Microcontroller needs owndedicated 3v3 regulator...

[Flyback]

Hello,

I have a microcontroller which controls an emergency light. This micro must charge the battery, which powers the light.
The micro also receives inputs from various opamps and comparators.
The micro and most of the opamps and comparators are all supplied by the same 3V3 linear voltage regulator. (this is the “main” 3V3 regulator)

The 3V3 regulator is supplied from the battery.
However, every now and again, the micro must simply switch the battery out, so that the battery does not get drained by the ~8mA current drawn by the opamps etc.

The problem is, that when the micro switches out the battery, it also switches off its own power supply, and this isn’t wanted.

Therefore, the micro must be fed by its own, separate, dedicated, “micro” linear 3V3 regulator. Then when the micro switches the battery out, this only means that the “main” 3V3 regulator looses power. The “micro” 3V3 regulator still draws power from the battery (but not much).

Anyway, do you see any problems with having two 3V3 regulators like this.? (ie one supplying the micro, and the other supplying the opamps and comparators etc).

One problem with this “two regulator solution” could be that some of the micro’s digital outputs connect up with circuitry supplied by the “Main” 3V3 regulator, and when the “main” 3V3 regulator powers down, the micro’s digital outputs could end up getting shorted to ground via the ESD diodes in the opamps etc.

Are there any known ways of preventing this kind of thing from happening?…or is it just a case of connecting series resistors to the micro’s digital outputs so that high short circuit currents don’t get drawn from the pins of the micro? What sort of size series resistors would be advisable?….(or more relevantly,what is the maximum current that could safely be allowed to flow through the ESD diodes of the opamps and comparators?)

 

[JimB]

Alternatively, when the micro turns off the power to the "Main" regulator, why not arrange that it also turns off the other outputs which may be "shorted" by the protection diodes?

Then when it is time to re-start the main supply, turn on the other outouts as well.

JimB

 

[KeepItSimpleStupid]

Flyback:

Your explanation got be a bit confused.

It isn't so much powering th OP amps and processor separately. but interfacing the digital signals created by the "OP amps" to the micro.

One way is to interface the "OP amps" with an I2C level translator which is smart enough to go into a high Z state when either poswer supply is missing.

Otherwise, depending on your design you could be passing way to much current through the protection diodes.
e.g. If the processor was unpowered and the op-amps were and you pass a 3.3 signal essentially to ground becaise the supply is at zero.

Now if the "OP amps/comparitors" were all open colector, then you would have no issues.

Here https://www.fairchildsemi.com/pf/FX/FXLA0104.html is just one example of interfacing the actual logic levels. I have been researching lately and there are LOTS of components in this area.

 

[ronv]

Presumably it is the micro that puts things to sleep. Why not program it to set the outputs low at the same time? You could probably use a little PFET to switch off the logic and save a regulator.

Oops, missed JimB's post.

 

[blueroomelectronics]

Would help if you posted a schematic.

 

[Flyback]

Some of the inputs to the micro are from opamps to the micro's ADC inputs........the ADC obviously draws some current from these inputs (to charge up the ADC capacitor), so I would rather not have resistances in series with the micro's inputs and outputs, -but believe I can get away with it.
I appreciate all the above, and series resistors still seems the cheapest and lowest component count way...............I am just not used to putting series resistors in micro I/O lines.

 

[KeepItSimpleStupid]

Again, move to I2C: **broken link removed**

Or shut the ports down. Tri-state them.

 

[samovar]

I would have one regulator. Then, when needed, I would use a MOSFET to switch the power off to everything except the controller.

 

[ronv]

Some of the inputs to the micro are from opamps to the micro's ADC inputs........the ADC obviously draws some current from these inputs (to charge up the ADC capacitor)

I don't think so.

so I would rather not have resistances in series with the micro's inputs and outputs, -but believe I can get away with it.
I appreciate all the above, and series resistors still seems the cheapest and lowest component count way...............I am just not used to putting series resistors in micro I/O lines.

What makes you think you need them? I mean if your mind is made up that's one thing..........

 

[Nigel Goodwin]

I get the impression he doesn't know what he's on about, at least he certainly can't explain it - I don't even have a clue what he's asking?, but it makes no sense.

 

[kubeek]

If you keep power for the micro and turn off everything else you will be fine, the other way around would make problems with esd diodes. Just make sure you keep the outputs going to the opamps low when their power is off.

 

[Flyback]

Just make sure you keep the outputs going to the opamps low when their power is off.

the problem is rogue pin states during programming......during programming, micro pins are supposed to be high z, but usually are not.....they could be output high, say

 

[kubeek]

What does this have to do with anything I said?

 

[Flyback]

I should have explained better, youre absolutely right in what you say...its just that I cannot always assure that those outputs that you speak of can be low, when the opamps power is off.
I was not disagreeing with you, I was agreeing with you, but stating more of the problem, in response to your excellent point.

I am just chucking this around my head, cuzz my boss does not want to make this change, he wants to keep our max1614 switch out ic.............I can see why to a point, as our micro is 64 pin, and ill have to rip up the whole layout and shovel in around 30 to 40 series resistors in series with with micro pins...its a big change.......but I have to convince the boss to allow it....he seems happy with the max1614...but that's a problem because it means we need a 5 cell battery solution to give the 5v max1614 needs.

I reckon we need to loose the max1614, and add the series resistors, but he wont do it.......I am trying to fish for help toward a way to convince him

 

[kubeek]

Well and is there a problem with powering the whole circuit while you program it?

 

[blueroomelectronics]

Guess we'll never see a schematic. Or any other detailed info. I wonder if this is the same wonky design that uses a SAM7 CPU to control an emergency light.

 

[KeepItSimpleStupid]

Why not just one or two I2C chips that do level translation and D/A. Here, read this datasheet: https://www.digikey.com/product-detail/en/PCAL9554CPW,118/568-9927-2-ND/3677033