make it clearly beta and betaeff

First of all, I'm assuming you're talking about the drain current with the devices in saturation - so are they in saturation? Secondly this is dependent upon the model - those simple equations break down for modern devices and modern models are more accurate. The documentation covers the details of the equations used. I would expect that for low level models (e.g. mos3) the equation you have will be a reasonable approximation, but it may not be for bsim4 (say) - but only in the saturation region. Thirdly, why not get the simulator to tell you the id rather than trying to work it out yourself?

Andrew.

Hi Andrew,
yes, i talk about the transistor in saturation mode, and i use the bsim3v3 model. i looked through the document(Spectemod), but i just found they mentioned betaeff, not give the detail explanation.the reason why i calculate the drain current(nmos saturation mode) is that i am designing a current source(current mirror), i found i can't get the copy current in the output branch, so i want to calculate it by hand , and check out my design fault. and i searched betaeff on the internet, most people talk about beta, not the betaeff in the spectre result, so comes this confuse.

thanks,
zfeng

betaeff is just "betaeff = ueff * coxe * weff / leff" where weff and leff are the effective width and length (i.e. subtracting any deltas to account for etching, bias dependencies and so on (there are equations for this in the models). ueff is the effective mobility (again, this is because it's length dependent) - there are equations for that.

So betaeff is essentially the beta of the device using the real width and length not the drawn width and length.

Andrew.

Hi Andrew,
thanks, so for calculating the drain current, we can use the equation like: Id=1/2*betaeff*(Vgs-Vth)^2, here, we neglect channel length modulation effect, is the equation right? for analyzing the basic current mirror, i use a vdd=2.5 w=2u and l=0.35u to configure the circuit, and i get the id=655.2uA, but when i calculate it by hand use the above equation id=1/2*662.3*(2.5-0.5484)^2=1.261mA, betaeff and vth comes from the op results, where i am wrong?

regards,
zfeng

Zfeng,

As I said, most likely the problem is that you're using a basic text book equation, which does not represent what happens in bsim3v3 (or real life) for modern small geometry devices. You're effectively using a 30 year old model equation which was probably OK for large geometry processes.

I have not tried to follow through the bsim3v3 equations to see if it can be simplified to the form you have above - that's a lot of work (and not especially useful).

Regards,

Andrew.

Hi Andrew,
    ok, i got it, thanks, i think i am out of time.hehe...

zfeng

Hi Zfeng

Just to add on a bit to what Andrew has explained, you are using a 1st order square law model which is derived from the more complicated Pao-Sah model using a number of assumptions (e.g. ignore hole current, combination and regeneration, etc).


Best regards
Quek

hi Quek,
     thanks,did you do some hand calculation before the circuit simulation? at present, for the small size device, is there any equation for hand calculation?

regards,
zfeng

Hi Zfeng

Yes, usually hand calculations using your 1st order equation is good for estimation.

Best regards
Quek

Hi Quek,
    yes, people usually use the same equation for estimation, but i got a so big difference between hand calculation and computer simulation? could you check a little about my current mirror (refer to above parameters)? i feel rather strange about the difference? is the difference is true, why people do hand calculation?

zfeng