I'm sure this question has come up before...

I need to convert +5V from my power supply (a USB cable) to -5V to drive the negative supply of some op-amps.

I found a couple of circuits that do this by converting DC to AC with a 555 and rectifying it:
http://www.ecelab.com/circuit-polarity-inv.htm
http://www.aaroncake.net/circuits/vinvertr.asp
http://ourworld.compuserve.com/homepages/Bill_Bowden/page10.htm

Trouble is, I'm not sure if these circuits can supply enough current for what I need. I have three op-amp ICs (two LM324s and one LM339) and will use all 12 op-amps. The LM324 datasheet says '0.18 mA supply current per channel' and the LM339 datasheet says '0.2 mA supply current per channel.' So, since I'm using all 12 channels, I believe the op amps will draw (0.18*8 + 0.2*4) = 2.24mA total.

The circuit in third link can supply 12mA, so I would assume the first two could supply at least a few milliamps. Is this feasible? Would I be better off getting a DC/DC converter? There are so many options that I wouldn't know which one to choose, and they're not cheap.

I'm probably just going to wire up each of these circuits and see what I get. What's the worst that could happen if I wind up drawing too much current from the 555?

 

What is your application?

You may not need a dual supply.

 

I need to take the difference of two voltages and integrate it. The difference could be positive or negative, so a negative supply is required.

 

I built Bill Bowden's circuit, and it works, but doesn't generate a sufficiently negative output voltage.

The 7660 IC looks promising. The datasheet doesn't say how much current it can output, but posts in newsgroups suggest it's somewhere in the tens of milliamps. Plus, I should be able to get it at a local store.

 

You can use two op amps for each signal and bias them to work with a single supply and then use a difference amplifier to caclulate the difference.

 

You can use two op amps for each signal and bias them to work with a single supply and then use a difference amplifier to caclulate the difference.

I could do this, but my application uses a lot of op-amps already and the ability to deal with negative voltages would greatly simplify the circuit.

I read the 7660 datasheet more closely and determined that I can draw about 35mA from the negative output before it gets above -3V. This should work for me.

 

Another option is to use DC to DC converter which produces a negative voltage from a positive voltage.

 

USB can deliver 500mA. You could just do a silly voltage divider through some power resistors, 10R or something.

You would have to tie your signal ground to the midpoint of the 10R voltage divider, so it would have to be a floating ground.

Then if you pull 30mA out of the 250mA already being driven, you'll still be roughly +-2.5V.

The obvious question then becomes, is +-2.5V enough?

 

My first thought was to split the 5V into ±2.5V, but these limits aren't sufficient for what I need. My input voltages could be ±2.4V, which would be too close to the rails.

I picked up an ICL7660 DC-DC voltage converter today (living in Silicon Valley is awesome) and it looks like it'll work.