nepdeep,

OK, let me try to clarify.

First of all, we don't need to keep stating the condition Vgs > Vth. This simply tells us that the mosfet is turned ON, and the ON condition is all we are discussing. So, for our discussion, its a given that the mosfet is turned on by a voltage at its gate greater than the threshold ...

-----------------------------------------------------------------------------------------
Case 1 (Saturation):
MOSFET is saturated. This means it is turned on as hard as it can go and is no longer operating in a linear region. The resistance of the mosfet is RDSon. This is how mosfets operate when used as switches in circuits like switching power supplies and motor controllers.

In saturation, the voltage drop across the mosfet is simply Vds = Id x RDSon.

Looking at a mosfet datasheet, the relevant curve for the saturation mode is the Vgs vs RDSon curve. This shows that a large Vgs such as 10V-12V (which is how you saturate most mosfets) gives you a very low RDSon (in really good mosfets this could be 10 milliohms or less)
---------------------------------------------------------------------------------------

Case 2 (Linear operation):
MOSFET operating in its linear region. This means its turned on partially and basically acts as a variable resistor, with the resistance controlled by the gate voltage.

In the linear region the voltage drop across the mosfet Vds is a function of the gate voltage Vgs, and the drain current Id.

If you look at the mosfet curves, the Vds vs Id curve is the relevant one in the linear region. You will be driving the gate at a voltage Vgs which is only a little above the turn-on threshold Vgs(th).

-------------------------------------------------------------------------------------------

Confusion factor:
Here is where the confusion comes in ...

My descriptions above are of the terminology commonly used by designers who use mosfets. We generally think of mosfet saturation as the case where the gate is driven very high (such as 10-12V) and the mosfet is fully on, and the on resistance is RDS(on), generally very low, maybe milli ohms.

But really, the proper use of the term saturation just means that you are operating the mosfet in a region of the Vds vs Id curve where increasing VDS will not get you any more drain current (i.e. the flat part, or top, of the curve). In this case, if you want more drain current, you need to increase Vgs.
-----------------------------------------------------------------------------------------

Its really not that hard to understand from a practical standpoint.

Saturation is when you turn on the mosfet fully by driving its gate to 10-12V (varies depending on the mosfet, so-called logic level mosfets saturate at lower gate voltages)

Linear operation is when you drive the gate at a lower voltage and the mosfet acts as a variable resistor. In this mode, you need to concern yourself with the mosfet curves.
------------------------------------------------------------------------------------------

Hope this helps.

 

nepdeep,

I'll let you continue your general mosfet discussion with Ron. But for your specific circuit...

You have an op-amp driving the gate of your mosfet in a closed loop. The op-amp will drive the gate of the mosfet to WHATEVER voltage is required to make the inverting and non-inverting inputs of the op-amp equal.

You are therefore setting a small voltage across your sense resistor in order to set your load current. If you have the small voltage across your sense resistor, and a 20V power supply on the drain of the mosfet, obviously the rest of the voltage (20V - Vsense) is dropped across the mosfet. This is certainly not in the saturated region, its operating linearly, like a voltage controlled variable resistor.

See post #34. Your definition of saturation is wrong.

 

Yes, I know. I tried to clarify in the previous post but didn't do a good job obviously

As a practical matter, when designing with mosfets ... if you want your mosfet fully on (i.e. using it as a power switch), then you drive the gate to a high voltage, using the Vgs vs RDSon curve in the mosfet datasheet to tell you how high to drive Vgs. You do not concern yourself with the Vds and Id curve in the mosfet datasheet. Engineers usually call this "operating the mosfet in saturation". The terminology is wrong I suppose and we shouldn't phrase it this way.

Sorry for any confusion.

 

Yes, I know. I tried to clarify in the previous post but didn't do a good job obviously

As a practical matter, when designing with mosfets ... if you want your mosfet fully on (i.e. using it as a power switch), then you drive the gate to a high voltage, using the Vgs vs RDSon curve in the mosfet datasheet to tell you how high to drive Vgs. You do not concern yourself with the Vds and Id curve in the mosfet datasheet. Engineers usually call this "operating the mosfet in saturation". The terminology is wrong I suppose and we shouldn't phrase it this way.

Sorry for any confusion.

You posted #41 while I was typing #42.

 

Hey, thanks for the good info! I've been a practicing design engineer for a long time now and the tendency is to forget the theory you don't use regularly, and focus on the theory and calculations that get you to well designed circuits quickly. You're making me go back and look at stuff that I haven't thought aboutin a while. I use mosfets constantly, but almost always as power switches. I haven't looked at a VDC vs Id curve in a long time until these recent discussions.

 

if you want your Mosfet fully on (i.e. using it as a power switch), then you drive the gate to a high voltage, using the Vgs vs RDSon curve in the mosfet datasheet to tell you how high to drive Vgs.

No.
The curves on datasheets are for "typical" parts. You cannot buy a "typical' Mosfet, You get whatever they have which might be "minimum" then your circuit WILL NOT WORK!

The datasheet lists 10V as the Vgs for a certain minimum on-resistance. If you use a Vgs of 10V then ALL of your circuits will work.

 

Thank to Jmac and Ron a lot....I think now I understand....in simple words..I even simulated my circuit and found following

when Vload=0.5, then VGS>≈15 [V], thus truning the MOSFET to saturation
so voltage drop across the MOSFET is 22*Rdson=22*10mohm=220[mV]
and the Rsense ≈ 200[mV] and parasitic loss≈100[mV]

Similarly,
when Vload=20[V], VGS≈=2-3 Volts, so MOSFET is at the linear region,
so Voltage drop across mosfet ≈= 18.98[V]...and
the voltage across Rsense ≈ 200[mV] and parasitic loss≈100[mV]

here the resistance of mosfet at 20[V] vds is ≈18.98[V]/22[A]=≈860[mOhm]
...phew ... thanks a lot...I hope I got it right...comment please

 

Keep in mind that Rds(on) is not a precision resistor. The value .022Ω Is the maximum value at Vgs=10V. Your value will almost certainly be less.
Calculating Rds when the transistor is in the saturation region (Vds>Vgs) is meaningless, because the transistor is very nonlinear at that operating point. It is not acting as a resistor.

 

Your calculations are wrong ...

P = (20V)(22A)(50% duty cycle) = (20)(22)(.5)=220W

Trise = (220W)(60C/W) = 13200 C

Your mosfet is toast

------------------------------------------------------------

Lets work the calculations in reverse to see what is the maximum power you can dissipate across the mosfet...

Trise + Tambient = MaxJunctionTemp

(P x 60C/W) + 25C = 175C

P = (175-25)/60

P = 2.5W

So the maximum average power you can dissipate in this mosfet is only 2.5W. This is 5W ant 50% duty cycle, etc...

------------------------------------------------------------------------------------------

BTW, as others said, your circuit makes no sense. Why are you making a current sink that has no load? All you're doing is drawing a constant current from your voltage source. Why?

Hi Jmac,
I have made a choice of MOSFET and calculated the power as instructed by you as follows:

Maximum power dissipation
The maximum power dissipation of the MOSFET is calculated as follows:

Junction temperature (TJmax) = 175[°C]
Thermal resistance (junction-ambient) θja =40[K/W]
Max. Power= Pmax
Temperature rise= Pmax* θja= 40*Pmax
Maximum junction temperature= Temperature rise + ambient temperature
175=40*Pmax+25
Pmax= = 3.75[W]
This means I can make the duty cycle 50% and Pmax 7[W]...or duty cycle 25% and Pmax 14 watts and so on...but now my question is ...what if the frequency is very low...or if it turns on for 1 min and turns off for 9 min...
I have basically two questions now:
1. How do I determine the max allowable time...
2. Or is this related to the frequency...which actually again determines the maximum allowable time ....

Please help....
P.S.
I was trying to search the answer in this graph of mosfet but got no clue...

 

Nepdeep,

I don't know the exact answer to your question - will have to think about it. Some thoughts ...

Multiplying the max power by the duty cycle is only valid, as you said, if the frequency is a reasonable value. If the frequency were REALLY slow, lets say ON for 1 hour, then OFF for 1 hour, then obviously its not valid.

I suppose the mosfet must have some kind of thermal time constant that I've never seen listed in a datasheet.

In practice, as long as you're switching the mosfet at a reasonable frequency (maybe 10hz or more???) then its probably a valid calculation.

BTW, is it really valid to use 25C for your ambient temperature in your equation? Will your circuit ever be used anywhere where the temperature exceeds room temperature?

Also, I would never run a mosfet at its full rated junction temperature. I always try to keep them under 100C worst case just to make sure the design is reliable and has margin. What is the circuit is used in a higher ambient environment? What if the surrounding components cause a temperature rise inside your enclosure?

Have you considered adding a heatsink to reduce the mosfet junction temperature?

 

Nepdeep,

I don't know the exact answer to your question - will have to think about it. Some thoughts ...

Multiplying the max power by the duty cycle is only valid, as you said, if the frequency is a reasonable value. If the frequency were REALLY slow, lets say ON for 1 hour, then OFF for 1 hour, then obviously its not valid.

I suppose the mosfet must have some kind of thermal time constant that I've never seen listed in a datasheet.

In practice, as long as you're switching the mosfet at a reasonable frequency (maybe 10hz or more???) then its probably a valid calculation.

BTW, is it really valid to use 25C for your ambient temperature in your equation? Will your circuit ever be used anywhere where the temperature exceeds room temperature?

Also, I would never run a mosfet at its full rated junction temperature. I always try to keep them under 100C worst case just to make sure the design is reliable and has margin. What is the circuit is used in a higher ambient environment? What if the surrounding components cause a temperature rise inside your enclosure?

Have you considered adding a heatsink to reduce the mosfet junction temperature?

First of all thank you very much.
I tend to learn more on this foum than anywhere...as we discuss more practical approach here...

Now back to our discussion.
I have considered adding heat sink...which would decrease the total thermal resistance...and other thing is...the device would be used in laboratory ..no other place or ambient temperature is around 25 maintained...and thirdly...if I choose the value such as 100 as max temperature...I get very low (maximum power dissipation) so I ultimately have to limit the voltage and current value...but I have the requirement of 20 v and 22A at the worst casees...but most importantly...let us continue with previous problem of claculating ....the mosget max frequency or on time...thanks Jmac..

 

I'm working on slew rate control. What range do you need?
Answer should be in amps/second, or amps/μs, etc.

 

I'm working on slew rate control. What range do you need?
Answer should be in amps/second, or amps/μs, etc.

5[A/us]is minimum requirement...higher is better.....yes...controllable slew rate

 

I don't remember if you ever said what exact mosfet you were using. But if its a TO-220 or similar, just bolt it onto a scrap piece of metal with some thermal compound and you will get better performance for essentially zero cost. Also, you said that this was a lab instrument of some sort? How about using an inexpensive fan to get airflow? I know you want to get back to the calculations - just making some suggestions

 

I don't remember if you ever said what exact mosfet you were using. But if its a TO-220 or similar, just bolt it onto a scrap piece of metal with some thermal compound and you will get better performance for essentially zero cost. Also, you said that this was a lab instrument of some sort? How about using an inexpensive fan to get airflow? I know you want to get back to the calculations - just making some suggestions

Yes, Jmac....as far as my feasibiliity study...uptill now shows that the device is make-able ...as you said i am stuck at the calculations......and help needed..


@ RON...m excited to hear from you

 

Yes, Jmac....as far as my feasibiliity study...uptill now shows that the device is make-able ...as you said i am stuck at the calculations......and help needed..


@ RON...m excited to hear from you

Do you have a signal generator or controller that can generate pulses of variable frequency, pulse width, and slew rate?

 

Do you have a signal generator or controller that can generate pulses of variable frequency, pulse width, and slew rate?

Yes exactly...i give the external input from the signal generator...and I can control the slew rate by using controlling rise time and fall time ...if your case is the same then i am looking forward to hearing from you...and yes!! eagerly...
thanks

 

OK, here's my design. It has the same topology as yours, but is optimized for transient response.
The discrete op amp gave me better transient response than any high speed op amp that I tried. The down side is the DC performance. You will either need to match Q2 and Q3 Vbe's to about 5mV or less, and also Q4 and Q5 Vbe's to about 5mV or less. Match betas to about 10%. Match Vbe's by connecting collector to base, then use your multimeter's diode test function.
The other option is to add a pot, and don't worry about matching. I can provide a schematic and adjustment procedure if you want to go that way.
I picked the MOSFET at random. If you choose another part number, you might have to change the value of C1 to optimize transient response.

You can change U1 to another part number, but it needs to have a slew rate of about 10V/usec or higher.

 

Ron..if you dont mind can i have the ltspice file..please

 

Ron..if you dont mind can i have the ltspice file..please

I added it to my previous post just now.