Dueling Voltage Regulators

Status
Not open for further replies.
Constantin

Constantin

Well-known member
From what I've read, it's not OK for a voltage regulator to feed a voltage bus along with a fellow regulator, unless they're designed for parallel operation and/or you use Schottky diodes to isolate them from another.

However, would it be OK for them to be both feeding the same 3.3V bus if one of them is about to drop out due to a rapidly decreasing power supply? I ask, since I would like to use the enable feature on a voltage regulator with a dual comparator to detect when 1) the main supply bus voltage has dropped 30% and 2) the battery has enough juice left. I am using a TS9002 comparator. Here is the circuit I propose using for the battery-powered supply.

Clipboard01.jpg

The VBUS power supply is at a nominal 5V. The drop-out voltage on these TLV70233 regulators should be on the order of less than 100mV in my application. VBATT is hooked up to a LiOn battery and that voltage should range from 4+V to about 3.6V when the thing is empty. I used a NAND gate to combine the two outputs and drive the input into the TLV70233 Enable Pin HIGH whenever the two conditions above are met and the two outputs out of the dual comparator are LOW.

The idea is that the TS9002 will sense the rapidly decreasing VBUS voltage (let's say the power plug is pulled on the wall wart) and as the bus voltage plummets past 3.5V, the battery power supply is enabled. A pull-down resistor will pull the Enable Pin on the TLV70233 down as soon as the NAND gate no longer supplies a HIGH signal.

For the short time that the two voltage regulators are potentially at odds with each other, would you expect any damage? I'd think the time period would be too short and the disconnects too infrequent for this to be an issue, but is this correct?
 
Running two power sources at the same time is always potentially exciting but I think you can do what you are trying. I take it the other regulator is the one onboard the Teensy?

Looking at the TLV data sheet it has a protective diode on the main FET (page 4). If VOUT is higher than VIN then you are going to have current through that protective diode. I think there is a high series resistance there but still potentially interesting. Now in theory your external 3.3 Vout will always be lower than the 3.6V minimum of your battery pack so the diode will remain reverse biased. Does mean there is a potential battery charge situation here pulling from your primary reg if the battery becomes very flat.

Getting the TLV to take over from a falling main regulator should be fine, though again you may end up with the 3.3V showing up on it's VIN if it has fallen all the way. Again not normally an issue.

So as long as both your VINs are either disconnected or higher than 3.3V this should work.

When looking at this sort of thing you always need to look at both data sheets, do a truth table of the potential input combinations. Check for things like power indicator diodes that will be light up by reverse current through protection components and confuse you mightily. Have had a battery backed device where a Mains OK led would dimly light under battery operation.

And all of the above only applies to linear regulators. Running power into a switchmode (AKA wallwart) is an unknown and apart from anything else can end up with interesting resonant effects as the two regulating circuits swap load back and forth.
 
Out of curiosity did a quick check and powering via the T3 3.3V in pin you get 2.9V on the Vin pin. This is on the far side of pair of diodes so there is a hefty series resistance (>100K) in there but the capacitors store enough charge to flash a LED before they discharge. Not seeing any particular case where the current flow would have any meaningful effect at <1 uA. Which is why the diodes are there of course.
 
Hi GremlinWrangler and thanks for your insights! I too just realized that my approach won't work. Looking through the TLV data sheet, they mention that the VOUT is actively pulled low whenever Enable is pulled low. That's going to be a problem! :p

One of the reasons that I wanted a battery power supply was to keep the MCP3910 front end powered and logging while momentary line power interruptions could take the wall-wart down. Since the MCP3910 has its own two power supplies (one for analog, the other for digital), it's VBUS that must be kept alive (this is what they use as a source), not the 3.3V bus for the MCU and the rest of the stuff on the board.

Back to the drawing board!
 
@Constantin,
Is there some kind of: fail over, hot plug, or dual supply priority chip out there? I would think with mobile devices being common with chargers and the like there would some sort of supply priority chip out there.

Also, maybe I am missing it, but why can't you use a diode? Is the supply too low?
 
a 3.4V battery voltage doesn't leave a lot of room for a diode drop in conjunction with the 100mV internal drop in the reg. I think the commercial solution is more normally a single boost converter with two Vins, but it's not a subject I've spent much time on. I've so far just engineered things to live with the diode drop, accepting the power losses in battery applications for bullet proof design.

In this case the simplest solution may be to run the Teensy below 3.3V, see
https://forum.pjrc.com/threads/10736-Can-Teensy3-be-powered-from-3-3V-if-USB-is-not-used?

via a diode. This will simplify the power switch logic since the reg can be enabled but not under load while standard power is present since the diodes will be reversed against the normal 3.3V reg output with disable only happening when the battery is too low. It might even be possible to place this under MCU control and skip the hardware comparators. The easiest design for this does have the flaw of course that while it will run on battery power, if you switch it off you can't start it again...

Running on a lower Vin will impact the output voltages on the pins and the scaling of analog read.
 
Status
Not open for further replies.
Back
Top