Maker Pro
Maker Pro

cost of measuring picoamps

P

Paul Rako

Jan 1, 1970
0
Allan said:
Perhaps. Being a mathematician, I gravitate to the ordering LCM
(least common multiple). The national semiconductor site hangs
my browser. I tried it in lynx and got some information. They are
talking in terms of selling them in lots of 1000, with a $9.46
rail charge. Jameco doesn't turn up the part. National Semiconductor
describes the part as obsolete. What is used now?

I would probably only need to measure nanoamps, since Melissinos' book
says that the Franck-Hertz tube would produce nanoamps under the
conditions of the experiment. On the other hand, one wants a good
picture of the dips, so I feel better about picoamps.

When I asked Paul Grohe about this in my previous pose I also
copied Bob Pease. He responded with some important clarifications
to my statements. My comments are preceded by a >, Bob uses ***:
==========================
but I want to respond about the op-amp thread as well. Is it correct to say:

"The LMP2012 is a chopper amp and would be unsuitable for
measuring picoamperes.
**** Actually, the LMP2011/ LMP2012 are reasonably good
for measuring moderate numbers or DOZENS of picoamps,
but like most Chopper-Stabilized amplifiers, it's a little
noisy, especially for frequencies above 10 kHz. It has
superior VOLTAGE offset errors, but so-so current offsets
and noises. At a 1 nA full-scale ( i.e. with Rf = 100
or 1000 megohms) it's a little noisy. The CMOS dc
(non-chopper) amplifiers are MUCH quieter at that range,
and much less Ib and Ios.
Maybe the LMC6484 but the real deal is that the LMC6001
is just a hand-selected LMC660.
*** Not exactly. The LMC6001 is a SINGLE Amplifier
so it is similar to 1/4 LMC6484 or 1/2 LMC6482 (dual).
But it's a different mask-set, for sure. Just similar.
"Any LMC part would be OK as well as JFET parts like the LF411.
Input bias doubles with every 10 deg C in temp so keep it cool."
*** LF411's can be typically as good as 50 pA, but even the
LF411A grade is only guaranteed better than 200 pA. Our
CMOS -input amplifiers are normally 1000 to 10,000 x less
input current than that. The reason is, their input protect
diodes are TINY junctions; the BiFETs have the reverse
leakage of the whole TUB, which is physically a much larger
diode. More than 100x bigger.

*** As with many other signals, the best accuracy depends
on the actual signal range of V and I, and the signal's
impedance, and the feedback resistor's value.

*** If you are trying to measure accurately a dc signal of
100 or 1000 pA, some of the amplifiers listed above might
be OK. But if you are trying to measure 1 pA, you want
one of the better CMOS amplifiers. The LMC6001 is guaranteed
less than 0.02 pA at room temp; the 1/2 x LMC662 is normally
4 x better, BECAUSE it has smaller input protect diodes.
(Which is why its ESD rating is only 800 volts, compared
to the more robust ratings of the LMC6001) . So the choice
of the "best" amplifier DEPENDS on how big your signal is,
in V, I and Z - as is true for most every amplifier application.
There is no one amplifier that is "best" for EVERYTHING.
After all, we are talking about a LARGE number of orders of
magnitude of I and V... On the other hand, if one of the
amplifiers is "good enough", and is cheap, save your money
and buy the cheap one.

*** Paul Grohe's comments on this are quite right, as he
and I did a LOT of measurements on the old CMOS amplifiers.

*** I hope this clarifies the question of "good, better,
best". / Best regards. / rap / Robert A. Pease / Staff Sientist, NSC
===========================
I hope Bob's comments above are helpful.
 
A

Allan Adler

Jan 1, 1970
0
Thanks very mich to Paul Rako for his two helpful postings about LMC6001
and other amplifiers. I'll be better able to look at the National Semiconductor
website when I get to a library computer. The actual signals will be of the
order of nanoamps but in some of the experiments it will be desirable to
measure them very accurately.

Basically, I'm reading the descriptions of how to do the Franck-Hertz
experiment in Melissinos' book and trying to figure out how to do it
without "simply" purchasing the whole Franck-Hertz kit from Leybold
or its representative for more money than I can possibly afford. I want
to purchase a gas filled tetrode from a contemporary tube vendor and
use it as described in the experiment. Since the current through the tube
will be in picoamps or tens of picoamps, that's what has to be measured.
For measuring multiple excitations, it doesn't have to be extremely
accurate, but for measuring single excitations it is desirable to have
very accurate measurements to be able to detect excitations to other
low lying excited states. The various op amps mentioned are probably
good enough, but I'll have to study them carefully. The more thought
I invest in it, the less money I'm likely to waste on the wrong parts
and I don't have much to waste.
 
Top