# difference amplifier offset trimming

Discussion in 'Electronic Design' started by john, Apr 17, 2007.

1. ### johnGuest

Hello,

I am using texas instrument's part number INA2133U and OPA 131 to
build the constant current source. The following link will take you to
the circuit and data sheet ( page 13, figure 16). The inverting input
of the amplifier is grounded.

I tested the circuit for Leakage current and with zero input voltage ,
load resistor of value 100kohm, R = 50kohm and rail voltages of +/-
15volts, I am getting current in the range of 1 to 10 micro amp, which
is undesirable for my application. The questions are as follows

1. Do I need offset trimming like the data sheet discussed it on page
number 9?
2. Is there any other way to stop this leakage current?

The desirable leakage current range is less than 1 micro amp.

John

2. ### martin griffithGuest

How long would it take to solder in a trimpot and find out?

martin

3. ### WinfieldGuest

john <wrote:
I calculate that your measured output "leakage" current should be
less than 0.075nA, based on the maximum offset voltages of the
ina133 of 0.45mV, and the opa131 of 0.75mV, summed and divided
by 50k, plus 0.05nA maximum bias current from the opa131. Solve
that big discrepancy, before thinking about zero-trimming circuits.

4. ### johnGuest

Hi,

I connected the load of 100kohm resistor at the output and measured
the voltage across the resistor using voltmeter. Is it a good method?

John

5. ### Winfield HillGuest

Is everything well shielded? Do you have AC pickup (75pA
is a pretty weak current, implying high-Z, subject to pickup
and rectification)? Have you looked at the various signals
with a scope probe? It's not the "schematic" part of a circuit
that can get you into trouble, as much as the parasitic part.

6. ### johnGuest

Hi,

What if the 50Kohm resistor is replaced by 5kohm resistor? will the
leakage current go higher?

John

7. ### Winfield HillGuest

The offset-voltage contribution will be 10x higher, yes,
but it's still under 240nA, low compared to your value.

8. ### johnGuest

Hi,
I tested the circuitry again and I am getting the leakage current with
5kohm resistor of approx. 240nA. Is there any way that I reduce it
less than 0.075nA.

John

9. ### Winfield HillGuest

You can program up to 1 or 2 mA with a 5k resistor, and the
0.24uA zero-current error you mention observing is 1/8000 of
that, which is pretty good. If you add a little zeroing (which
can be on the input-voltage side) you may be able to improve
that by 10x or more. This gets you into the 10ppm area, but
the matching errors of the difference amplifier may also begin
to cause trouble and need attention.

If you want 75pA zero accuracy in a 2mA full-scale circuit,
that's only 0.04 ppm. To attempt to achieve that you'd need
an entirely different approach, a true current source made with
precision electronics, and perhaps with range switching, etc.

What are you trying to do?

10. ### johnGuest

Hi,

If you want 75pA zero accuracy in a 2mA full-scale circuit,
that's only 0.04 ppm. To attempt to achieve that you'd need
an entirely different approach, a true current source made with
precision electronics, and perhaps with range switching, etc.

So, the current source that I am using is not a true current source.
Would you advice that what is a true current source and how can I
build it?

What are you trying to do?
I am trying to build a system that can stimulate Lab. rat's retinal
tissues. Thats why I am concerned about leakage current.

Thanks
John

11. ### Winfield HillGuest

Those that use the collector of a transistor or the drain of
a FET for the output current, with an opamp providing the
programming control.
Aha! Then do you need really 2mA fullscale? If you use a
higher resistor value then offset voltage is less of an issue.
Maybe you can have a range-switching scheme?

But I wonder if your concern is over leakage to ground, which
could be eliminated with a battery-operated current source?

12. ### Marte SchwarzGuest

Hi Winfield,
No, in long term stimulation there would be a problem with dc leakage
current, too. Like in all good TENS the OP should stimulate biphasic and
dc-free. This is easyly done tith a output transformer.

Marte

13. ### johnGuest

Hi,
Can you advice any circuit diagram or FET ?
How a battery operated current source can not have a leakage current?

John

14. ### johnGuest

No, in long term stimulation there would be a problem with dc leakage
Can you advice the transformer setup? Will the transformer be able to
handle frequencies in the range of 1kHz to 40Khz.
current amplitudes between 1uA to 1mA?

John

15. ### WinfieldGuest

I like the transformer suggestion. If you feed it with a current
source, the output will also be a current but without any DC
component. A 1 to 40kHz range is easy, you can get three
or even four decades of frequency range if necessary. There
is little practical low limit to the currents that can be handled.

16. ### Marte SchwarzGuest

Hi John,
With such "easy" things you can use simple audio transformers. May be ISDN
transformer will work too. just try out. This is not a big thing.
If you want to do this work "science" like, then go to a manufacurer's site
and use their transformer calculation tool for their cores.

Marte

17. ### Marte SchwarzGuest

Hi Winfield,
That ist the typical way to do this in transcutan stimulation (TENS). Other
way is to couple out via capacitor. But a transformer is prefereable. Often
it is seen in a saturated coil design. Just interrupting a few mA flow
through a primary coil and plug the secondary winding to patient. But this
one is not prefereable for use in battery powered devices.

Marte

18. ### johnGuest

Hi,

I found the following transformer

http://www.vishay.com/docs/34063/lpe3325.pdf

What do you think about it... Is it appropiate to pass peak to peak
1nA to 1mA of AC current. I am also confused aboout grounding the
primary and secondary of the transformer. Would they have same ground?

John

19. ### Marte SchwarzGuest

Hi John, This is only a inductor, no transformer. There is only one winding. You need
two. If you want good and stable signals then I would use a little bigger
cores. What is the impedance that you expect to drive (including
electrodes)? The core dimension is depending at voltage-time-product not
only the current.

Marte

20. ### johnGuest

Hi Marte

The impedances are between 100Kohm to 1Mega ohm.

John