# Varying DC Offset

Discussion in 'Electronic Basics' started by bubbles123, Aug 1, 2007.

1. ### bubbles123Guest

Hi!

I'm getting an output signal (from my transimpedance amp) that has a
varying DC offset. The maximum size of the offset is about 200mV but
varies depending on whether there is a current or not. Ideally, I
would like to stabilise the offset or at least reduce the variation
down to about 50mV. Can someone suggest how I can do this?

Thanks!

2. ### Tom BruhnsGuest

Transimpedance, like current to voltage converter? What is the
source? Is it DC coupled to the amplifier? (If it is, how can you
tell what's an offset and what's the voltage because of a DC current
from the source, when there's a source current?) What's the rated
input bias current of the op amp? What size feedback resistor?

In other words, what is the source of the offset? Is it caused by op
amp input bias current, or is it caused by a source resistance that's
low enough to give you significant voltage gain from the input offset
voltage to the output, or is it caused by a DC current from outside
(like from the source)? Or--is is caused by rectification of some
large RF signal, or some other "parasitic" effect?

Suggest you first identify the source of the offset, and attack it
there, rather than putting a band-aid on the wound that may never
heal, so to speak.

Cheers,
Tom

3. ### bubbles123Guest

Hi Tom,

This is the circuit I'm using:

IR Photodiode -> Transimpedance Op Amp -> LPF -> Capacitor for DC
block -> Voltage Divider to fix offset -> Buffer etc.
So the problem lies in the voltage divider's offset when the distance
between the infra-red tranceivers is varied.

I assume that this is due to a variable current from the
transimpedance op amp.

I've tried to use a zener diode instead of a voltage divider to
overcome this variable offset, but the zener just sinks far too much
current and is impractical.

Also the source is an IR photodiode. I'm using Burr Brown OPA380 as
the transimpedance amp, with a 10k feedback resistor. The input bias
current is typically 3pA, however I doubt that this is the cause of my
problem.

Any help/suggestions would be greatly appreciated!

4. ### Tom BruhnsGuest

It doesn't make any sense to me that you'd be getting a DC offset
through a capacitor. Is there a way you can post a schematic? It
_sounds_ simple enough, but something is missing. What kind of
capacitor is it? Could there be any DC leakage in the capacitor? I'm
assuming the DC voltage at the transimpedance amplifier end (the LPF
end) of the capacitor is higher than at the output end...

Cheers,
Tom

5. ### Guest

Phenomena that don't make any sense around high gain amplifiers often
make sense when you look at the amplifier's output with a fast scope
and see that it is oscillating.

Signficant amounts of AC - more than a volt for FET-input amps, more
than about 30mV for bipolar inputs - at frequencies higher than can be
followed by feedback from the output, drive op amp input stages into
non-linearity, and can produce DC offsets.

6. ### Phil HobbsGuest

This makes no sense. How do you get a DC offset through that blocking
capacitor? And what does a voltage divider do to fix it?

It sounds as though you're trying to use a pot to force the buffer stage
to have zero offset, and the buffer's input current isn't constant with
signal. Are you using an emitter follower as the buffer?

Cheers,

Phil Hobbs

7. ### bubbles123Guest

The capacitor I'm using is of the ceramic type. Without the capacitor,
I get a DC offset of approximately 2.5V. The capacitor removes most of
the offset so that I get an output signal centred at approximately
200mV but this varies depending on how much current the photodiode is
sourcing. Simply put, I want to remove this variation or at least
minimise it such that it is never more than 50mV. I've used the
voltage divider to try and stabilise the offset variation (two
resistors connected to +/- 5V) but it doesnt seem to be working. My
guess is that the photodiode current varies slightly causing the
output voltage at the voltage divider to also vary slightly, thus
producing the slight offset variation. The buffer is being used as a
voltage follower before the comparator. I anticipate that the buffer
input current should not be constant with signal, as it depends on the
amount of current being sourced by the IR photodiode (which is
dependent on the intensity of IR light received). However, my issue is
not to do with the buffer, more so how to ensure that the
transimpedance output signal is centred at 50mV or less.

8. ### Phil HobbsGuest

If there is no circuit path from the TIA to the buffer that doesn't pass
through a good quality capacitor, then the DC photocurrent can't affect

Maybe you'd better post a schematic at alt.binaries.schematics.electronic.

Cheers,

Phil Hobbs

9. ### Tom BruhnsGuest

What are the values of the resistors in the voltage divider? What
kind of buffer amplifier are you using? Have you tried a Mylar/
polyester capacitor in place of the ceramic? (I've known some
ceramics in the old days long ago to be quite "leaky" though modern
ones aren't too bad as a rule.) If the input to the voltage divider
comes through the capacitor, and the output goes to the buffer
amplifier, and the "bottom" of the divider is grounded, and you have a
good high-impedance volt meter, what DC level do you measure at the
input to the voltage divider? What DC level at the output? Is the DC
divided by the expected voltage divider ratio? How, exactly, are you
determining the DC level?

I ask these questions thinking that (1) the capacitor could be bad, or
(2) the buffer amplifier input bias current could be causing the
divider unequally on positive and negative going parts of the
waveform, or (4) you're not measuring quite what you think you are.
There are other possibilities, but these are the first that come to
match what you actually have," so again, a schematic could be helpful.

Cheers,
Tom

10. ### neon

1,325
0
Oct 21, 2006
DC offset on amplifiers come from inheretance of the particular amp. the missmatch of input voltages on the diodes hat will get multiplied by A. The other is miss-match input inpedances. if the f/b resistor is very hi in value then there is an additional offset caused by current offset. without a schematic to look at it can be all of them compounded.

11. ### JamieGuest

How about correcting the route of the problem?

like for example, Is power supply also moving that much?
are you using shielded Twisted balanced pairs ? etc...

questions, questions etc...