LOW current supply (high I) Horowitz&Hill 6.7
Albert M. Putnam
impedance in the 1 GOhm range (BIG - 1000 MOhm for the pikers)
It only needs to be precise to about 5%.
I would use a LM334 (if I could just get my hands on one tomorrow)
Page 354, Exercise 6.7
"Make a 10uA current source with a 317 voltage regulator."
Horowitz and Hill
Art of Electronics
Second Edition
Cambridge, New York 1989
What is the answer? (grrrrrr)
I assume they mean without any op-amps, FET's, or transistors.
I.e the just 317 and passive components.
Every text seems facinated by "precision" and "high current"
circuits. I want the opposite.
BUT high impedance is important.
The circuit is to drive "power sensitive" high impedance devices/loads
for voltage measurements.
Albert Putnam
put...@msc.cornell.edu
Cornell Materials Science Center
Douglas P. McNutt
(Albert M. Putnam) wrote:
>I need a low current source (1-10uA) that will have an output
>impedance in the 1 GOhm range (BIG - 1000 MOhm for the pikers)
>It only needs to be precise to about 5%.
>I would use a LM334 (if I could just get my hands on one tomorrow)
>
If you're in space you can use a radioactive beta emitter.
From the USA. The only socialist country that refuses to admit it.
John E. Lundgren
constant current. That's done by putting a resistor between the output
and the adj leads. The load is taken from the adj lead to ground. A
resistance of 1.2 ohms gives a constant current of 1 amp. 1.2K gives 1
mA, etc.
--
Greg Thoman
>I need a low current source (1-10uA) that will have an output
>impedance in the 1 GOhm range (BIG - 1000 MOhm for the pikers)
You have quite a challenge here. Even an improved Howland
current pump tends not to approach the Gohm range except perhaps
with tender, loving, care.
You'll find an improved Howland on page 388 of H&H, 2ed,
in figure P ("high-current bi-polarity current source") and you
can learn more about it by consulting pages 61-72 of the October
15, 1992 issue of Electronic Design.
[chomp]
>Page 354, Exercise 6.7
>"Make a 10uA current source with a 317 voltage regulator."
>Horowitz and Hill
>Art of Electronics
>Second Edition
[chomp]
>What is the answer? (grrrrrr)
>I assume they mean without any op-amps, FET's, or transistors.
Your assumption has sabotaged you. The answer is to adapt
figure 6.38 (B) to your needs, in which a unity-gain op-amp buffer
allows the 317's adjustment pin current to flow without affecting
the output current.
Have fun!
--
-----
Greg Thoman: The opinions expressed herein are mine alone, and I am
solely irresponsible for them.
Greg Thoman
Note that this only works when the regulator's adjustment pin
current and its variations are negligible compared to the desired
output current. The LM317's adjustment pin current can reach 100 uA,
so that simply using a resistor isn't sufficient for the original
poster's query about a 10 uA current source.
However, inserting a unity-gain op-amp buffer from the
"low end of the resistor" (the output of the current source) to
the adjustment pin circumvents this problem, since the output is
then the resistor current +/- the input bias current of the op-amp.
Winfield Hill
On 5 Aug, Albert M. Putnam <put...@msc.cornell.edu> wrote:
- I need a low current source (1-10uA) that will have an output
- impedance in the 1 GOhm range (BIG - 1000 MOhm for the pikers)
- It only needs to be precise to about 5%. I would use a LM334
- (if I could just get my hands on one tomorrow).
-
- Horowitz and Hill, Art of Electronics, Page 354, Exercise 6.7
- "Make a 10uA current source with a 317 voltage regulator."-
-
- What is the answer? (grrrrrr). I assume they mean without any
- op-amps, FET's, or transistors.
- I.e just the 317 and passive components.
OK, sorry, we could have been more clear. The exercise is inside
the paragraph describing the extra opamp solution (along with a
schematic, fig 6.38-B); the exercise is simply for you to expand
on that design (make it adjustable) and do the calculations.
jlu...@eis.calstate.edu write, in answer to Albert:
- Since the LM317 is a constant 1.2 volt source, you have to change
- that to a constant current. That's done by putting a resistor
- between the output and the adj leads. The load is taken from the
- adj lead to ground. A resistance of 1.2 ohms gives a constant
- current of 1 A, 1.2K gives 1 mA, etc.
As later pointed out by Greg Thoman, John Lundgren's answer was
wrong; the ADJ-pin current must also be counted - that was the little
"gottcha" point of the subsection, and of figure 6.38B, showing an
opamp follower providing the LM317 ADJ pin's current.
-----------------------------------------------------------------------
In his original posting, Albert M. Putnam continued:
- Every text seems facinated by "precision" and "high current"
- circuits. I want the opposite. BUT high impedance is important.
- The circuit is to drive "power sensitive" high impedance
- devices /loads for voltage measurements.
In the power-supply chapter we were talking about power parts,
and many amps, etc. But, it's clear that what Albert needs is
just a standard current source made from bipolar transistors.
We discusss them in detail in section 2.06. (And mirrors, sec 2.14)
Al says he wants 1 G-ohm output impedance. That's di = dv / R or
a change of 10nA for 10V change. At 10uA, that's just 0.1% change.
The Early effect (small changes in Vbe with Vce) makes this hard
to achieve with just one transistor, but in Fig 2.24 we show the
cascode connection, which is one way of solving this problem.
We measured its current stability at about 0.1%.
The circuit of Fig 2.25 also eliminates the Early-effect variation,
since the reference transistor, Q1, always has about Vce = 2 Vbe.
______________ Vcc
| |
| R2
Q1 e |
b-----* fig. 2.25
c |
| e Q2
*-----b
| c
R1 |____ output current
| changes less than 0.1% with Vout
gnd
It also provides positive output currents, and with R1 = 1M and
R2 = 62K + 500K or 1Meg pot, should meet Al's need nicely. Q2
should be high H_FE, e.g. 2N4250, since changes in base current
with Vce lower the output impedance.
Obviously, this circuit can be improved with an opamp feeding Q2-b
and a 1.2 volt reference, like a LM385-1.2 (it wants 20 uA min)
__________________ Vcc
| |
LM 385 R2
| |
| -----------*
| | |
| -- - e Q2
| >---b
*----- + c
| |
R1 |____ output I = 1.2 / R2
|
gnd
This method provides a stable output with supply and temperature
changes. The opamp must operate with inputs near Vcc. You can
see which ones will go all the way to Vcc by looking at the "Swing-
to-supplies? In, +" column in Table 4.1. Also, the opamp should
use FET inputs, since its input current adds into R2 along with Q2.
Some of the choices are LF411, OP-41, TL070 (this whole TL0x1 family),
etc. Some opamps can run single supply (Vee to gnd). If you use
a TL061, then one 9V battery is fine.
Q2 could be a p-channel MOSFET for 0.01% performance. Read
section 4.07 for more ideas.
Winfield Hill hi...@rowland.org
Rowland Institute for Science
100 Land Blvd.
Cambridge, MA 01242
p.s. I'm suspect there may be something wrong with my usenet
feed - maybe it's stuck on local only, etc; if someone can read this,
would they kindly repost a copy of it, so the others can see it?
Winfield Hill
Current sources and current mirrors are one of my favorite topics.
So I've reposted this exchange, which was orginally under the
subject "LOW current supply (high I) Horowitz&Hill 6.7", which I
felt might not alert others who share my interest: