high speed cmos current mirror?

[SUNBELT]

I am trying to mirror 80u to a differential switch wich operates with 1g speed.
I tried the cascode mirror but it didnot work
do you have any suggestion or know any paper about high out impedance current mirrors?

 

[eda_freak]

please explain with brevity...it seems ur trying to say that u need a current sink with 80uA capacity...can u give more details about the technology used and topology...?

 

[sankudey]

Hi,
U can try with increasing the length (2u or more) and avoid the cascoding.....but I don't know whether it will give a better speed....

again...if u don't stop the mirrored current MOS, instead u use a differential swith, i.e., when u try to push it to the o/p on the swith u employed and when off, u turn on another swith which will push the current to some dummy (like gnd/vdd etc.)....in this configuration u have to bother about only the switching speed of the mos.....and the magnitude of the current will not affected very much (but a little).....

sankudey

 

[venkateshr]

Hi ,
I dont think it(mirror) has to do anything with the freq. except for the impedance it offers then ( parasitic cap could bring down the imp. at that freq. ) .basically cascode mirror shd work at any freq. . please let me know if I am wrong.

thanks

 

[sankudey]

Actually that depends on the scheme of operation.....As discussed earlier...the detail is yet to come up....but if the current mirror is always ON with a differential switch....as I have mentioned in above post.....it helps in keeping the current value steady for the mirror....only little transient effects are observed.....this topology is popular in current steering DAC...and it works fine.....SUNBELT may try this.....only extra requirement will be one inverter for the extra complementary switch to be driven....some times people also do it by taking Q and Qb from a DFF...if a DFF could be accomodated.....

 

[SUNBELT]

I have attached the simple view of the circuit, the inputs are not (invert) of eachother with the 1g freq. this reduces the out put impedance of the current mirror I think I should change the current mirror in the way that its output impedance increases,

 

[montage2000]

maybe active cascode is seful, that is use OP to boost output impedance

 

[showtime]

if you use gain-boost, be care of the feedback speed and stabilty.