The use of electronic measurement equipment in the research environment:

Environment:
cryogenics with samples at 4 to 0.01 kelvin
samples with dimensions at nanometerscale, detecting single electrons 
(quantum-dots, single-electron transistors, nanotubes)
Large magnetic fields (Tesla's), long wires (meters)

Measurements:
V-I and I-V with associated gate sweeps and magnetic-field sweeps
Resistance values range from 10ohm to 10Gohm, Voltage 10nV to 10mV,
Currents fA to uA.  
The measurement bandwidth is usually << 1kHz with fast read-out up to
a MHz. 
Excitation is dc, pulses and rf up to 40GHz all usually in the uV/uA range.

Essential:
The effective temperature of the samples (0.01kelvin) must be protected
from rising caused by injected energy of the environment, including the 
measurement system. This leads to noise & interference level spec's like: 
< 160nVrms integrated over a bandwidth from d.c. to 100MHz
and < 0.5pV/sqrtHz for 100MHz to 6THz. 
Filtering on sampleleads is used at roomtemp (Pi-filters) and cryogenic
 (RC-filters and Cu-powder filters).
Dedicated front-end (analog) measurement equipment prevents excess
noise&interference on the sample and is tailored to the application. 
-----------------------------------------------------------------------

Problems associated with (digitally based) measurement equipment:

-Noise&interference  is specified (and optimised) for it's value 
 presented at the output these measuring devices, however more important
 in measurements on nanoscale/nonlinear samples is the amount of 
 noise&interfence generated at the input terminals.
 Example: Averaging reduces output noise but the sample gets disturbed 
 by the input noise and so the details of an I-V curve to be 
 measured are still smeared out by this and will not be measured.
 
-The input circuitry contains switching elements for
 periodic cycling like: autozero, gainadjust, chopping etc.
 This creates spikes in the input-connected device (sample).
 Example:Keithley 2000 DVM at 100nV resolution creates 20 mV
 (200000x higher level) spikes in a 10Mohm sample resistor. 
 (These spikes  don't disturb the DVM reading but appear at the 
 input terminals and on the sample)

-The GPIB-bus(or other) PC-communication creates charge injections 
 at the input terminals. 
 (Example: on a 10Mohm sample this created >10mV jumps
 which dissapeared fully by using an analog buffer-amp first)

-The input terminals on most DVM's are unshielded 4mm plugs
 and the equipment itself is usually housed in a poorly shielded 
 case (receiving EMI) with digital displays which are scanned 
 (generating EMI). Again, this does not disturb the DVM reading
 but it disturbs the sample.

-There is usually no battery supply possible, only mains supply 
 introducing interference spikes by capacitive coupling and stray
 magnetic fields. (some even use a switched mode power supply) 
 Again, this does not disturb the readout itself but it disturbs
 the sample.

          ********** Conclusions: ************
It seems that the designs for (digitally based) measurement equipment 
are optimised for showing a clean/accurate/stable result on the 
display, but not for keeping the input terminals clean.
Digital equipment has input interference that usually exceeds its
measurementresolution by several orders of magnitude.
Analog circuitry on battery-supply has no input (or output) 
interference.  


R. Schouten 
University Delft, Applied Physics dept.  Quantum Transport group