30+ volt power supply, used as a -15,+15 volt project.

[Supper_slash]

Here is my dilemma, I have a basic project that is based on the TL072 chip.
It requires a +15v and a -15V rail. I have a 30-volt buck converter that I was going to use a resistor voltage divider to create a grounding point at the 15-volt mark and use for a ground. Will this be a workaround?
As far as the power requirements the converter should more than be able to handle the small load that the circuit demands. I am more or less looking for confirmation that if I create a stable voltage divider at the 15-volt mark {half of the power supply's power output} then use that as a ground reference for the circuit, the TL072 would just read the rails as +15,-15-volt rails.

Here is the circuit that will be under test.

[colinza]

You could bias your input signal up after decoupling it and then you don't need a negative ground rail and then decouple the output. Alternately you can implement a virtual ground using an op-amp, tons of examples if you google search it.

[Zero999]

I don't understand what you mean by the 15-volt mark. The usual way to do this is to add a potential divider and treat the centre point of the potential divider as 0V, with either side of it being +V and -V.

[Supper_slash]

That is just what I had in mind Hero999. I just wanted to be sure I was thinking correctly. Some times when you are meshing 4 projects together you can start getting a bit meshed up yourself.
Once I get a working model together I hope to post my finished project up for the community.
Thank you again.

[bd139]

I'd use a TLE2426 rail splitter virtual ground. It can source/sink 20mA unlike voltage dividers and doesn't have a horrible startup time constant like decoupled voltage dividers. In your design there's an 88k resistor/cap combo connected to the virtual ground rather than very high impedance inputs so you need a relatively low impedance ground here. TLE2426CLP is a nice TO-92 package as well.

Datasheet: http://www.ti.com/lit/ds/symlink/tle2426.pdf


You could use another opamp as a unity gain follower driven from a voltage divider though.

[Supper_slash]

bd139 That's just what I was looking for thank you.

[Zero999]

The turn on transient can be minimised by bypassing both R1 and R2 in the potential divider.

An op-amp or the TLE2426 can be used but is it really necessary? The current into the 0V node in this circuit is due to the bias currents in the TL072 which are tiny, so a potential divider should be adequate.

[drussell]

The current into the 0V node in this circuit is due to the bias currents in the TL072 which are tiny, so a potential divider should be adequate.

Yeah, a simple divider with a capacitor across each of your new virtual +/- rails is more than adequate for a TL072. 

I've done that many times successfully. 

[bd139]

That'd work for simpler cases but the following node needs a lower source impedance than the divider will give it:



A follower after a voltage divider or TLE2426 will do the job. I like the TLE2426 because it's stable with capacitive loads and has very low quiescent current of 170uA compared to a divider stiff enough to hold that at low impedance.

[Zero999]

That'd work for simpler cases but the following node needs a lower source impedance than the divider will give it:



A follower after a voltage divider or TLE2426 will do the job. I like the TLE2426 because it's stable with capacitive loads and has very low quiescent current of 170uA compared to a divider stiff enough to hold that at low impedance.

No, that node does not need a lower impedance, than the divider will give it. The 18pF capacitors prevent any DC current flowing in to or out of the potential divider. The maximum total current into the potential divider is equal to the total bias currents of two op-amp input pins, which is 2×7nA = 14nA worst case, over the entire temperature range of the standard TL071xC, or 2×50nA = 100nA over the temperature range of the TL072M.

A potential divider with 1M resistors, each with a 1µF capacitor in parallel, will give a low enough impedance for your circuit. It will only consume 15µA, less than a tenth of the power taken by the TLE2426 and tiny fraction of what most op-amps use.

[bd139]

Damn it. You're spot on. I'm being an idiot! 

Going to sit at LTspice for a few minutes and make sure my brain is fixed.

[Zero999]

Go easy on yourself. I made a mistake too. I forgot that there are two op-amp inputs at the potential divider node, so the worst case bias current is double what I said previously (I've edited the post, adding 2×). Still it should be low enough not to be a problem for the potential divider, which has an output impedance of 500k, so even with the worst case bias current of 100nA (though 130pA is more typical), the 0V node will only rise by 50mV.