Voltage dividers on power supply rails

[new299]

Hi all,

I've been presented with a design that has voltage dividers on the power rails. These are used to take +/- 5v down to +/- 3v (100,150Ohm IIRC). They are used to drive the rails of an opamp. The opamp is a precision transimpedance amplifier for signals in the pico to nanoamp range (LMP7721). It draws about 1mA. My tests show that current can vary by about 500uA with input signal (mostly when it starts hitting the rails). Current draw is supposed to vary by ~1mA with temperature.

The LMP7721 datasheet suggests using a regulator and filter network on the rails, but not a voltage divider.

I've always understood that using voltage dividers on the supply is bad practice. It does appear to work with this part. But I've never seem it done. I can't see any advantage over using a voltage regulator (aside from a slight cost advantage).

I feel like using voltage dividers here is bad practice and has the potential to cause other issues down the line.

Am I just crazy? Can anyone show me an example where this has been done and was a reasonable design choice? Or have any other input as to why this might be a good/bad idea?

[bob91343]

The problem with a voltage divider is that it has a high impedance, and thus its output voltage will vary with current load.  This may not be acceptable for your design.

Depending on how much variation you find acceptable, there are several schemes that will work.

The obvious zener diode.  Temperature instability and soft knee.

Series diodes, similar issues.

Active regulator.  Not simple, takes many components.

In any case, if the load doesn't change much then a voltage divider may be the best solution.

[mikerj]

What resistor values are used for the divider?  If the current through the divider is significantly higher than the op-amp current then voltage change will be small, but obviously this increases power consumption.

Presumably the divider is well decoupled with a suitable capacitor? If not it's inviting performance problems such as instability.

[Zero999]

It sounds like a bad idea. The LMP7721 is also only specified to a total power supply voltage of 5.5V and +/-3V is a total power supply voltage of 6V.

[OM222O]

It sounds like a bad idea. The LMP7721 is also only specified to a total power supply voltage of 5.5V and +/-3V is a total power supply voltage of 6V.

That's not correct. Page 4 of datasheet, absolute maximum ratings:
"SupplyVoltage (VS= V+– V−)(3)    –0.3    6.0V"
"(3)The voltage on any pin should not exceed 6V relative to any other pins."

so +-3V supply is within the specified limits, albeit a bit dangerous.

[Zero999]

It sounds like a bad idea. The LMP7721 is also only specified to a total power supply voltage of 5.5V and +/-3V is a total power supply voltage of 6V.

That's not correct. Page 4 of datasheet, absolute maximum ratings:
"SupplyVoltage (VS= V+– V−)(3)    –0.3    6.0V"
"(3)The voltage on any pin should not exceed 6V relative to any other pins."

so +-3V supply is within the specified limits, albeit a bit dangerous.

No it is correct. From the data sheet.

Absolute Maximum Ratings indicate limits beyond which damage to the device may ccur. Recommended Operating Conditions indicate conditions for which the device is intended to be functional, but specific performance is not ensured. For ensured specifications and the test conditions, see the Electrical Characteristics Tables.

In other words, it won't blow up with a supply voltage of 6V, but it's not designed to work at that voltage. -0.3V is within the abosolute maximum rating, but don't complain when it doesn't work at that voltage.

[oschonrock]

No it is correct. From the data sheet.

Absolute Maximum Ratings indicate limits beyond which damage to the device may ccur. Recommended Operating Conditions indicate conditions for which the device is intended to be functional, but specific performance is not ensured. For ensured specifications and the test conditions, see the Electrical Characteristics Tables.

In other words, it won't blow up with a supply voltage of 6V, but it's not designed to work at that voltage. -0.3V is within the abosolute maximum rating, but don't complain when it doesn't work at that voltage.

Yeah, operating at "absolute max" is always a bad idea. Zero room for variance or error.

Re OP: Resistor voltage dividers are not normally used (or suitable really) to "supply power" to anything. Too many disadvantages. In order for the voltage to be stable under load you need ~ 10x the load current flowing down your divider. Very inefficient, and still not stable. Active voltage regulator is a much better choice.

What is done sometimes is to "buffer the output of a voltage divider" with a non-inverting op-amp unity gain voltage follower. Even then, that buffered voltage is not often used to "supply power" to other devices, but commonly used to provide, for example, a "half-rail" reference voltage, which might be used in multiple places. For more accuracy people often use voltage references in the same way, ie buffer them with a voltage follower, then distribute to where they are needed.

TLDR: to "supply power" to something, use a regulator, most commonly for an opamp (particularly a precision one) use a low noise LDO.

And yeah, don't design to operate at "Absolute max".

[new299]

No it is correct. From the data sheet.

Absolute Maximum Ratings indicate limits beyond which damage to the device may ccur. Recommended Operating Conditions indicate conditions for which the device is intended to be functional, but specific performance is not ensured. For ensured specifications and the test conditions, see the Electrical Characteristics Tables.

In other words, it won't blow up with a supply voltage of 6V, but it's not designed to work at that voltage. -0.3V is within the abosolute maximum rating, but don't complain when it doesn't work at that voltage.

Thanks, I'd missed this and this is really useful information.

[David Hess]

An NPN and PNP emitter follower can be added after the voltage dividers to reduce their impedance by at least 2 orders of magnitude without adding undo complexity.

[new299]

Re OP: Resistor voltage dividers are not normally used (or suitable really) to "supply power" to anything. Too many disadvantages. In order for the voltage to be stable under load you need ~ 10x the load current flowing down your divider. Very inefficient, and still not stable. Active voltage regulator is a much better choice.

What is done sometimes is to "buffer the output of a voltage divider" with a non-inverting op-amp unity gain voltage follower. Even then, that buffered voltage is not often used to "supply power" to other devices, but commonly used to provide, for example, a "half-rail" reference voltage, which might be used in multiple places. For more accuracy people often use voltage references in the same way, ie buffer them with a voltage follower, then distribute to where they are needed.

TLDR: to "supply power" to something, use a regulator, most commonly for an opamp (particularly a precision one) use a low noise LDO.

And yeah, don't design to operate at "Absolute max".

Thanks!

Yes, I've seen a op-amp used to drive a supply rail in an older Lecroy oscilloscope, it makes sense to me that it could provide a reasonably stable supply (though the approach seems rare, and I'm not clear why you'd want to do it).

But I've never seen a voltage divider used to create rails before. I'm kind of confused as to why someone would want to do this (having been presented with the design).

Are there any examples voltage dividers on rails in production designs? I don't think I've ever seen it.

[oschonrock]

Are there any examples voltage dividers on rails in production designs? I don't think I've ever seen it.

Only in bad designs or if you are talking about a few uA and don't care to waste 10x of your uA in the divider and are not fussed about the exact voltage.

It's just a bad idea, if you can afford an LDO (price, space, etc) , use that...

[David Hess]

Yes, I've seen a op-amp used to drive a supply rail in an older Lecroy oscilloscope, it makes sense to me that it could provide a reasonably stable supply (though the approach seems rare, and I'm not clear why you'd want to do it).

Operational amplifiers used to be relatively expensive so using one that way would be considered wasteful.  Their relatively high open loop output impedance also limits performance unless you add a shunt capacitor and most people do not know how to do this without causing the operational amplifier to oscillate.

But there are cases where they are exactly what is needed, like to power a circuit which has limited power supply rejection like a VCO or oscillator.

But I've never seen a voltage divider used to create rails before. I'm kind of confused as to why someone would want to do this (having been presented with the design).

Are there any examples voltage dividers on rails in production designs? I don't think I've ever seen it.

There are plenty of examples where a resistor divider provide low current bias supplies for things like cascode transistor output stages which require neither precision nor high current.

[Zero999]

We need to see the schematic, to make a full judgement. The output impedance of the potential dividers is 60Ohm and it seems plausible a low frequency op-amp, will work with that sort of power supply impedance, without any problems, as long as not much current is taken from the output.

To get a total power supply voltage of 5.5V, rather than 6V, change the lower resistor to 100R, on the negative supply. The reason why I'm suggesting to reduce the negative, rather than positive supply voltage, is because the works closer to the negative voltage, so reducing that, will have less impact.