And post 3 isn't like any data sheet I've seen out of 30 I looked at. I did find it in appnote.
But in the data sheet it's using a diode with resistor to -10 volts.

Some datasheets are better than others. I've noticed a trend to no longer include information that was previously included, specifically schematics.

I first saw the circuit in question in the 1976 National Semiconductor Linear Data Book. The picture I posted is from the 2004 National Semiconductor datasheet, page 15.

 

The zener or two diodes not going to be as good as it could be done with a regulater that can go to ground here little easier way from eeblog

What a horrible idea. The voltage drop of the diodes depends on load current and is nonlinear. Kind of defeats the purpose of using a voltage regulator.

 

It's horrible lol its all in what you need.
Horrible is all these people that can't look out side of the box. We still be texting in stone.
This is how bad it is maybe off by 200 mV
Heck most USB supply's are off by up to half a volt its all in what one needs

 

It's horrible lol its all in what you need.
Horrible is all these people that can't look out side of the box. We still be texting in stone.
This is how bad it is maybe off by 200 mV
Heck most USB supply's are off by up to half a volt its all in what one needs

The USB power specification is +/- 5%, and most devices have a bit of margin around that. Comparing that to other environments without qualification is reckless and dangerous. And irrelevant, since the topic of this thread is output voltages below 1.25 V. FPGA's and GaAS devices routinely have voltage specs below 2 V. For a 0.8 V CPU core, 200 mV is a 25% error that can cost hundreds of dollars per device.

Of the various 2-diode approaches to compensating for the LM317 minimum output voltage, the approach in the video is the worst performing. Not only are the two diodes outside the regulation control loop, and not only do their Vf change with temperature, but they change even more with output current, significantly degrading the regulator circuit's overall performance. And, the current-induced error is worst at the low output voltages that the circuit is specifically created to provide.

Fairchild's datasheet for the 1N400x series indicates Vf = 0.6 V @ 10 mA, and 0.9 V at 1 A, a 0.3 V change. Two of them in series yields is a 0.6 V change that is not compensated for by the regulator. Even if we discount that error by 50% because the regulator application is steady-state while the datasheet data is for pulsed operation, and the steady-state temperature rise Vf decrease will offset somewhat the current-induced Vf increase, at a nominal 1.0 V output voltage that still is a 30% error, 30 times worse than the LM317's max load regulation error.

ak

 

FPGA'S don't use a lm317 do they ??
Oh and the op changed his first post theres now more to work with he added center tap transformer. Google get the job done for him now I'm out of this one. Got to figure how to use a LM 317 for fpga.