Remove DC offset from analog signal

[Heher]

I am redesigning an oddball signal conditioner and I need a little help. Does anyone know of any method to reliably remove DC offset from the input signal?

The overall idea of the circuit is to take an input waveform, remove the noise, and output a standard 0-5V TTL square wave of the same frequency as the input waveform. Phase shift is not an issue with this ciccuit and can be ignored. The input waveform will be in the range of -10V to +10V with an amplitude as low as 100mV. It will be a square wave, sine wave, or triangle wave from 10Hz to 10kHz.

The problem lies in the fact that the input wave is usually riding on a large DC offset (up to a few volts) when the signal comes in. I need to remove this DC offset prior to my pre-amp and filter stages of the signal conditioner. Does anyone know of a good method for removing DC offset without using a trim pot?

I have tried using passive RC high pass filters amongst other solutions but I am not able to remove enough of the offset to get a signal that I can work with.

Anyone have any ideas that might be able to help me out? Thanks in advance!!

 

[OperaHouse]

Is there some reason you can't do it with just a blocking capacitor? If you need something more active, you can do a long time constant minimum detector and subtract that value with an op amp. That would wor if you wanted to keep signal from zero to positive value. Subtracting the value of an intergrator would give you an average value centered atound zero volts.

 

[melone]

As OperaHouse suggested, just use a series cap. The value will need to be chosen based on the frequency of the signal that you need to pass....

 

[Heher]

Since my input signal frequency is 10Hz to 10kHz wouldn't this require an extremely large value for the cap to ensure the frequency response I am looking for?

 

[IRstuff]

How was it done originally?

TTFN

 

[Heher]

I've got to admit guys...I feel like an idiot now! :~/

I was simulating the circuit and was basically looking at the output prior to the capacitor charging up! The 10Hz high pass filter works great - when you let the cap charge up! LOL

Sorry for the wild goose chase, I'm sure I'll be back with more brain farts soon!

Thanks to OperaHouse and melone!

 

[melone]

I think that there is a MUCH simpler, and robust solution....

Input signal (w/DC offset) ----||----- Output signal (n/DC offset)

 

[Heher]

I don't know if subtracting one SM resistor from a simple RC passive filter and going to just a blocking cap makes it MUCH simpler but I think I would need an extremely large value for the cap at 10Hz to 10kHz, wouldn't I?

 

[sreid]

All of us have been caught out at one time or another by forgetting that there are start up transcients which are different than the steady state case. In your problem you probably have a high pass break point of about 1 Hz to reliably pass a 10 Hz sine wave. It would then take about a second for the start up DC trancient to decay. Can you simply wait one second after the signal is applied before you start looking at the squared up signal?

 

[Heher]

sreid,

Yes, that is what I am doing at the moment and it seems to be functioning correctly now. Once it reaches steady state everything looks great.

 

[peebee]

I agree with melone and OperaHouse -- why do have the resistor in there? Is it specific to your RC filter, or is it stray resistance somewhere else? Seems to me things would work better if you scrapped the resistor.