unbalanced supply for op-amp in pure DC operation?

[Eliminateur]

Here's the issue:
i need to measure DC current (with a large dynamic range, i'm measuring from microamps to 10A~) with a 10A/75mV low-side shunt, tested device is battery operated and the testing rig as well(test rig would probably use a 9V battery).

So my problem is the op amp selection to amplify the shunt to 5V, all the op amps i have are NOT RtR in/out, all i have at hand are regular FET input amps and the fancies i have is a OP4277 which has a output swing of +-2V to vcc/vss and the common mode input as well, so i can't feed it from the +5Vcc of the digital circuitry if i want the full 5V output.

Now, the positive rail is no issue, i can use the +9Vbat and that gives me the clear 5V output, the problem is the ground/negative.
I've been toying with using a 555 for negative rail, but that has significant loss leaving me with ~-7.2v for +9V input.

¿what happens to the output if i feed the op amp with  +9v/-7.xV?, both circuits grounds are joined (the battery under test and the tester battery)?, for example, under no load, with 0V on the NI input of the amp, ¿what happens to the output? will it stay at 0V or will it go to 0.9Vdc(9-(9+7.2/2)) due to the imbalance?, i've googled around bu can't find a definite answer(some sites say it won't do anything, others that it will)

another option i was considering is using two 9v batteries, but ofc they won't have the exact same voltage

[graybeard]

It will work fine.  The rails can be anthing so long as the total meets the minimum, and and does not exceed the maximum voltage rating of the op-amp.

Make sure the common mode input range inad output swing will work for the chosen rails.

[T3sl4co1l]

So you're, uhh, measuring nanovolts with an op-amp?

Tim

[Eliminateur]

So you're, uhh, measuring nanovolts with an op-amp?

Tim

¿what's the problem?, i don't think i'll measure down to that scale considering it's all cobbled up together and the ADC resolution won't go so down, at least i must have sub mA accuracy(like i have in my multimeter already in the 10A range)

[T3sl4co1l]

Alright, fractional mA in the LSBs, is reasonable.  You will need a precision uV offset, probably autozero/chopper type, and yes, bipolar rails are a good idea.

A typical way to derive split rails from a single battery, is to use a voltage divider and op-amp voltage follower.  The amp's output is assigned ground, and the +/-4.5V rails are then used for whatever.

Charge pumps are a runner-up, but are very noisy so may not be so attractive here.

Tim

[Eliminateur]

Alright, fractional mA in the LSBs, is reasonable.  You will need a precision uV offset, probably autozero/chopper type, and yes, bipolar rails are a good idea.

A typical way to derive split rails from a single battery, is to use a voltage divider and op-amp voltage follower.  The amp's output is assigned ground, and the +/-4.5V rails are then used for whatever.

Charge pumps are a runner-up, but are very noisy so may not be so attractive here.

Tim

I considered using an opamp as virtual ground, but AFAIK that would kill half my usable range since i'll be bringing the output to half of vcc(for example) and since vcc is +5v it doesn't leave me with ANY usable range for the output, less than 1v (unless i use the 9V unregulated input, that still leaves me under the output swing as the OPA277 has 2v of marging either way, so it leaves me with +-2.5v at the most of output swing, not nearly enough.

i'm testing with a charge pump based on a 555 for now

[T3sl4co1l]

Why does your VREF have to be >2.5V?  Is this not actually going into an ADC?

If you're just buffering for a DMM or something, it could just as well be a 1 or 2 or 10V range I suppose.  But then I don't know why you're concerned about digital supply range if not for an ADC.

Asymmetrical rails are a good idea, for an otherwise single-supply application, especially considering it should remain operational down to say 6V or so, otherwise you're leaving a reasonable bit of unused charge in the battery.  A +4/-2V supply with 3.3V digital supply would be quite workable.

Tim

[nigelwright7557]

I use ICL7660 for negative rails.
OK for a few mA.

[David Hess]

i need to measure DC current (with a large dynamic range, i'm measuring from microamps to 10A~) with a 10A/75mV low-side shunt, tested device is battery operated and the testing rig as well(test rig would probably use a 9V battery).

10 microamps through a 7.5 milliohm shunt is 95 nanovolts.  This level of resolution is impossible in this application but not because of limits in operational amplifier performance.  This will be the domain of chopper stabilized operational amplifiers and the lowest drift bipolar low noise operational amplifiers.  Or combine the two for the best characteristics of both and use several sets in parallel to further reduce noise.  The real problem will be thermal effects in the shunt itself which will swamp this level of precision no matter what you do.

¿what happens to the output if i feed the op amp with  +9v/-7.xV?, both circuits grounds are joined (the battery under test and the tester battery)?, for example, under no load, with 0V on the NI input of the amp, ¿what happens to the output? will it stay at 0V or will it go to 0.9Vdc(9-(9+7.2/2)) due to the imbalance?, i've googled around bu can't find a definite answer(some sites say it won't do anything, others that it will)

Besides characteristics which may vary due to the total supply voltage, there are two effects when the supply voltages are not identical to the test conditions listed in the specifications:

1. Different but still symmetrical levels affect the offset voltage through the power supply rejection.

2. Asymmetrical levels affect the offset voltage through the common mode rejection.  Using tracking power supplies avoids this error.  Of course if the common mode input voltage changes, the same thing happens but then bootstrapping the supplies or using the inverting configuration can correct even this.

Whether these matter depends on your requirements for precision and the application.  In the worst case, the additional offset voltage can be nulled out assuming that the supply voltages are stable.

[Eliminateur]

Why does your VREF have to be >2.5V?  Is this not actually going into an ADC?

If you're just buffering for a DMM or something, it could just as well be a 1 or 2 or 10V range I suppose.  But then I don't know why you're concerned about digital supply range if not for an ADC.

Asymmetrical rails are a good idea, for an otherwise single-supply application, especially considering it should remain operational down to say 6V or so, otherwise you're leaving a reasonable bit of unused charge in the battery.  A +4/-2V supply with 3.3V digital supply would be quite workable.

Tim

yes, the input range of the ADC being 0-5V, if i use the floating supply i cut that in half, right?