The accurate matching of Fets presents a number of logistics problems that isn't
apparent till you actually start matching them in large numbers. Especially finding 'Quad'
sets (four matched Fets) can be particularly troublesome. A large basic sample is
required, just to start with.
The main problem is the 'thermal versus time' factor.
Fets measure differently with temperature. Thus if you measure a batch in the early part of the morning, you
cannot match them up to a batch measured at noon. This is unless you have a very
accurate temperature controlled environment (available in some labs). This
is not practicable in this case.
I always measure batches of Fets when the temperature is
close to 20 degrees centigrade. But this means I can only match Fets
accurately when measured at the same time. I have built in a couple of additional
steps at the end of the matching procedure to counter this 'time' factor.
More on this later...
First Rule: No Hands
Use plastic pliers when matching Fets. Your
fingers are too warm.
This makes the whole procedure rather time consuming. Each Fet, when measured, changes
its own thermal characteristics. This is because we can only measure it by
making it conduct an amount of current. Guess what? The temperature of the device
now changes with current, so we need to wait while the Fet drifts to its own
equilibrium (which only apply to that set of circumstances right there and then) and can be
compared to others measured close to each other in the same time period.
That is why, right at the end, the Quad (4) or Pair (2) Fets that 'seems' suited
as a set are finally checked within a five minute period. Effectively in
the end, Fets end up as generally been measured as much a four times, sometimes even more.
Here is a run down on general sequence in matching large quantities of Fets:
The Hard Yards: Grouping Fets
Initially, a hundred Fets are measured and grouped into 11 trays. The very lowest measured
Fets, when is R less than 1000, (see further explanation of
R below) are put aside and not matched. The very highest (R greater than 1300) are generally only matched as pairs. The other
9 groups are put aside. Sometimes Quads become available where R is
greater than 1300, please enquire before ordering.
Initial Pairing
Now each group is individually recorded (R = xxxx) and those that measure close
to each other are collected into a single bundle from which we may get a Quad
from and some Pairs. This requires a third testing and the candidates are
grouped as actual Quads and Pairs.
Those that are not matchable are put aside but not forgotten. When a
group either side of the one you have just processed, then the values
that neighbours that group will be taken into account. Finally, the
unmatched Fets go back into the 'kitty' to be added to the next batch of 100 to be matched in the future.
Final Pairing
The final sets of Quads and Pairs are now measured for the final time within the
above specified 5 minute period.
Ratings
At this point two numbers are recorded on the label attached
to the bag containing the Quad or Pair. This may typically look something like
this:
R = 1090
S = 8
QUAD or PAIR
An explanation of those numbers: These are arbitrary numbers generated by the
measurement method. The first one R is the approximate median of the Quad (or Pair)
and S is the spread. Generally speaking an R rating of 1000 is close to an IDSS
of 10mA, 1100 is close to 11mA etc. The S rating is more important. Generally if S is 10 or lower, then the matching of the Pair or Quad is well within
one percent, so S = 10 or lower is very acceptable indeed. Rarely will this go lower
than 5 as it tends to be the practical limit of the measurement method.
That's It!
Hope this will give you an idea as to the care that is taken to provide this
service and the value it delivers.
Regards
Joe Rasmussen
BUY FETS ON-LINE
NOW
