Current Steering DAC, Switch Transistors in Saturation or Triode??

[jgk2004]

Hello all,

I have always designed my current steering DACs with the switch transistors operating in the Triode region and to reduce the glitch energy not turn them completely off and have a slight overlap as to not kill the mirror. Lately, I was told by a senior designer to have the switches operate in Saturation..but I am seeing problems... When looking at mismatch, I have higher harmonics when the transistors are in Saturation and smaller when they are in Triode. I feel this is correct since, with a higher Vov, with threshold mismatch, the percent will be smaller, therefore Triode should be better.... Is this correct or do I have something wrong with my setup. What it sounds like from this senior guy was that saturation is better... But I would disagree.

What do you guys think? Do you design with the switches in Triode or Saturation?

JGK

 

[lamoun]

Hello JGK,

I design them in triode region too.

Could it be that the senior designer was talking about bipolar transistors where the saturation region would be the equivalent of MOS in triode?

 

[jgk2004]

Hello Lamoun,

Thanks for you feedback. As for the senior guy, he works in MOS only, so there is not mistake. I was also just thinking of the good old square law equations, even thou they dont apply in nm techs, the mismatch would be squared in Saturation, therefore it should be smaller if in Triode. Doesn't this further backup my reasoning which matches my simulations?

Any other DAC designers feedback would be great!

Score: --> 2 triode - 1 saturation (counting the senior guy)


JGK

 

[erikl]

Hello John,

Score: --> 2 triode - 1 saturation (counting the senior guy) JGK

Didn't you ask him for his reasoning?

I increase your triode score: always used triode region in such case.
Of course you can still reduce the mismatch if you use even larger FETs (= going to lower inversion coefficients) - but I'm sure you know this.

erikl

 

[jgk2004]

Hello Erikl,

Well he said for matching and also glitch energy.. But I feel if you control the switching from sink to source, it doesn't matter when considering the glitch energy.

As for the matching... I have both benches right now with the same area(only difference is region), but now I will keep my switches in Triode operation and thus increase the size till my harmonics are below my spec.

Score:---> 3 triode - 1 Saturation.

Any more votes with reason?

JGK