Hello everyone,

After doing almost 2 days research, I ended without knowing 100% when to use a bjt or a fet.

Consider these examples all using an µC:

Driving an 5V relay;
Driving an 12V relay; (I suppose its the same case)
Driving a high power circuit;
other examples you can thing of.

I know how FETs and BJTs work, so to save you time and work, you don't need to write about that (Vgs, internal resistance, V drop, etc).

In each example, if possible, please state a usable FET and/or BJT example.

Thanks a lot!

 

With a BJT (not Darlington), for an NPN common emitter (saturating) stage, you generally need base current to be about 10% of collector current to guarantee saturation. Your considerations are then
1. Collector (load) current.
2. Collector current capability.
3. Output current of the μC, to provide the required base current.
4. Power dissipation probably won't be an issue, because μCs can't provide enough base drive to allow enough collector current to overheat the transistor, so long as you observe the rule Ic/Ib≈10.

With a MOSFET,
1. you need an n-channel logic-level device whose Rds(on) is specified with Vgs which is at or below the Vcc of your uC.
2. Rds(on) must be low enough that Iload^2*Rds(on) is well within the power dissipation specs of the MOSFET.
3. At high switching speeds, gate capacitance can slow down the switching speeds to the point that overheating could become an issue.

Maximum voltage ratings must always be observed for any device.

Other members may come up with consideration that I forgot to include.

Darlington transistors require more of an explanation than I have the energy to get into right now.

 

With a BJT (not Darlington), for an NPN common emitter (saturating) stage, you generally need base current to be about 10% of collector current to guarantee saturation. Your considerations are then
1. Collector (load) current.
2. Collector current capability.
3. Output current of the μC, to provide the required base current.
4. Power dissipation probably won't be an issue, because μCs can't provide enough base drive to allow enough collector current to overheat the transistor, so long as you observe the rule Ic/Ib≈10.

With a MOSFET,
1. you need an n-channel logic-level device whose Rds(on) is specified with Vgs which is at or below the Vcc of your uC.
2. Rds(on) must be low enough that Iload^2*Rds(on) is well within the power dissipation specs of the MOSFET.
3. At high switching speeds, gate capacitance can slow down the switching speeds to the point that overheating could become an issue.

Maximum voltage ratings must always be observed for any device.

Other members may come up with consideration that I forgot to include.

Darlington transistors require more of an explanation than I have the energy to get into right now.

thanks for your answer.
About the darlington configuration, I know the fundamentals.
actually, I know the theory how darlington configuration, BJTs and FETs work.
I'm only confused about when to use wich (I know that I need to use darlington when I need a amplification that I cant achieve with a single one. for example, gain of 100 in just 1, darlington configuration woud be 100*100 = 10000 gain - at least is what I have learned).

Once I used a darlington configuration to be able to drive something wich I couldnt with the output of the µC.

But not speaking about darlington configuration, in wich cases would you use BJTs and FETs with an µC?

I know that when switching speed is critical, one should go with FET.
And it seems that for higher currents, it is difficult to get small fets in comparation to BJTs.

If I dont want too much dropout, I think I should also go with the FET, right? (with low Ron, or else it wouldn't matter).

(sorry for the big text)

 

Packaging is another concern. I don't know if you can find a logic-level MOSFET in a through-hole package. Are you willing to use surface-mount devices? If you are, the choice still may be just about availability. Do you have a stock of BJTs? If you do, I would use them, unless the load is more than about 250mA.
A MOSFET may not switch faster than a BJT when you are driving a moderate current load (up to a couple hundred mA). BJTs have storage time. MOSFETs have gate capacitance.
High current BJTs are slow, but you can't drive them directly from a μC anyway, because the base current will be too high.

 

Packaging is another concern. I don't know if you can find a logic-level MOSFET in a through-hole package. Are you willing to use surface-mount devices? If you are, the choice still may be just about availability. Do you have a stock of BJTs? If you do, I would use them, unless the load is more than about 250mA.
A MOSFET may not switch faster than a BJT when you are driving a moderate current load (up to a couple hundred mA). BJTs have storage time. MOSFETs have gate capacitance.
High current BJTs are slow, but you can't drive them directly from a μC anyway, because the base current will be too high.

Well, the idea is what to use for prototyping. I would like to have the prototype as close as possible to the final "product" (not for selling though).

On the end-product, yes, I actually prefer SMD, but for prototyping not.
That is another thing... for SMD i find a lot os FETs suitable for every need, but not in THT.

About the BJTs, I have some very nice BJTs that I normally used until now. But I was thinking about using FETs to increase power efficiency.
I have the SS8050, and D882. Both BJTs.

What would you suggest?
Using BJTs (if i have low current, I would user darlington configuration) for prototyping, and then, in the final product, using FETs?
Note that I'm refering only for switching purposes, like driving LEDs, Relays, valves, anything switchable.

Btw. thanks again for your time and fast response.

 

If you want to prototype with SMDs, you could use adapters.

 

I was going to give you a long list of differences between a JFET and a BJT but then you wrote this:

Note that I'm refering only for switching purposes, like driving LEDs, Relays, valves, anything switchable.

You have to make the distinction between small signal amplification and a power driver.

If you are referring to low to medium power drivers then a BJT is an ok choice.

If you are talking about switching applications with medium to high power then we are looking at either a power MOSFET or a power BJT.

MOSFETs in general have low input current requirements and low drain-source on resistance. Hence they can handle higher currents.

 

If you want to prototype with SMDs, you could use adapters.

That would be an option. But I was hoping I could avoid using adapters.
but if I must... then I have no choice.

I was going to give you a long list of differences between a JFET and a BJT but then you wrote this:



You have to make the distinction between small signal amplification and a power driver.

If you are referring to low to medium power drivers then a BJT is an ok choice.

If you are talking about switching applications with medium to high power then we are looking at either a power MOSFET or a power BJT.

MOSFETs in general have low input current requirements and low drain-source on resistance. Hence they can handle higher currents.

Well, from what I see from the small THT Mosfets and BJTs, like TO92, the bjts can handle much higher current.

The advantages/disadvantages I would have in this case is that Mosfet requires virtualy no current at the gate, while the BJTs needs current and have a rather high V drop (Vec) compared to the Mosfet (not in all cases but most of it, it is).

When I go for SMD, then (for switching purposes) it seems that the only advantage from the BJTs is... they are lot cheaper. But then again... If you use Mosfet your electric bill won't be as high (even if the diference is very small) xD.

So, in the end, from what I read and I understood from you guys, is to use BJTs for switching low to medium-low current hungry devices, and MOSFETs for medium upwards hungry devices. right?

so, in that case, to minimize costs and have some efficiency, if the µC can't drive directly and BJTs, could I use a cheap, low current capable, even if it has rather high Ron to drive one BJTs wich would handle the current? Is it a good option?
(this would be to not stress the µC)

 

It comes down to: "What do you want to drive?"

 

My rules are: If a bjt will do the job, I have drawers full of them, but a MOSFET will always beat a bjt at high currents, and they do just fine at low currents. I am expecting that all uC output switches will become MOSFETs soon. As a matter of opinion, go with the flow and use the (many recently improved) MOSFETs.

 

I told the OP he would need logic level MOSFETs, if that's what he decided on. Not sure if he got that.

 

I am expecting that all uC output switches will become MOSFETs soon. As a matter of opinion, go with the flow and use the (many recently improved) MOSFETs.

Aren´t they already? I thought that all todays digital and probably lots of analog stuff like drivers etc. is made in CMOS or similar logic, so the outputs would inherently be mosfets as well?

 

With a MOSFET,
1. you need an n-channel logic-level device

With a dissertation as long as yours was, it is often difficult for a beginner to tell which points are more important.

Now it has been pointed out for him.

@kubeek: I was referring to the external drivers that are necessary because the Uc's have such limited current capabilities and that designers have to choose among, because this thread is a design question about external drivers.

 

With a dissertation as long as yours was, it is often difficult for a beginner to tell which points are more important.

Now it has been pointed out for him.

@kubeek: I was referring to the external drivers that are necessary because the Uc's have such limited current capabilities and that designers have to choose among, because this thread is a design question about external drivers.

I thought I was being pretty succinct, considering the scope of his original post.

 

Thank you all for all your replies. They have been all helpfull.

One last question:
I found some p-mosfets.
Well... I was searching for a logic level p-mosfet, but I found a standart mosfet that has the same Vgs as the logic level. shouldn't both be the same type?

Mosfets:
ZVP2106ASTZ
BS250(KL)

 

The voltage required to turn the mosfet on is the determining factor, not what some underpaid clerk did or did not happen to type that day.

 

I don't know if I'm doing something wrong or not, but... would this be ok?

(NPN Bipolar + N MOSFET version)

EDIT: I hope what I posted isn't something barbaric

 

For a load switch (like a relay), you could use either. It reality, there are pros and cons to both. Operationally, A FET can mimick a BJT and a BJT can mimick what a FET does. The person designing it should know what he can get away with, so that is what dictates whether or not you use a BJT or Fet (also, cost does too, especially when you have to buy over 100K). I prefer using FETs just because I think they are much easier to design around, but I have also encountered problems where using FETs would cause more harm than good, so it was worth while for me to do the "extra work" and design around mosfets.

 

I don't know if I'm doing something wrong or not, but... would this be ok?

(NPN Bipolar + N MOSFET version)

EDIT: I hope what I posted isn't something barbaric

That will switch a load, but I can't imagine why you would want to do it that way.

 

That will switch a load, but I can't imagine why you would want to do it that way.

the load consumes over 500mA
so i need +/- 20mA at the gate of the BJT, i think.
To put away that "stress" from the µC, I drive the BJT with an mosfet, so that the µC itself, consumes the less current possible.

Is there something I shouldn't do?