Proper Design Procedure in Selecting Components for LM317 Voltage Regulator Circuit.

[David_]

I haven't actually tested this i real world circuit, I was about to order the MOSFET when I realized that some thing was wrong. So far it is all LTspice, is it possible that this is a LTspice fluke?

In any case here is the circuit:


What will follow here is first a plot of the input and output voltage and load current but the green trace is a collection of 76 steps, the upper resistor in the regulator feedback voltage divider is 1meg in the plot while the lower resistor starts at 25k and in 76 steps goes up to 400k.
The second plot is just the same only I have flipped drain and source of the FET:



I am in the process of designing a neat PCB to test this setup in reality while fixing a few things that was not good in the last prototype, and I intend to prove that your(Klaus) hypothetical regulator control circuit can indeed go down to 0V or just about.

Regards

 

[FvM]

PMOS is wrong, either as shown or flipped. Would work with a NMOSFET in it's place, but you get uncomfortable high dropout voltage with FETs in any case.

 

[KlausST]

Hi,

Darasheet say, the lower feedback resistor should be less than 250k.

Klaus

 

[David_]

PMOS is wrong, either as shown or flipped. Would work with a NMOSFET in it's place, but you get uncomfortable high dropout voltage with FETs in any case.

Well I guess that shows what I know...

Is that large voltage drop something that is a effect of how FETs work and can't be helped or does it have anything to do with the situation of a N-Channel MOSFET not having a higher Gate-Source voltage than its Drain-Source?

I don't know how big the difference really would be(Gate voltage same as drain VS higher gate voltage than drain) But do you think that the following circuit concept could be worth anything?:


The idea being to reduce the voltage at the N-Ch MOSFET drain... Or is it the source voltage that matter, my thoughts are somewhat fussy and I think I might be very wrong in the idea shown above.
Or does the diodes in my concept circuit not matter at all since the source never exceeds 50V?

I suppose that it is worth wile to include footprints for both a Nch MOSFET and a NPN to the PCB for testing, I don't know why I always prefer to use MOSFETs rather than BJTs.

I don't know why I have such a hard time grasping the hole transistor concept, I read enough but I find that I need to read things expressed in a certain way for me to understand and honestly I should do more simple but really boring lab demonstrations. I think at least that especially if you read the theory and don't quite get it then experiments is the next step(which always should be done anyway).

I should have been thorough enough to read the datasheet but I didn't read more than the info page on linear.com, the regulator I will use in the real circuit is a TI part TPS7A4001 and there are a spice model for that but only a encrypted PSPICE model and I don't use PSPICE. I downloaded a free student version of PSPICE but I did not like it at all and you can contact TI and ask for a not encrypted version of models but that require you ta have a NDA whatever that is between you/your company and Texas Instruments and I don't think I have any opportunity for that.

 

[ads-ee]

you can contact TI and ask for a not encrypted version of models but that require you ta have a NDA whatever that is between you/your company and Texas Instruments and I don't think I have any opportunity for that.

Non-Disclosure Agreement, usually involves lawyers, or if you are a consultant you read it (good luck understanding it) and sign it (hoping they don't have a stipulation that requires your left kidney ;-)).

Basically it is so they can take legal action against you if you divulge any information on anything in the model. If you keep your mouth shut about what is inside then you would be fine.

 

[D.A.(Tony)Stewart]

An output impedance Bode Plot determines the stability. Also a Pulse step load.

try this.

 

[David_]

I am reading about how to do this in LTspice, I know that is a .AC analysis and that Z should = I/V but it can't be as simple as plotting the output voltage / the load current can it?
Why not? Why would it? Its easy to produce many different kind of plots for different things but its another thing to know that any of the plots show anything that is relevant or real.