ACTIVE vs PASSIVE

[MY PIXEL]

Posted to CCD-world:
-+-+-+-
John Geary,

It is interesting to watch what is happening to the CCD load resistor these
days. The trend at the CCD manufacturer is to make the on-chip amplifier
smaller and smaller (the node sensitivity is apparently winning out over the
increase in white and 1/f noise leading to less noise electrons). In turn the
MOSFET transconductance is getting smaller too . . . today it is < 100 umhos
for high performance CCDs. 

Questions because of this trend  .. . . 

For a given node sensitivity (i.e., V/e-) does a passive off-chip load
resistance degrade signal to noise? Are we forced to an active load resistor?
It appears this is the case from the following analysis.

The CCD signal goes with the voltage gain of the output amplifier as,

		ACCD = gm RL / (1 + gm RL)

gm = transconductance
RL = load resistance

Note as the MOSFET is made smaller the voltage gain decreases for a given
RL.		

***********
Example given: 

gm = 100 umhos
RL = 20 k ohms

ACCD = 10-4 x (2 x 104) / (1 + 10-4 (2 x 104))

ACCD = 0.6667 V/V

Ideally this should be unity.
 
**********

The white increases as the MOSFET is made smaller as,

Noise = (4 k T Rout)^1/2

T= temperature (K)
k = boltzmann's constant
Rout = RL / (1 +gm RL)

***************
Example given:

gm = 100 umhos
RL = 20 k ohms


Rout = (2 x 104)/[1+10-4 x (2 x 104)]

Rout = 6.6 x 103 ohms 

Noise =(4 x (1.38 x 10-23) x 293 x 1 x 6600)1/2

Noise = 10 nV / root-Hz

About what is measured.

***************

There is nothing we can do about white noise (it will go up with output
impedance). However, we should attempt to keep the CCD voltage gain near
unity. Note a 20 k load resistance is too small for todays CCDs (from the
example above we loose 34 % in gain and a corresponding loss in S/N). We could
increase the load resistance to help maintain a high voltage gain. However,
the increase is limited because a finite amount of drain current must flow to
maintain transconductance. This current requirement is about 1 ma.		

>From this analysis it appears that all future CCD designs should include an
active load. In addition, a second amp stage should be included. That way you
can go back to a passive load resistor (not to mention higher speed
performance).

The Pluto/Cinema devices recently fabricated have single and dual stage amps
(the latter having an active load). Measurements show that noise floor is near
identical. However, the on-chip load design and its doping must be properly
selected (past active loads were very noisy . .. e.g., Thomson). 

To conclude the 20 k load resistor is probably outdated. The future should
bring dual stage with active load.

As for finding a good constant current diode for current single stage CCD
units. One might consider to design and fabricate a load specifically for the
CCD (thousands could be made on a single run - want to go into business?).

Jim
-+-+-+-
For information about CCD-world, send email to owner-CCD-world at cfht.hawaii.edu.