Maker Pro
Maker Pro

Is this mosfet good for my project?

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Looks fine, and true, you no longer need a logic level mosfet.
 

Xenobius

May 15, 2012
125
Joined
May 15, 2012
Messages
125
Hi Steve,

Thanks great I learned a lot :D So basically when looking for a mosfet now all I need to know is:

- Max Voltage (anything greater then 12v)
- Max current (anything greater than 1A as I will be passing 1A) or should I get a higher current mosfet?
- RDSon (Resistance between the Drain and Source when turned on. The lower the better because it heats less... are there any uses for high resistance ?
- What about gate voltage? Should this be EXACTLY 12V? or how does it work? For example if the Mosfet is -30V, but the Source will be at -12V in my case, will it still be fully off if the gate is at 12v? I dont think it needs to be -30v right?

I think for a mosfet to be fully off, the gate needs to be at source level irrelevant if the mosfet is rated 30v and I am using it at 12v right?

Thanks Steve :)
 
Last edited:

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
I would choose a Vgs (max voltage) of around 30V just to have plenty of margin.

I would choose an Id (max current) well in excess of your requirements, at least 2A, but probably 5 because there are other factors at play, and you want to ensure you get the current you desire, and to have plenty of margin.

Rds(on) should be some small fraction of an Ohm. 0.25 ohms would be fine, and most devices with this voltage and current rating are going to easily meet this. Don't go chasing super low values.

There are several gate voltages. Vgs(max) needs to be at least 12V for the circuit you've drawn (preferably 15 to 20) since the gate can be pulled 12V from the source. The other important one is Vgs(th) which is the voltage that needs to be between the source and gate before the mosfet even starts to turn on. This was important previously (logic level mosfet means Vgs(th) around 3V or less). Again, not so important now, because you'll probably get at least 10V swing and that should drive the mosfet hard on. The datasheets will tell you how hard on if you're interested.

As soon as the voltage between the soirce and gate falls below the Vgs(th) value, the mosfet will be off. Your circuit will get that to almost zero, so no problems there.

The datasheets will have a graph which shows Id vs Vgs. If you draw a line across the 1A Id (2A to be sure), and see what Vgs that is, you can know for certain. I expect it will be something like -4 to -6 volts (perhaps even lower). Many of the voltages you see on the datasheets for P channel mosfets are -ve because they are typically references tot he source, and that is the most positive part of the circuit in many of these cases. I have used +ve voltages up until now, and I'm not sure which would be the most confusing.
 

Xenobius

May 15, 2012
125
Joined
May 15, 2012
Messages
125
That's great information. Couldn't have expected anything better.
I would like to thank you for your time :) I think I have enough information to finalise the triggering part of the circuit.

X
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
The 10K resistor in the level shifter will cause the MOSFET to switch off very slowly. This may or may not be a problem, depending on how fast you need to switch. At what frequency is this going to be switching?

Bob
 

Xenobius

May 15, 2012
125
Joined
May 15, 2012
Messages
125
Hi Bob,

Thanks for your feedback. I didn't know that it was going to turn off slowly. As for the switching frequency, to my understanding if I am going to switch this device ON twice per second, than this is said to be 2Hz... am I right?

If this is the case, what frequency would I be using if I intend to pulse it once and only once? 1Hz?

Thanks
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
The RC time constant between the 900pF gate capacitance and the 10K resistor is 10uSec, which is fine when you are operating at less than about 10KHz, which you are by a factor of 10000, so I think you are ok.:)

If you ware trying to switch at 100KHz you would be in trouble.

Bob
 

Xenobius

May 15, 2012
125
Joined
May 15, 2012
Messages
125
The RC time constant between the 900pF gate capacitance and the 10K resistor is 10uSec, which is fine when you are operating at less than about 10KHz, which you are by a factor of 10000, so I think you are ok.:)

Hey thanks Bob, much appreciated but how did you calculate that I am going to operate less than 10KHz? For example I want to turn on a mosfet for 100mS and off for good (until I reset the MicroController) What frequency is that?

Thanks once again :rolleyes:
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
If you are only switching something once, the frequency is 0. In that case, what you have to worry about is only the delay in switching it. If 10uSec is fast enough, then you are fine.

Bob
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
BobK is alluding to two issues with switching.

The first is that while the mosfet switches, its resistance changes from Rds(on) (or something similar) to infinity (or something effectively the same). During this process, the current through the mosfet falls. HOWEVER, the power dissipation first rises, before falling again. This is because the power is related to the current squared across a resistance.

The upshot of this is that during the switching period, the mosfet dissipates more energy (possibly far more energy per unit time) than if it is on or off. So the more often you switch it on and off, the more time it spends switching, and the greater power is lost. SO... The mosfet gets hotter, and the load gets less power.

This is the first consideration. How much power is lost in the mosfet?

I'm assuming a 12V supply and a 1A load.

Well, if you're switching it on and off 20 times per second (10Hz), and each time you turn it off (10 times) it takes 10us, then it is switching for 100us out of each 1000000us, or 0.01% of the time.

Imagine that during switching, we calculate the losses to be about 3W. That's 3W per second, so over a second that's 0.3mW. For completeness, lets assume that while on, the Rds is 0.1 ohm and that the duty cycle is 50%. So the power lost while the device is on is 0.1*1^2*0.5, which is 0.05W (50mW)

This means the total power lost in heat is about 50.3mW. The loss due to switching is tiny, and as Bob suggests, can be ignored.

Now let's try that at 10kHz...

now there are 10,000 switching (off) events each taking 10uS. This means that the device is now switching 10% of the time. So the switching losses rise to 0.3W, and lets assume the other losses remain the same.

So at 10kHz, the dissipation rises to 50mw + 300mW = 350mW. The switching losses dominate. But, the device does get turned on and off -- it's just that your mosfet gets hotter. (and this is one reason why I suggested a TO-220 device. Even if your switching losses get quite high, it's not going to get too hot)

There is another problem. As Bob suggests, don't go trying to run this at 100kHz -- why?

lets do some math... At 100kHz, you are trying to turn the device off 100,000 times per second. But that device takes 1/100,000 of a second to turn off! This means that the device will never get to turn off before its asked to turn back on again. If your circuit relies on the device turning off, it's not going to work.

So that's the second problem. You may not even be able to turn it off completely.

This is also the reason why I told you not to go chasing very low Rds(on) values. As you can see, the Rds(on), once fairly low, contributes minimally to your power dissipation -- especially at low currents (1A in this case is a very low current!).

Rds(on) is reduced, essentially, at the cost of increasing gate capacitance. And it is that gate capacitance which contributes to slower switching speeds. And as you've seen, the losses during switching can be significant.

I guess you're beginning to see that the answer to "what mosfet should I use" is rather a complex one, and one where the person assisting may need to make assumptions about your project or ask questions that seem strange.

I think I assumed the switching speed was low... How did I do that and was it justified?

Ah.. OK, your question included the following:

I am looking for a P Channel, Mosfet which will be turned ON for 10 - 100mS and off for 30 minutes.

If you had said 2uS and off for 10uS, the solution would have been significantly different and we would have been talking about gate drivers for your mosfet.
 

Xenobius

May 15, 2012
125
Joined
May 15, 2012
Messages
125
This is the first consideration. How much power is lost in the mosfet?

I'm assuming a 12V supply and a 1A load.

Yes this is exactly my load.

And this s a great explanation, I'm literally printing it and keeping it in my notes :D

I guess you're beginning to see that the answer to "what mosfet should I use" is rather a complex one, and one where the person assisting may need to make assumptions about your project or ask questions that seem strange.
:eek:

Indeed this is a complex question so I should have provided you not only with the load, but also with the switching ON speed required (I definitely don't care about switching off speed as long as the solution switches off completely.

So the current configuration switches on in about 1mS which is acceptable but I am not sure if I should be switching it faster because the final product will actually be controlling this mosfet over wireless... so considering the transmission delay, and processing delay from the arduino, the delay to shift the registers in the 74HC595 and the 1mS delayto switch this off, I think it's too much.

What should I be changing in the circuit to make it switch faster? You mentioned gate drivers... I will be looking on the internet for these to understand more but do you have anything in mind for my situation?

Thanks a lot for your time and detailed post. In a previous post I had also posted a datasheet of a new Mosfet that I found with the specifications that you suggested. Is it good?
 

twister

Feb 12, 2012
172
Joined
Feb 12, 2012
Messages
172
BobK is alluding to two issues with switching.

The first is that while the mosfet switches, its resistance changes from Rds(on) (or something similar) to infinity (or something effectively the same). During this process, the current through the mosfet falls. HOWEVER, the power dissipation first rises, before falling again. This is because the power is related to the current squared across a resistance.

The upshot of this is that during the switching period, the mosfet dissipates more energy (possibly far more energy per unit time) than if it is on or off. So the more often you switch it on and off, the more time it spends switching, and the greater power is lost. SO... The mosfet gets hotter, and the load gets less power.

This is the first consideration. How much power is lost in the mosfet?

I'm assuming a 12V supply and a 1A load.

Well, if you're switching it on and off 20 times per second (10Hz), and each time you turn it off (10 times) it takes 10us, then it is switching for 100us out of each 1000000us, or 0.01% of the time.

Imagine that during switching, we calculate the losses to be about 3W. That's 3W per second, so over a second that's 0.3mW. For completeness, lets assume that while on, the Rds is 0.1 ohm and that the duty cycle is 50%. So the power lost while the device is on is 0.1*1^2*0.5, which is 0.05W (50mW)

This means the total power lost in heat is about 50.3mW. The loss due to switching is tiny, and as Bob suggests, can be ignored.

Now let's try that at 10kHz...

now there are 10,000 switching (off) events each taking 10uS. This means that the device is now switching 10% of the time. So the switching losses rise to 0.3W, and lets assume the other losses remain the same.

So at 10kHz, the dissipation rises to 50mw + 300mW = 350mW. The switching losses dominate. But, the device does get turned on and off -- it's just that your mosfet gets hotter. (and this is one reason why I suggested a TO-220 device. Even if your switching losses get quite high, it's not going to get too hot)

There is another problem. As Bob suggests, don't go trying to run this at 100kHz -- why?

lets do some math... At 100kHz, you are trying to turn the device off 100,000 times per second. But that device takes 1/100,000 of a second to turn off! This means that the device will never get to turn off before its asked to turn back on again. If your circuit relies on the device turning off, it's not going to work.

So that's the second problem. You may not even be able to turn it off completely.

This is also the reason why I told you not to go chasing very low Rds(on) values. As you can see, the Rds(on), once fairly low, contributes minimally to your power dissipation -- especially at low currents (1A in this case is a very low current!).

Rds(on) is reduced, essentially, at the cost of increasing gate capacitance. And it is that gate capacitance which contributes to slower switching speeds. And as you've seen, the losses during switching can be significant.

I guess you're beginning to see that the answer to "what mosfet should I use" is rather a complex one, and one where the person assisting may need to make assumptions about your project or ask questions that seem strange.

I think I assumed the switching speed was low... How did I do that and was it justified?

Ah.. OK, your question included the following:



If you had said 2uS and off for 10uS, the solution would have been significantly different and we would have been talking about gate drivers for your mosfet.

Thanks Steve, for that lesson!
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
So the current configuration switches on in about 1mS

if this is still your circuit

4384d1337236060-mosfet-good-my-project-schematic.jpg


Then it will turn on MUCH faster than it turns off. Bob suggested 10us to turn off. it will probably turn on in under 1us (not 1ms)

which is acceptable but I am not sure if I should be switching it faster because the final product will actually be controlling this mosfet over wireless... so considering the transmission delay, and processing delay from the arduino, the delay to shift the registers in the 74HC595 and the 1mS delayto switch this off, I think it's too much.
Even if it was turning on in 1ms, as long as you were leaving it on for significantly longer than 1ms (you said 100ms) then it would probably not be an issue.

However, maybe this is the read-back of the state. In this case, yes, you would need to allow for the delay. 1us may be short enough to ignore (maybe).

What should I be changing in the circuit to make it switch faster? You mentioned gate drivers... I will be looking on the internet for these to understand more but do you have anything in mind for my situation?
Let's hope we don't have to go there :)

Thanks a lot for your time and detailed post. In a previous post I had also posted a datasheet of a new Mosfet that I found with the specifications that you suggested. Is it good?
I'll try to take a look. Is it a lot cheaper?
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
I still see a link to a FDC6506P.

I think that's a little marginal unless you're very confident about your load characteristics.

I recommended something a lot larger, more suited to some random load you might put on it, and one that might even survive some bad mistreatment. If you're confident this won't happen (no accidentally shorting the load, or replacing it with a higher current load) then you may be OK.
 

Xenobius

May 15, 2012
125
Joined
May 15, 2012
Messages
125
Hi Steve,

Thanks as always. Ok so if it turns on in 1uS than its pretty much perfect. I was worried about the delay not because the mosfet wouldn't handle it but because the final application needs to be as precise and fast as possible to the neares 250mS so anything faster than 250mS is ok to me.

Also this is the mosfet which I said was cheaper. The previous one was 60c while this one is like 12c eur BUT it has 2 zener diodes between the gate and the source so I'm not sure. Also this one is 5.5A like you suggested.


http://www.vishay.com/docs/66826/sq3427ee.pdf
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Those zener diodes protect the gate from excessive voltages. Because they conduct at around 17 to 18 volts, this device can be used safely in this circuit. You may want to place a 100 ohm resistor in series with the gate of the mosfet to prevent damage to the transistor if there is a significant transient voltage.

12c is good :)

And you're doing this as surface mount, right?
 

Xenobius

May 15, 2012
125
Joined
May 15, 2012
Messages
125
Thats awesome... I can finally make an order. I will put a 100ohm too then.. (more to buy) hehe but 1000 resistors only cost 2eur so its not a big deal.

Thanks buddy :D
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
I would recommend you breadboard it up, or at least make a prototype before you commit to buying all the components. There may be some gotchas you haven't seen yet.

Maybe make 5 first...

If the cost is low enough, you may be able to absorb the loss. Still worth buying a few extra of everything so you can make a few prototypes Maybe +10% in quantities? It's really up to you.
 

Xenobius

May 15, 2012
125
Joined
May 15, 2012
Messages
125
Yeah infact I was thinking of buying a few before but thanks for the tip :D
 
Top