applying power to positive rail before the negative one or viceversa

[ricko_uk]

Hi,
assuming a the following:
- simple dual supply op-amp circuit with only a single op-amp or single instrumentation amplifier
- no input signal as initial condition (until both positive and negative supplies are switched on)
- very little output load

is there any issue in applying power to one rail first (for example the positive one) then several seconds later apply power to the other rail (the negative one)?

Also the same question for instrumentation amplifiers and specifically the attached one (in case the answer depends on the datasheet)?
If the answer depends on the part used, what do I need to check on the datasheet?

Many thanks

[schmitt trigger]

My understanding (and please correct me if I am wrong) is that the most negative supply provides the necessary reverse bias to isolate the substrate from everything else.

If you don't have a negative bias, all sorts of parasitic junctions could form between the +supply and the I/O pins.

[ricko_uk]

thank you Schmitt Trigger,
do you mean that in dual supply op-amps I should always apply the negative supply first?
Thank you

[duak]

ricko, have a look at page 8 of the data sheet - i shows the absolute maximum ratings of the device.  In particular, the input voltages relative to the supply rails.  If a device has a supply sequence requirement it will be spelled out in this section or in the power supply section.  Look also at the input voltage section on page 22.

What you won't want is to allow the negative supply to ever go positive relative to circuit common (or the positive supply to go below the same).   Schmitt trigger points out that the substrate is the most negative voltage on the chip and allows the use of reversed biased diodes to isolate the devices on the chip.  Some parts required a bias voltage before, or at least no later than any of the other supplies.  Dynamic RAM chips and microprocessors with 2 or 3 supplies from the 70's needed this.  I don't know if there was damage if the negative supply came up late, but I'll bet it wouldn't work well if the supply was low.

This device probably has some input circuitry dielectrically isolated from the substrate so their voltages can go outside of the supplies, but this is the exception for analog parts.  Many have protection diodes that clamp the inputs to the supplies.  Older parts have parasitic diodes to the substrate but not usually to the positive supply.

In general, I'd say op-amps should have their supplies come up at the same time.  I could imagine if the negative supply came up first, it could apply a negative voltage to the following circuitry which, if digital, could respond in an unexpected way.

Hope this makes sense,

[David Hess]

Even without signal inputs, applying the positive supply before the negative supply, or the reverse, can exceed the input common mode range of the operational amplifiers causing phase inversion.  Or there can be other problems.

In the past when this could be an issue, power supplies might use tracking regulators instead of independent regulators.  Or circuits were added to disable or control outputs until the supply voltages were in operating range; I have bench powder supplies which have exactly this to prevent output glitches when turned on or off.

[ricko_uk]

Thank you all!

[ricko_uk]

If you have a prototype board with dual rail supply and have only single supply bench PSU, what's the best solution to have dual supplies which also turn on together?

[magic]

If there are equal capacitances between each rail and ground, simply applying power between the rails will set the ground in the middle due to capacitive divider action.
To keep it there permanently, use a resistive divider or active rail splitter, depending on how much current imbalance is expected between the rails and supply impedance requirements.

[CatalinaWOW]

AFAIK there is no universal answer for this.  You must check the data sheets.  If you don't have access to the sheets negative first is a good guess.

Many parts are designed to be insensitive to brief reversals of supply voltages, but many others will latch up and a sizable number will fry.  The only solution is to repeat - check the data sheets and conform with the requirements.

[exe]

An opamp doesn't know anything about power rails, it is actually supplied by a single voltage applied between Vss and Vcc. Split rails are only used to bring input levels or output levels within required range (e.g., common-mode input doesn't include ground, or output cannot swing all the way to the ground).

So, applying only one rail... what does this actually mean? Does it mean the other rail at ground potential? If so, I'd apply a positive rail first, keeping in mind that either input will not work, or output will get above ground.

If there is no path to ground (unlikely) then the power sequence doesn't matter, as it is equivalent of no power.