What is the best way to remove DC offset

A battery of two AA alkaline cells for 3 volts total.

I'm trying to decide if you're being subtle, since the transformer may behave poorly with 3V DC feeding it.

I'm curious why the cap divider or AC coupling failed.

Cannot speculate on solutions while half the problem is unstated.

A few clues about your real problem would help us understand.

John Fields does not do subtle.=A0

Me too.

Sure we can - and do it all the time, even though we shouldn't waste the bandwidth. The speculations are mostly useless and irrelevant.

When the original poster tells us something about the background to their problem and why they want to solve it, we can often be a lot more helpful.

-- Bill Sloman, Nijmegen

The original circuit is as follows: astable multivibrator -> unity gain buffer -> capacitative divider ->

unity gain buffer The astable multivibrator is being excited by some variable slope ramp, the output at the end of the capacitative divider looks fine, but the output of the second unity gain buffer is just a flat -ve offset. The output of the astable multivibrator has a +ve DC offset that is to be gotten rid of.

Air gap ?

Remove the capacitive divider. It may be forming a charge pump in combination with one diode at a terminal of one digital IC. That cyclically boosts a voltage. Instead, use a resistive divider or op amp at low gain.

You need a DC path for the input of the second buffer or the bias current will be integrated by the capacitive divider and it will drift into saturation. A relatively large value resistor to ground will do.

You're playing with fire loading the first buffer with a capacitive divider too.. that will reduce the phase margin and it may oscillate.

Best regards, Spehro Pefhany

P.S.

You can avoid both issues in one swell foop by capacitively coupling into a resistive voltage divider.

Best regards, Spehro Pefhany

Here is a framework *to start with*, Daku.

When posting a question, always answer the 'big six questions': *Who, What, Where, Why, When, How Much* You will only know which answers were important when your mystery is correctly solved.

You answered 1-1/2 questions of the six.

Who: Daku. Example: Tech in a college lab

What: Example: Interface between an 'optical color sensor front end' and a small 3 V microwave telemetry module with 120 ohm pullups in the transmitter

Signal with a troublesome DC offset. * Normal amplitude of the signal? Example: +- 50 mV p-p

• Type of signal? Example: Differential NRZ, 10 KHz

Where: Example: All on a wood lab bench isolated from other experiments

Why: Example: Plant biology experiment Recording leaf color changes

When: Example: Operated normally when I powered it from a lab supply for the 'color sensor' and two "D" cells for the transmitter, but +3 V DC offset appeared as soon as I removed the cells and plugged in a 'wall wart' to power the transmitter :)

How Much: Example: +3.0 V DC relative to earth ground

--Winston

On May 12, 8:25 am, Spehro Pefhany <

Thanks a lot. This did the trick.

You're welcome!

So, you are going to find a pair with exactly 3 volts total? Measure the terminal voltage of a new alkaline cell and you'll see that it's higher than 1.5 volts.

Please specify the type of data that is on this signal? Clean analog? Digital ? etc.. It makes a difference on the method used.

They make virtual grounds that can split the power supply so that your common is actually at center point. This means you need to have a isolated power source for this one circuit.

If you don't want to go that route, you can create a small (-) voltage source to network against the +DC offset you have via a divider.

But, like I said, with out knowing a little more of what you have, it's hard to select the proper way. Because I was, myself, employ a op-amp with minus offset uses as a buffer+amp if needed.. Hell, with rail To Rail types, one could use a dual type, one unit for the (-) volt generator and the other for the buffer+gain..

Jamie

I once designed a transistor driver for a piezo strip using a couple batteries so the circuit could work off a single single hv supply.

Ac coupling needs grounding resistance after the cap.

Greg

Or alternatively use a low pass filter to remove any fast signal variations and hence a DC reference is created. Using a differential amplifier across the LPF will show the fast variations (AC) only.

Not common in electronics design, but this approach is used much in image processing to remove illumination variations e.g. before edge detection.

The circuit you describe is sounds like what is referred to as a "servo" input on high-end audio equipment. Such gear typically claims "DC coupled inputs" to alleve audiophile fears of musicality corruption from the dreaded input coupling capacitor in the signal path.

The idea is that the signal to be amplified does indeed pass through a DC-coupled stage, which also subtracts off a low-passed version of the signal from a parallel path. The LPF capacitor (typically in a "leaky integrator") is seeing the input signal, but essentially filtering it out completely to leave only DC (or maybe a fractional Hz, etc).

This allows AC coupling using a rather small and cheap capacitor... but I suspect the main advantage is that they can claim it as DC coupling. That removes any question that might have arisen about a conventional AC coupling circuit.

Otherwise, they'd need to use an exotic cap made out of passivated platinum titanate or something, and claim that it was better than their competitor's organically grown nanotubular zirconium capacitor.

Best regards,

Bob Masta DAQARTA v6.01 Data AcQuisition And Real-Time Analysis