HI. I doing a current mirror design using a 0.18um technology.
but i found that if i used the minimum length transistor in the design, than the design will have problem with layout mismatch....
so i try to increased the length of the channel. but the technology that i using does not allow me to changes the length. the only option that i have is to series the transistor.
here is my question, how do we series a few shorter length transistor to form a longer channel transistor which have behavior like a single long channel transistor???
Re: Longer channel MOSFET from 2 or more shorter channel MOS
Connect drain to source in a serial manner. Connect all gates together, and all bulk contacts together (all MOSFETs in same substrate resp. well).
It won't have exactly the same properties as one single long transistor, but rather close. (And it seems you have no other choice, anyway) ;-)
Re: Longer channel MOSFET from 2 or more shorter channel MOS
thank for the reply. I have try series the transistors and run some simulation on it. but the simulation result that I get is different than what I have calculate especially the Id.
for example, Id calculated is 27.09uA but the simulated id is 84.9uA.
the equation that i used is Id = 1/2 (UnCox)W/L(Vgs - vth)² neglecting channel modulation effect.
what seem to be the problem??? any suggestion on how to predict the id more accurately?
Do you see a similar discrepancy for a single transistor?
In this case may be your calculation method is not exact enough, because it doesn't include second order effects.
Re: Longer channel MOSFET from 2 or more shorter channel MOS
Short-channel effects are reducing the effective channel length of your unit-sized devices. If your model and model parameters are good then the simulator is taking all these into account correctly, making the total effective L smaller than simply the sum of the individual nominal Ls.
If your hand calculation is using the quadratic mos model then it's certain to be *wrong* even if you were not putting devices in series. 0.18um MOS transistors can't be modeled by the simple MOS model.
Why can't you change (increase) the channel length ? Are you constrained to fixed sized devices ?
Re: Longer channel MOSFET from 2 or more shorter channel MOS
thank you for the reply guys..... I check already the schematic and ya you are right, one of the series transistor is in triode region. that why i can't get the correct Id.
guys, another question, can anyone suggest any biasing circuit to bias a transistor which will have a constant Vgs? before this i have try to bias using current mirror but because of mismatch in the layout, it seem can't not work. thank you
Re: Longer channel MOSFET from 2 or more shorter channel MOS
ansonyeap said:
thank you for the reply guys..... I check already the schematic and ya you are right, one of the series transistor is in triode region. that why i can't get the correct Id.
This is the normal situation when you use this series configuration of transistors. This is not the cause of the error. The first (top) transistor will be in saturation, and all the other transistors of the series will be in linear region.
Check how you calculate your estimate of the Id. The simulated Id may well be right, or closer to the right value.
Re: Longer channel MOSFET from 2 or more shorter channel MOS
Hi,
In such cases, source degeneration really helps. This reduces the Vt mismatch effect. Larger the degeneration resistor, lesser the effect of Vt mismatch. However, "Larger resistor" is compromise between area and performance.