some questions about high precesion constant current source

hi,everybody.
I have some questions about hight precesion constant current source.My
goal is 1uA whose error is blow 0.01%.

In my circuit I used a op-amp,p-mosfet and 5V voltage reference.
But the result is so bad,the error is almost 1%.

Could somebody give me some hints?
Thanks very much.

eehinjor

 

[eehinjor]

added:
1,the voltage of the load is between 0V~5.5V.

2,the output of voltage reference is 5V.

3,the sources of op-amp are +12V and -12V.

thanks

 

[John Popelish]

eehinjor@163.com wrote:

hi,everybody.
I have some questions about hight precesion constant current source.My
goal is 1uA whose error is blow 0.01%.

In my circuit I used a op-amp,p-mosfet and 5V voltage reference.
But the result is so bad,the error is almost 1%.

Could somebody give me some hints?
Thanks very much.

eehinjor

So you must have a current sense resistor that drops 5 volts when 1 uA

passes through it. That would be a 200k resistor, right? In order
for the closed loop to have .01% accuracy, not only do you need
sufficient loop gain, but the opamp input impedance would have to be
more than 10,000 times this 200k, or greater than 2 giga ohms. What
opamp are you using?

 

[Chris]

eehinjor@163.com wrote:

hi,everybody.
I have some questions about hight precesion constant current source.My
goal is 1uA whose error is blow 0.01%.

In my circuit I used a op-amp,p-mosfet and 5V voltage reference.
But the result is so bad,the error is almost 1%.

Could somebody give me some hints?
Thanks very much.

eehinjor
eehinjor wrote:
added:
1,the voltage of the load is between 0V~5.5V.

2,the output of voltage reference is 5V.

3,the sources of op-amp are +12V and -12V.

thanks

Hi, eehinjor. Welcome to the newsgroup. For things like this,
typically

* 1% is easy -- just cobble together a cookbook circuit

* 0.1% requires quite a bit of attention to detail, but is usually not
too tough and somewhat more expensive

* 0.01% starts getting into lab quality -- really tough and much higher
cost for precision components

From your description, you have something like this (view in fixed font
or M$ Notepad):

`
`
` VCC VCC
` | +
` .-. | -----> +
` | | '----------o
` R1| | 1uA
` '-' .----------o
` |5.000V VCC | <---- -
` o-----. + |
` | | IC2 | |
` IC1/-/ | |\| ||-+
` ^ '-------|+\ ||-> Q1
` | | >---||-+
` | .--|-/ |
` | | |/| |
` === | === |
` GND | GND |
` '------------o
` |
` .-.
` R2 | |
` | |
` '-'
` |
` ==` GND
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

(You might want to download Andy's ASCII Circuit -- it's a lot easier
to use a picture to describe what you're doing.)

If you would, a few questions:

* If the above circuit is correct, could you please specify what you're
using for IC1, IC2, R2 (wirewound/metal film/metal oxide/carbon film,
with tolerance), and Q1? Each one of these components could easily be
the culprit. If the above isn't correct, what's different?

* What's your budget? If you actually had to pay much more than a
couple of bucks for 0.01%, would you still want it as badly? Can you
live with less?

* Do you need accuracy as well as precision? In other words, do you
need an 0.01% DC current source at somewhere around 1uA, or does it
have to be exactly 1.00000uA? That's going to be more difficult.

* What's the temperature range for this circuit?

* You've said you've got a burden which will cause a load voltage
between 0 and 5.5V. But you didn't describe the load, or what you're
doing with it. Is this an instrument which has banana jacks, and you
want 200 feet of wire going out to the load and back? Is the load
purely resistive? Are you planning on unplugging it while the circuit
is on? Or only off? Please describe.

* Have you put a scope on the output, looking for oscillations? This
is basically a unity gain non-inverting amplifier, and is prone to
misbehaving. Depending on your components, your circuit may actually
be operating correctly as a high speed oscillator.

* While we're on the subject, what are you using to measure current and
check your circuit? Even measuring DC current to that accuracy
requires a bit more than a handheld DVM.

If this is a dedicated circuit with a purely resistive load in very
close proximity to your circuit, it might be easier to lose the FET and
do something like this:

` .---. R(load)
` | | + ___ -
` === | .-o--|___|--o---.
` GND | | |
` | | |
` /-/ | VCC |
` ^ | + |
` | ___ ___ | |\| |
` -5.000Vo--|___|-o-|___|---o---|-\ |
` | | | >--------'
` .-. .-. .---|+/
` | | | | | |/|
` | | | | | -
` '-' '-' === Vee
` | | GND
` - ==` Vee GND
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

By using a pA input low drift op amp and high precision low temperature
coefficient resistors, you might be able to get close to 0.01%
precision without too much difficulty within a lab temperature range if
your reference is up to the task. Getting it accurate to 1.00000uA,
though, would pretty much be luck, or would require careful nulling of
the op amp offset and an adjustable reference, which has its own set of
problems (like potentiometer drift with temperature, &c).

Of course, this basic circuit won't help you if you've got those
hundreds of feet of wire, a reactive load, or you have to be able to
disconnect the load hot. It also is very susceptible to ESD damage, as
you've got an op amp input floating in the breeze.

A better response awaits more information from you.

Good luck
Chris

 

[John Popelish]

John Popelish wrote:

eehinjor@163.com wrote:

hi,everybody.
I have some questions about hight precesion constant current source.My
goal is 1uA whose error is blow 0.01%.

In my circuit I used a op-amp,p-mosfet and 5V voltage reference.
But the result is so bad,the error is almost 1%.

Could somebody give me some hints?
Thanks very much.

eehinjor

So you must have a current sense resistor that drops 5 volts when 1 uA
passes through it. That would be a 200k resistor, right? In order for
the closed loop to have .01% accuracy, not only do you need sufficient
loop gain, but the opamp input impedance would have to be more than
10,000 times this 200k, or greater than 2 giga ohms. What opamp are you
using?

Damn, I have to get some caffeine in me before posting. 5 volts
reference with 1 uA current implys a 5 mega ohm current sense resistor
and an opamp impedance of 5 gig ohms for .01% accuracy. Of course,
the sense resistor also has to be accurate and stable to better than
..01% and so does the reference. The opamp offset voltage must also be
less than a half millivolt. The frequency response of such a source
will also be accurate only for very low frequencies where capacitive
current is less than .01% of the source current.

 

[eehinjor]

Thank you very much.
I know this is very difficult,I have done some experiment,but all the
result is so bad.

The resistor is 5000K.In my circuit,the op-amp is OPA602.

 

[John Fields]

On 29 Aug 2005 07:23:45 -0700, "Chris" <cfoley1064@yahoo.com> wrote:

....

From your description, you have something like this (view in fixed font
or M$ Notepad):
`
`
` VCC VCC
` | +
` .-. | -----> +
` | | '----------o
` R1| | 1uA
` '-' .----------o
` |5.000V VCC | <---- -
` o-----. + |
` | | IC2 | |
` IC1/-/ | |\| ||-+
` ^ '-------|+\ ||-> Q1
` | | >---||-+
` | .--|-/ |
` | | |/| |
` === | === |
` GND | GND |
` '------------o
` |
` .-.
` R2 | |
` | |
` '-'
` |
` ===
` GND
This
is basically a unity gain non-inverting amplifier, and is prone to
misbehaving.
^^^^^^^^^^^^^

inverting

....

---
--
John Fields
Professional Circuit Designer

 

[John Popelish]

eehinjor wrote:

Thank you very much.
I know this is very difficult,I have done some experiment,but all the
result is so bad.

The resistor is 5000K.In my circuit,the op-amp is OPA602.

The opamp is pretty good (bias current within minimum requirements and

offset voltage probably also). But what is the gate leakage current
of your PMOS transistor, and how do you build this thing with surface
leakage current to the sense node much lower than the .1 nA error budget?

 

[eehinjor]

the type of my p-mosfet is VP0610L.its gate leakage current is only
10nA(max.).

Unfouturnately VP0610L is obselete from last year,I don't know which
one can replace it or better than it.

I used a intrument of Agilient,the type is 3458A.it can measure current
as low as 1nA.

 

[John Popelish]

eehinjor wrote:

the type of my p-mosfet is VP0610L.its gate leakage current is only
10nA(max.).

Yikes. That is 100 times your leakage current budget of 0.1 nA.

Unfouturnately VP0610L is obselete from last year,I don't know which
one can replace it or better than it.

I don't know what a better one would be, but it must have lower gate
leakage and probably would be a much smaller die, maybe rated for a
higher voltage.

I used a intrument of Agilient,the type is 3458A.it can measure current
as low as 1nA.

 

[eehinjor]

Thank you very much,John.

But when I try to use this circuit to realize 1mA constant current
source,the result is also unsatisfied.

I don't know the reason,so I am afraid the circuit is wrong.

 

[John Popelish]

eehinjor wrote:

Thank you very much,John.

But when I try to use this circuit to realize 1mA constant current
source,the result is also unsatisfied.

I don't know the reason,so I am afraid the circuit is wrong.

I haven't seen your circuit, including the test setup that you are
using to get this unsatisfactory result.

 

[mike]

John Popelish wrote:

eehinjor wrote:

Thank you very much.
I know this is very difficult,I have done some experiment,but all the
result is so bad.

The resistor is 5000K.In my circuit,the op-amp is OPA602.

The opamp is pretty good (bias current within minimum requirements and
offset voltage probably also). But what is the gate leakage current of
your PMOS transistor, and how do you build this thing with surface
leakage current to the sense node much lower than the .1 nA error budget?

At what level of accuracy do you have to start worrying about
thermoelectric effects of the interconnect?
mike

--
Return address is VALID but some sites block emails
with links. Delete this sig when replying.
..
Wanted, PCMCIA SCSI Card for HP m820 CDRW.
FS 500MHz Tek DSOscilloscope TDS540 Make Offer
Wanted 12" LCD for Compaq Armada 7770MT.
Bunch of stuff For Sale and Wanted at the link below.
MAKE THE OBVIOUS CHANGES TO THE LINK
ht<removethis>tp://www.geocities.com/SiliconValley/Monitor/4710/

 

[eehinjor]

Thanks all.The circuit is shown as below.
Q1:VP0610L;
Q2:2N4402;
RL:the load(resistor);
IC1:OPA602;
R1:the resistor(5000K ohm),it is composed of 4990Kohm fixed
resistor(0.1%,15ppm/c) and a 20Kohm variable resistor;
+12V,-12Vower;
+7V:it is produce by INA105 and AD586(+5V voltage reference);
+5.5V:to avoid the voltage of RL over this value;

In fact,my goal is 1mA/100uA/10uA/1uA when R1 is 5K/50K/500K/5000K.

By the way,the VP0610L has been obseleted from last year,would you like
recommand one?

From simulation result by pspice,the circuit is right,but the real
result is so bad.I don't know how to modify it.

+12V
|
.-.
| |
R1| |
'-'
|
|-----------------------o
| +12V |
| + |
| IC1 | |
| |\| ||-+
o-------------|-\ ||-> Q1
| >---||-+
+7V--------------|+/ |
|/| |
| |
- |
-12V |-------------------o
| |
|< .-.
+5.5V----| Q2 | |RL
|\ | |
| '-'
| |
o------------------ o
|
===
GND

 

[eehinjor]

Thank you,Chris.
There is a little difference between our circuits,but the theory is the
same.

I don't know what IC1 is in your first circuit.

Maybe I am confused between precesion and accurary.I only mean the
result shoule be 0.9999uA~1.0001uA.

The temperature is from 20C to 70C,of cource the wider the better.

The load is resistor,when its value is 5500Kohm and the current is
1uA,the voltage will be 5.5V.

waiting for your response.
best regards.

 

[John Popelish]

mike wrote:

John Popelish wrote:

eehinjor wrote:

Thank you very much.
I know this is very difficult,I have done some experiment,but all the
result is so bad.

The resistor is 5000K.In my circuit,the op-amp is OPA602.


The opamp is pretty good (bias current within minimum requirements and
offset voltage probably also). But what is the gate leakage current
of your PMOS transistor, and how do you build this thing with surface
leakage current to the sense node much lower than the .1 nA error budget?


At what level of accuracy do you have to start worrying about
thermoelectric effects of the interconnect?

That not only depends on the level of accuracy, but on the reference
voltage. This design, supposedly is based on a 5 volt reference.
That is pretty huge compared to thermoelectric effects.

 

[eehinjor]

Just now I simulate the Chris's first circuit,found the result is
worse.
especially when the load is smaller than 500K,the output is not
constant.

 

[Chris]

eehinjor wrote:

Thank you,Chris.
There is a little difference between our circuits,but the theory is the
same.

I don't know what IC1 is in your first circuit.

Maybe I am confused between precesion and accurary.I only mean the
result shoule be 0.9999uA~1.0001uA.

The temperature is from 20C to 70C,of cource the wider the better.

The load is resistor,when its value is 5500Kohm and the current is
1uA,the voltage will be 5.5V.

waiting for your response.
best regards.
eehinjor wrote:
Just now I simulate the Chris's first circuit,found the result is
worse.
especially when the load is smaller than 500K,the output is not
constant.

Hi, eehinjor. Apparently you've cross posted your question over at
sci.electronics.design as well as s.e.b. It's now clear that you are
trying to make a DC current calibrator with switchable currents in
decades down from 1mA to 1uA with 0.01% precision. Here's the key post
from s.e.d.:

eehinjor wrote:

Thanks all.

For me this task is a chanllege.I have done some experiment during past
months.
My goal is 1mA/100uA/10uA/1uA,the accuracy is 0.01%,the result is that
the output can be switched between them.Then some mux-chips will bring
error.

I have tried three ways.One,single op-amp & resistor.Second,two
op-amps,one of them as the feedback.Third,the op-amp & PMOSFET.Before
experiments,I simulated them by pspice.

The first two ways can acheive 1mA/100uA easily,but they can not
realize 10uA/1uA.
So I have to change the way to the third.But until now,I can not
realize 1mA by the third way.Maybe because I am not familar with
PMOSFET.

I will buy some chips tommorrow to do some experiments on your advice.I
am sure I have to study more from all of you.

best regards.

A couple of points:

* I believe the biggest hangup you're having with your P-MOSFET circuit
is getting a 5 Meg resistor to stay stable to 0.01%. Higher resistance
values are a bear. Changes in ambient humidity are particularly
troublesome in making effective resistance values drift all over the
place. I would also be concerned about leakage current in the FET, and
possible oscillations.

* The second circuit I mentioned is only useful if you can put a
resistive load on the circuit board within a couple of inches of the op
amp. The circuit takes advantage of the "virtual ground" at the input
of the op amp. The divider of your voltage reference is set to inject
1uA into the inverting input. The op amp will work to put 1uA through
your load to balance the injected current, and keep the input at 0V.
This doesn't work if the feedback loop goes 2 meters through test probe
wires to your load or a meter, and then back. The inductance of the
leads and any noise pickup on the wires guarantees oscillation. The
simple op amp current null circuit I suggested is not useful for you.

You're discovering the hard way why current source meters with the kind
of accuracy you're specifying usually cost thousands of dollars new
(Keithley in particular is a good source for this kind of instrument).
They produce a dial-in, rock-stable current under varying loads. You
should know, however, that even these instruments are not specified to
the temperature range you specify -- they generally are only spec'ed
over lab temperature range.

If you need a calibrator, you can take hope in the thought that used
instruments are available at a fraction of the price of new. You may
be able to get something close to what you want (if you can lose the
commercial temp range) for hundreds rather than thousands of dollars.
An 0.01% dial-in or switchable current source is not a trivial project,
and not suitable for newbies.

If you still want to pursue this as a project, reread all the responses
to your posts in both newsgroups, particularly those of Mr. Popelish,
Ban, and Winfield Hill. They've given you a free helping of really
good practical advice.

Then go to instrument manufacturer websites, and take a look at what's
out there commercially, what it can do, and what it costs. These
instruments are generally a real bargain, and give you good value for
your money. You will see, if nothing else, that what you're proposing
isn't exactly a trivial newbie project. If somebody could slap
together a little perfboard circuit that does what you propose, they
wouldn't be spending thousands of dollars for one of these instruments.

Also, get the operation/service manuals on one of the older Keithley
current sources, and take the time to examine the circuits in detail.
There's a real education in doing that, and may help you in getting the
knowledge you need to approach your job. But you might want to shoot
for 0.1% at best, especially considering your temp range.

Good luck
Chris

 

[eehinjor]

Thank you Chris.
This is my first time to google's group.This question is postd at
sci.electronics.design.

During these two days,I have learn more.There are little description
about MOSFET in our book,so my fault is some knowledge about it.

On your advice,I will do more expeiment step by step until the goal.

best regards.

 

[Bob Masta]

On 30 Aug 2005 18:38:16 -0700, "eehinjor" <eehinjor@163.com> wrote:

Thank you Chris.
This is my first time to google's group.This question is postd at
sci.electronics.design.

During these two days,I have learn more.There are little description
about MOSFET in our book,so my fault is some knowledge about it.

On your advice,I will do more expeiment step by step until the goal.

best regards.

One thought you might want to consider:
I have seen super-accurate *voltage* calibrators
that work via digital counters. The basic idea is
that you use a counter with some logic to go high
for the proportion of time you want to divide the
master reference by. For example, to produce
1.0 V from a 10.0 V master, you generate a pulse
that is high for one count out of 10, then use that
to switch the master. You have to filter the result
to get DC. That's easier at higher clock rates, but
higher clock rates have more time spent in switching
which needs to be compensated. Other than that,
the clock frequency is not critical since the whole
trick relies upon duty cycle, which being digital is
rock-solid.

Application of the above to *current* sources is
"left as an excercise for the student". Just giving
you an alternative approach.


Best regards,


Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com
Home of DaqGen, the FREEWARE signal generator