L7812 Voltage sag with 12V DC motor

I'm experiencing a voltage drop across a pretty simple circuit and don't understand why.

I have a L7812 setup to give my circuit a 12V DC power. The L7812 has bridge rectifier, caps, and transformer 120V to 15V. From the LM7812 I pass the current through a pnp transistor that is fully saturated. Looks like I get about 18-19V into the L7812 and when the motors (small pc fan motors) are off (pnp open) I see a stable ~12V.

I have a small resistor in series (out of motors to ground) with the motors (3 in parallel) to limit the current. The motors are expected to draw 100mA. So I expect to drop a bit over 300mA across the circuit. My transformer and L7812 are all spec for up to 1A. When I turn the motors on (pnp saturated) the voltage sags to near 9V pulling about 300mA. So I'm not too sure why I'm seeing the sag as I should be able to drive plenty of current and with my resistor in series of the current path there should be current limited.

Any ideas? Thank you.

 

To BobK: I'm measuring from the LM7812 output to ground. So in a sense it is across the pnp, motor and resistor. As for the resistor need, I wasn't too sure how the motor would act and have thought it to be unnecessary too, but I'd rather be safe than sorry.

 

I'll get a schematic out soon. But the values are measured. The current measurement is from the LM7812 output to the rest of the circuit.

 

Thanks for all the help so far.

Here is a sample diagram more of a hack job of the real circuit. Many of the switches don't exist and are here for me to test the sim. My real circuit doesn't have SW1, but uses some other things to mimic a switch, but I get the same effect where the voltage at QE2 drops from ~12V to closer to ~9V when SW1 goes to ground (lower position).




Steve: it is a TO-220 package. I have a heat sink on it and it stays pretty cool

 

1) thanks for the suggestion. I don't have an o-scope, but have increased the cap size. I imagine my voltage has less ripple now, but I still the the voltage sag.

2) The short answer is the SW1 really isn't in the circuit and I have it there to ensure that some of the other components in the real circuit have a ground path (though quite limited).

 

I'm starting to think it is the L7812 that is creating the problem. I bypassed the pnp transistor and see the same sag. The exact 7812 that I am using is L7812ABV
from digitkey: http://www.digikey.com/scripts/DkSearch/dksus.dll?vendor=0&keywords=L7812ABV

Also, I did a few measurements around the 7812. The Vin is about 22V +/- 0.6V, where the 22V is a DC measurement and the 0.6V is an AC measurement. Vout when the motors are off is 11.75 +/- 0.005V. When the motors are on I see it drop to 10V +/- 2.5V.

Any new ideas are greatly appreciated.

 

A quick measurement shows just about 0.1V drop across the emitter and collector. You can see with the diagram now posted that I am using the transistor as a switch.

 

I'm off to buy some larger caps now. The 100uF seem to have helped some.

 

Just added the 1000uF on the input and the voltage sag is gone.

Thank you (and everyone) for all the ideas and help.