Invert Battery and Maintaining Low Noise

[dmorris619]

I am trying to invert the the output of a battery so I can run an op-amp double ended with 5 volts and -5 volts. The battery itself produces 7.5 volts so I am passing it through a linear regulator to drop it down to 5 volts with low noise, at the expense of heat/efficiency. This is all great until I try to invert the voltage from the battery for the other rail, I want to pass that through ICL7660CPA which will invert the signal, but I am afraid that by using this I will put too much noise on the rail even if I pass it through the regulator and use large decoupling capacitors.

To give some perspective on why low noise is so important in this design is because our input to the double ended op-amp is on the scale of microvolts and our output will be on the scale of millivolts. To add to that a small fluctuation will give results that are not accurate enough, so we are trying to minimize noise from the device as much as possible. We also cannot use capacitors that are very large because the device will be repeatedly turning on and off and the charge time required could prove to be a problem.

Thus I am looking to see if anyone can provide two things, an explanation of just how the inverter proposed above works and if there is another less noisy alternative (other than a second battery) to the proposed inverter.

Thanks in advance

 

[berkeman]

I am trying to invert the the output of a battery so I can run an op-amp double ended with 5 volts and -5 volts. The battery itself produces 7.5 volts so I am passing it through a linear regulator to drop it down to 5 volts with low noise, at the expense of heat/efficiency. This is all great until I try to invert the voltage from the battery for the other rail, I want to pass that through ICL7660CPA which will invert the signal, but I am afraid that by using this I will put too much noise on the rail even if I pass it through the regulator and use large decoupling capacitors.

To give some perspective on why low noise is so important in this design is because our input to the double ended op-amp is on the scale of microvolts and our output will be on the scale of millivolts. To add to that a small fluctuation will give results that are not accurate enough, so we are trying to minimize noise from the device as much as possible. We also cannot use capacitors that are very large because the device will be repeatedly turning on and off and the charge time required could prove to be a problem.

Thus I am looking to see if anyone can provide two things, an explanation of just how the inverter proposed above works and if there is another less noisy alternative (other than a second battery) to the proposed inverter.

Thanks in advance

Well, the lowest noise solution would be to use two batteries. Is that an option?

If not, definitely invert and post-regulate with a linear regulator. I'd recommend using a low-dropout regulator, so it has a good Vin-Vout difference.

 

[Topher925]

I work a lot with low noise analog circuits for very sensitive instrumentation. A lot of my projects involve working with femptoamp currents and nanovolt potentials and I run into your problem a lot. Usually the way I get around it is by using an external power supply that has an inverted rail, maybe at -15V, regulate it to -12V, follow that with an LC filter, regulate it again to what ever voltage I need, then follow with another LC filter.

In your case I would probably look at using an inverter with a very high switching frequency, as high as possible. Like 500KHz, minimum. I would then follow that with two LDOs in series. Even 79XX's are pretty good at cutting out noise but won't like large low frequency transients. You said that you can't use large caps but what about an LC filter? Do you have any room/budget for some inductors?

 

[dmorris619]

I work a lot with low noise analog circuits for very sensitive instrumentation. A lot of my projects involve working with femptoamp currents and nanovolt potentials and I run into your problem a lot. Usually the way I get around it is by using an external power supply that has an inverted rail, maybe at -15V, regulate it to -12V, follow that with an LC filter, regulate it again to what ever voltage I need, then follow with another LC filter.

In your case I would probably look at using an inverter with a very high switching frequency, as high as possible. Like 500KHz, minimum. I would then follow that with two LDOs in series. Even 79XX's are pretty good at cutting out noise but won't like large low frequency transients. You said that you can't use large caps but what about an LC filter? Do you have any room/budget for some inductors?

I think this design could work, the only difference is that I will be working with batteries rather than a power supply. I could definitely put in some inductors provided they are not excessive large. I apologize for using such vague terms as excessively but as of right now I don't know what the turn on time of my device has to be. So let's say I want my turn on time to be around 2mSec and go from there.

Could you possibly suggest some inverters that have a high switching frequency? Also would the LDO's regulate to the same voltage or would they drop the voltage in stages? I definitely have room to place large capacitors and inductors on my board to help smooth things out but I am not sure I have the time required because in order to preserve battery life we will be turning everything but the microprocessor off and then turning everything back on, I just remembered that capacitors store their charge even when not connected to a voltage source, so maybe large capacitors can be used provided I find a way to keep them from discharging.

 

[dmorris619]

So I think I found an inverter that fits my needs, its the LTC1550CGN inverter. It has an input between 4.5 and 6.5, so I'll have to run the battery through a positive regulator first and then through this. It says that the output is adjustable in the 16pin model by attaching a voltage divider to the IC. Do you guys think this is a good choice for an inverter following the method of regulating it to 5 volts, then inverting it with a voltage divider attached to increase it to 5 volts at the output and then putting it through a series of regulators to smooth out the output?