Voltage Sources in Parallel

[mribble]

I'm designing a modular system where different devices can plug into each other.  Some of the slave devices need power from the master device so I had to provide power on the plug.  However, I also want to allow two of the powered master devices to plug together for communications.  While each device should have 3.3V on this line, I know the linear regulators can have slightly different voltages.   I'm wondering if those different voltages would cause issues?

I know putting a diode or resistor would make it safe to connect these devices, but I'd rather not have the voltage drops from these.  Is there some standard solution or chip to solve problem?  On a side note I looked at the Arduino circuit and they have some 6 pin SPI headers that have power.  If you connected two Arduinos together with their SPI header they would have two regulated power sources meeting in parallel without any protection.  Maybe in the real world this isn't an issue because of internal resistance of the regulators or maybe Arduino isn't supposed to be connect this way?

Here's the regulator I'm using in case it matters: http://www.mouser.com/ProductDetail/Texas-Instruments/LP38692MP-33-NOPB/?qs=sGAEpiMZZMvu8NZDyZ4K0c0kBHVNK554

[IanB]

The operation of a voltage regulator, in simplified terms, is like this:

1. If V < Vref, send more current to the output
2. If V > Vref, send less current to the output

So if you put two regulators in parallel with slightly different Vref values, the one with the highest Vref will win. Assuming it has enough capacity, it will send current to the output until the voltage rises to its Vref value. The other regulator will see a voltage that is too high and so it will reduce its output current to zero.

If you increase the load too much, beyond the capability of one regulator, then the voltage will sag until the second regulator cuts in.

You will not get equal load sharing between the regulators unless you put some resistors on the regulator outputs.

If there are no resistors and both regulators become active, there is a possibility of oscillation as each regulator will see an active load with unexpected feedback from its control actions. In practice I don't know how bad this could be.

[mribble]

That sounds not nearly as bad as I thought it might be.  I don't care at all about having an equal load balance.  Either regulator can provide more than enough current. 

I was thinking that that the lower voltage regulator would try to keep the voltage low and you'd get something like a short circuit, but it sounds like you don't think that would be an issue.

If anyone has additional insight into the possible oscillation issue IanB mentioned I'd be very interested in that.

Thanks!

[Richard Crowley]

The concept of trying to "share" or "bus" the final, regulated power rail is a method rarely used because of the very problems you are encountering.  OTOH, "busing" the raw (pre-regulation) power is much easier on several levels.

[mribble]

That's my current plan.  I was just thinking it would be nice if we didn't need each slave module to have it's own 3.3V regulator, but that isn't so bad...