Kn with the EC N-mosfet ... Advanced user notes 1 ... Nmos setup Part B

253

03:38:15

See here for previous basic Nmos setup inc. Rds:   http://everycircuit.com/circuit/6063244598050816

Kn and its relevance in EC:

Mosfet...Advanced user notes 1:

Let's start from the basis that you are now able to remain in Ohmic on a real but simple Nmos board setup with a fixed Rds which conforms to its datasheet Rds. The key thing here is that you need to stick in Ohmic for a while, because we need to be in there to obtain another fixed value from the board which is specific to that individual mosfet, ie:  the Kn

Take V and I board measurements of a simple Nmos setup using this formula to obtain the Kn ... keep Vds very low ... and stay in Ohmic, ie: where Vds << Vov   having already obtained that Nmos Vth:

Kn = Ids / [2(Vgs-Vth) x Vds]   ... keep in Ohmic

A few different V and I attempts should provide a reasonable average result for that real Nmos fixed value Kn 
... a high Vov (overdrive) is generally better, so that the Rds is held low at its datasheet state.... Lowering the Vgs risks increasing the Rds value above its datasheet stated value.

Then use this formula below to find the altered (raised) rds which occurs if Vov is lowered, eg if we now enter saturation where Vds > Vov:

rds = 1/[2Kn(Vgs-Vth)]

That non-ohmic-like raising of rds when Vov is altered (ie Vov reduced) is why we needed to find Kn whilst we were in Ohmic, where the rds could be kept at one single value ... its datasheet value.

 Kn, like B stays fixed:  B = 2Kn    ie: Kn = 0.5B

So...
We have obtained the fixed Kn for our specific real board Nmos by taking simple board measurements.
(That's why you don't find Kn on a datasheet).

Now find B from that Kn, as described just above here  ....B=2Kn

Then use any combination values for B and put them in as EC Nmos settings which form 
B = KP(W/L)

Note that any combination will work ... 
eg: if you ensure that W/L = 2, then you could simply put in the found Kn value as KP ... job done.... but try any combination of  W,  L, KP  if you wish, as long as the B ends up the same as the measured-calculated B from above  ...
 ie: B = KP(W/L)

The real absolute values of a mosfet just don't matter to us ... it is their combination as B (or likewise Kn) which is important.

By doing that, we have in effect set the Kn in the EC mosfet as fixed. So then, this fixed EC Nmos is a good simulation of our real Nmos and can then be used in other EC situations eg for saturation setup which simulates this real known Nmos with those parameters. 

The earlier mentioned useful formula above for Ohmic can then be used at any time, now that we know the fixed Kn ... here it is again:

Kn = Ids / [2(Vgs-Vth) x Vds]  ... for Ohmic

but now, we know Kn, so we can re-arrange that formula and find other required parameters, eg:

Ids = Kn x [2(Vgs-Vth) x Vds]  ... for Ohmic

So now, without any board, we are able to use EC with reasonable confidence, to simulate a known real mosfet with similar internal paramaters and behavior.

Keeping that same Kn, we now need a different formula if we are in saturation, because in that state a new aspect arises ... the changing slope of the saturation curve due to an increase in rds beyond the factory datasheet value (which now pales in to insignificance compared with the new value).... 
but not here today.

eg for above:
See all of the parameters in the schematic.

Alter the resistor to be anything you like 
between 1Ω and 1kΩ

At each alteration, view the shown resistor current (Id)   and the shown drain volts between that resistor and the Nmos (Vds)

Now do  Vds/Ids to get Rds

You will find that Rds always remains the same, regardless of whatever resistor value you use.

ie: we have set the Nmos Rds as fixed.

Likewise... the B is fixed in that Nmos as 1000
and the Kn= 500   fixed 

Use the formula above and confirm any parameter as seen in the schematic.

With a little practice, you can now obtain Rds and Kn for a real mosfet then set its Kn and B in to an EC Nmos knowing that they are fixed, regardless of load resistor alterations... You are simulating a known Nmos in EC, and you can deduce or confirm any parameter with the Ohmic formulae here.
( B is called the mosfet Beta:  β )

See here for Simulation Guide for EC:  http://everycircuit.com/circuit/6260747729633280

published 7 years ago

nikisalli

7 years ago

Useful, bookmarked

2ctiby

7 years ago

@nikisalli ... many thanks, much appreciated

lenzrulz

7 years ago

Well done 2ctiby, you’ve addressed an issue which remained unanswered for several days, your description is awesome, keep up the tutorials...👍

crake

7 years ago

I like it, 2ctiby

2ctiby

7 years ago

Thanks guys ... I hope it can be helpful to someone

jason9

7 years ago

If the Kn is individual to every MOSFET but the Rds is not, then how can I make two EC MOSFETs each with the same Rds but their own individual Kn without changing the Vov?

2ctiby

7 years ago

@jason9 ... The Rds and the Vth and the Kn are all slightly individual and even more so for individual circumstances just as many things are with heat etc.... the whole thing must be taken with a bit of leeway ...it's not an absolute thing like on a computer ... and that is partly why the individual mosfet Rds board reading may be slightly different to the datasheet. Even the datasheet itself often gives a range rather than an absolute specific value fot parameters like Vth etc. All we can do is make our best estimate and then apply necessary tweaks... that's the way electronics are compared with maths/computing.

jason9

7 years ago

Yes, but what if the Rds is one value and the Kn is another, but when you try to do that in the simulation then if the Kn is correct the Rds is too high, but if the Rds is made correct then the Kn ends up to high? Does that just mean that EC can’t even simulate some MOSFETs at all?

jason9

7 years ago

Let’s say you find a MOSFET, don’t know where the data sheet is, and do some testing, so you determine the Kn, Vth, Rds, and all that fun stuff, and there is no range other than that permitted by measurement inaccuracies, and the Kn and Rds were measured with a Vgs of 10V, but when you try to simulate that MOSFET in EC, the Kn and Rds don’t match such that EC won’t permit both to be correct with a Vgs of 10V. Like perhaps it has an Rds of 70ohms, a Kn of 2mA/V^2, and a Vth of 3V (and thus a Vov of 7V), so you try to simulate that transistor in EC, but when you make it have a Kn of 2ma/V^2, it has an Rds of 35ohms (at a Vov of 7V), and when you make it have an Rds of 70ohms (at a Vov of 7V) it ends up with an Rds of 1mA/V^2. What then?

2ctiby

7 years ago

@jason9 ...The Kn is physically fixed, so when you have entered that as Beta: KP(W/L) in to the EC mosfet, then it does not alter. It is mathematically related to the Vov as described by the formula, The Rds however is not fixed, it is only the minimum amount from the factory, as described on the datasheet which is 'fixed' rather like a bought resistor which may be increased by additions but not reduced out of play. So we will see the rds alter as the Vov alters. Bear in mind though that the V=IR idea is not applicable generally, because the Vgs is not connected physically to the drain... so we need to consider how the Vds and Ids are interacting. We don't actively try to alter the rds, it alters as a consequence of the other parameter variations. It is best if you stick to the formula: Kn = Ids / [2(Vgs-Vth) x Vds] and Rds = Vds/Ids with fixed Kn and known Vov until the introduction of saturation where lambda becomes significant.