOA chart. That doesn't mean they don't work there, just that the manufacturer just doesn't grantee anything in that mode of operation. It is even possible that transistors made in one factory blow up sooner in linear operation as compared to the same partnumber made 3 years ago in a different factory. They don't test them in the factory for that because they were not supposed to be used for that in the first place.

When FETs are used in the linear region on purpose, then there is always some form of feedback loop that affects the gate voltage and tweaks it to keep the transistor at the desired linear operating point. Because of this the exact gate voltage does not need to be known beforehand, the circuit designer just needs to know the rough min and max range of Vgs that their circuit needs to be able to provide in order to keep the transistor happy.

[MrAl]

MrAl, hi Again, it seems you are on to something here, may I present you a day's worth of work, I don't know if my testing methods are good enough, but to me, these seem at least sensible I hope it become useful to somebody or at least some odd curiosity to be appreciated

Before I forget, the power supply I used is a 5V capable of spitting out 5A it is a SMPS, gave me some headache with ripple voltage that is why I had to use the caps gotta make or buy a linear adjustable PSU to get a clean signal whenever i get to do this, on this note, for all voltages, please consider +/- 0.2V margin, my multimeters aren't the professional stuff I am a hobbyist.

I'd post a pic of the test rig, but it is so ugly that my schematics and data results will have to do:
Te test Rig for the N-Channel, please pay attention to the voltmeters and ammeter, that is where I collected the results, the thermocouple is a type K cheap stuff but good enough, I used a small clamp to get it close to the FET right on the dain plate, no heatsink was used and on idle the thermocouple shows 27ºC(yes it is in Celsius, if there are any complaints I'll use Kelvin next time Haha)

(Attachment Link)
Figure1: N channel test rig

(Attachment Link)
Figure2: P channel test rig

At the end I included PDF and a spreadsheet file, I used xlsx format, but it was made on libreoffice, I figured somebody may play with it or make charts, not much data, but enough to make sense of things.

Conclusions

Aside from FET #06 that was somewhat not a good performer, but on the datasheet it states that Vgs(th) max is 5V so, that is that... I was actually surprised, I did not tested with different loads, but I did tested with something somewhat worst case scenario for what I intend to do, but I presume that under these conditions even a 500mA load would warrant a heatsink, even small, the lines where  the results are with "NULL" are cases where I deemed the test unnecessary due to previous results. Transistors #08, #10-#13 were the best performers, of course #10 is rated as a linear FET, I bought for a project never came through, #11-#13 are P channel, so long as Vgs is kept at -4.9 to -5V it will be fine, but -4.5 things go sour but FET #08(irf540n) goes fine even with Vgs of 4V which was surprising
 
Bellow is the data of one IRLZ44(Manufactured by Vishay)
Cg = 0.990nF
Rds = 0.11Ohm
Vth = 1.6V
diode voltage=462mV

Vds = 0.175 Id=1.10 T(30s) 33.2ºC @ Vgs 5V
Vds = 0.223 Id=1.09 T(30s) 36.3ºC @ Vgs 3.62V
 

Here it is interesting to see even with this low Vgs a LL FET still maintains "cool" and low Rds,,  too baad I could not find the rest of my IRF540N and I bet there were 3 more IRF95440 somewhere but I can find it later.

with this I can say  with confidence that it is possible to use power MOSFETS on 5V Vgs, but only on light loads and maybe medium  with a beefy heatsink and like many above said,  if it is for repair, it is not worth replacing a signal with power FET.  and if it is design, remember to account for the large gate charge  and gate capacitor or you  might just burn the IO pins on  your IC.

To magic:

I see, I wish more manufacturers would put this on the datasheets I remember seeing one with Vgs(th) lines for -55/25/175 ºC  but it's been some time, don't even remember what it was. but that is interesting, reminds me of the tubes,  but self starting!

Edit: sorry, had to delete and repost because I am still figuring out attachments in here.

Hello again,

That looks like some great testing there which is informative.

It looks like with some of them Vgs=4v could get funny, but it would have also been interesting to see the numbers with Vgs=3v and Vgs=3.5v, just to get an idea how they perform with some 5v logic families.

I suppose if you use it with a microcontroller i/o pin and you want to be sure to limit the current, you can use a series gate resistor with the tradeoff being reduced turn on and turn off rise and fall times, with the accompanying slower overall switching speed.

It must have taken some time to compile that chart, I have to say congratulations for getting that much done I know it's not easy to do and get all the measurements right.  That's the way to do it though, and with several transistors from various places you can get a good idea what to expect.

[zinn]

To Berni,

I assumed such, I mean International rectifier does not put DC in the SOA, but Fairchild does in many of what I seen and so did Samsung on the IRFS640. another interesting thing is that if you lookt at the datasheets there may be a hint of how well it works if you look the chart that relates gate charge with Vgs, in some the miller plateau start right before 5v and in some a bit after, of course it s hard to relate because the chart was made with much higher Vds and Id than what I did, but I cannot see how since I do not own an oscilloscope, I have a FNirst, but it is only good up to 200KHz, and has low resolution, maybe with a high enough gate resistor I could see the curve, however I do not see the point of doing so now, nor if there is any use for it. I might just try this today with one after work out of sheer curiosity.


To MrAl,

Hello Again!

Yeah... they do not like 4V Vgs, with a few exceptions, 4V Vgs is just abuse IMO, I made the test rig to ensure I did not went bellow that, and during test it proved to be a wise decision, with a few exceptions, all of them at 4V Vgs got really hot really fast and I decided it would be counter productive to go bellow that, turns out they do not compare with logic families, that is why I did not went bellow 4V, in fact you can see that my PSU was 5.17V and I used 2 series 1n4007 to make a 1.2-1.3V range, just to ensure that(I forgot to put on schematic, but my potentiometer was a 10K linear), I considered the idea of 3-3.5V, but after seeing the thermocouple rising. I gave up on it.


And yes, a large enough gate resistor would help on an driving IC but as you mentioned, such slow speeds, anything switching  woud be off the table, and SOA is something that has to be regarded when operating these devices at slow speeds unless your load is super light, i.e. under 500mA draw and even then I think that is pushing it, just imagine, needing a heatsink to operate at 250mA, at this point it may be better to use a power BJT instead. But if you are doing single pulses or using it to turn something like a relay of a lamp on and off, or even to just drive another transistor it can be useful, but I still do not think it is ideal because in almost all cases you will need a heatsink if the transistor stays on for a long time, ans this at full 5V Vgs, perhaps you can get away with 4.75, 4.8 depending on the part, but THAT IS IT! any less and a heatsink is mandatory because in all cases at 4.5Vgs the temperature continued to rise, in some slowly but few stopped.

and lastly, thank you, I figured since you guys helped me a lot in this matter, the least I could do is to contibute with a a bit of data, it is quite the task, but it taught me the value characterizing the parts I have on hand instead of just relying on datasheets

[Geoff-AU]

it taught me the value characterizing the parts I have on hand instead of just relying on datasheets

Can be very useful - the datasheets give you a minimum guarantee about the performance of the part.  If you are designing a commercial product for production then the datasheet is your bible, because you get guaranteed functionality if you obey it.  For your own prototypes or one-off devices, absolutely you can characterise parts individually.  You just need to be aware of what the device will do if the operating conditions change.

[MrAl]

To Berni,

I assumed such, I mean International rectifier does not put DC in the SOA, but Fairchild does in many of what I seen and so did Samsung on the IRFS640. another interesting thing is that if you lookt at the datasheets there may be a hint of how well it works if you look the chart that relates gate charge with Vgs, in some the miller plateau start right before 5v and in some a bit after, of course it s hard to relate because the chart was made with much higher Vds and Id than what I did, but I cannot see how since I do not own an oscilloscope, I have a FNirst, but it is only good up to 200KHz, and has low resolution, maybe with a high enough gate resistor I could see the curve, however I do not see the point of doing so now, nor if there is any use for it. I might just try this today with one after work out of sheer curiosity.


To MrAl,

Hello Again!

Yeah... they do not like 4V Vgs, with a few exceptions, 4V Vgs is just abuse IMO, I made the test rig to ensure I did not went bellow that, and during test it proved to be a wise decision, with a few exceptions, all of them at 4V Vgs got really hot really fast and I decided it would be counter productive to go bellow that, turns out they do not compare with logic families, that is why I did not went bellow 4V, in fact you can see that my PSU was 5.17V and I used 2 series 1n4007 to make a 1.2-1.3V range, just to ensure that(I forgot to put on schematic, but my potentiometer was a 10K linear), I considered the idea of 3-3.5V, but after seeing the thermocouple rising. I gave up on it.


And yes, a large enough gate resistor would help on an driving IC but as you mentioned, such slow speeds, anything switching  woud be off the table, and SOA is something that has to be regarded when operating these devices at slow speeds unless your load is super light, i.e. under 500mA draw and even then I think that is pushing it, just imagine, needing a heatsink to operate at 250mA, at this point it may be better to use a power BJT instead. But if you are doing single pulses or using it to turn something like a relay of a lamp on and off, or even to just drive another transistor it can be useful, but I still do not think it is ideal because in almost all cases you will need a heatsink if the transistor stays on for a long time, ans this at full 5V Vgs, perhaps you can get away with 4.75, 4.8 depending on the part, but THAT IS IT! any less and a heatsink is mandatory because in all cases at 4.5Vgs the temperature continued to rise, in some slowly but few stopped.

and lastly, thank you, I figured since you guys helped me a lot in this matter, the least I could do is to contibute with a a bit of data, it is quite the task, but it taught me the value characterizing the parts I have on hand instead of just relying on datasheets

Hi again,

Yes the data was very interesting thanks for sharing that.

As to the gate resistor, a lot of designs use up to 10 Ohms in series with the gate for various reasons.