How to go about designing a folding amplifier?

[Karthikeya]

can ne1 plz tell me how to go abt designing a folding amplifier?
it is to be a 8 folding factor cmos differential type

regards

 

[ocarnu]

folding amplifer

You have two options:
-connecting differential pairs with opposite signs in parallel. (You obtain differentuial inputs with a "differebtial adaptor").
- cascading (stacking) differential pairs.
Minimise Vgs in both cases.

 

[Karthikeya]

Re: folding amplifier

hi
suppose i go for cross connection of diff pairs then what shud be the separation of neighbouring voltage references to have good interpolation. just for example take a 6 bit folding adc 3/3 then for 8 folding factor what shud be the slope, the value of resistance?
i have a folding amplifer with f=8 but the problem is there is a tapering the vout/vin characteristic(ideally the outputs must cross at same dc value) which is due to the increase in vcm as vin increases....now how do i correct it?

i m sorry i cant paste the circuit or the characteristic but i hope you can understand the above problem

regards

 

[electronrancher]

folding amplifier

it sounds like you may need common mode feedback in order to keep your amps in the right cm range.

 

[Karthikeya]

Re: folding amplifier

electronrancher can you please explain that in some detail and also on how do choose the resistance?

thankz in advance

regards

 

[ocarnu]

Re: folding amplifier

The separtaion should be VGSeff<sqrt(2)*ΔVref. When you add the caractheristics they should be separtaed enough to generate a traingular charact with maximum aplitude. If it is too separated will become trapesoidal, that's fine but is not optimal. In they are too close is still trinagular bu sub-optimal.
2. You cant make even folding rate with X-connecting pairs without using a dummy current source to equalize de dc level at the output.
3. What resistors? You mean, the load?

 

[Karthikeya]

Re: folding amplifier

ocarnu
i have made sure of the extra decoupled diff pair to ensure proper dc value but still the characteristic remains approx same as previous. yes, what should be the value of resistance for load. if i use Gm amp then the output swing is reduced as a tradeoff for BW, then what is the soln? i have also tried mirroring current and use R-load but with same result. i want good BW and swing.

regards

 

[ocarnu]

Re: folding amplifier

Use an active load. The problem with this folding circuits is the huge DC value of the current (4Xtail current in eachh branch of the output) while only one tail current is used to charge/discharge the load capacitor. So you will have to use a bypass: two current mirros will take most of the DC current, and in parallel you have a diode connected transistor (or an R), but only a small part of the current flows there => you can increse the resistance.

You will never get good BW with this configuratin becuase the Cgd of 8 (or 9) pairs apper as a load. You want more current, but you will have the same ΔVref, so u want same Vgs, so you have to increase trasistors => increased Cgd=> speed stays the same.