[SOLVED] Not understanding the reason(s) why the regulator fails with discrete reference and buffer

Not understanding reason(s) why regulator fails with discrete reference and buffer

Hi,

This circuit works with a simulators voltage source of 1V as a reference, but when I replace it with a discrete design it fails to regulate the voltage regulator.

I sort of see that it may have to do with source resistance - the VS2 1V voltage source has 0r resistance, and when changed to more than 25r the circuit also begins to fail to perform as intended.

Also, I guess that maybe at start-up the regulator has no reference to work from when connected to the discrete reference...which needs the regulator to work before it can set the comparator reference voltage...

It's not a very good circuit, but that's not the point. Each block works on its own, i.e. simulates okay/well for arbitrary waveforms (regulator), transient, temperature, and DC transfer characteristic - except the regulator across temperature: it can't handle -50 to 150ºC, that starts to drop around ~70ºC and is dead by 100ºC.

The ref + op amp buffer can operate down to 1k load and maintain a 1V output. The voltage regulator also works down to 1k before beginning to go below 5V out (it only needs to carry a ~1.8k load).

What options are there to make this circuit work?
How do I get round the chicken and egg problem of the discrete reference - if that is one problem - and its needing a start-up that turns off because it has to get the kick-start from VS3 (the 6 to 30V supply voltage)?

JPG is not a good image, gets blurred when expanded, not sure why, so also attached is a PDF of the schematic(s). First schematic is the circuit broken into two, with ref and buffer on a 5V supply - that works, i.e. VF4 is 5V out. Second schematic is the version that does not work, i.e. does not regulate to 5V out on VF2



View attachment vregref112017.pdf

Thanks for any help in understanding the main issue(s).

- - - Updated - - -

I need to add, so as to clarify something not explained above, that if you look at the PDF, the top schematic (where the ref and buffer are powered by a separate 5V supply) also works as hoped for - with the discrete reference as the comparator reference.

Re: Not understanding reason(s) why regulator fails with discrete reference and buffe

I'd start (heh) with where it says "Start-up (not used)".

Beyond that, look out for positive feedback when you
try to power a reference block from the regulator it is
the reference, for. This can be a startup issue or a
small signal stability issue (once started). My preference
has been for a crude "LDO" to get reference PSRR up,
and forget being so clever.

Re: Not understanding reason(s) why regulator fails with discrete reference and buffe

Hi dick_freebird,

Thanks for the pointers, much appreciated. The comparator is very, very bad (Grade A device? No - a solid grade Z), so I can dimly associate your description/advice with things I see happening. I just didn't want to copy the LM139 (Love it! Want to shake the hand of the designer(s) for that current source.) or similar.

Anyway, I couldn't see a way round the start-up but I've managed to "solve" a few issues today by rethinking the reference and its needing supply independence, so ignorantly throwing components at a basic idea and juggling with them a bit I got this - by luck, not skill or knowledge:



Shame it droops a little from 0ºC to -50ºC, because I always work on the theory that these things are never quite as good in the real world as the simulation...

If you have the time to reply - What pieces, for you, constitute a "crude LDO"? I'm not good at this stuff to begin with, but discrete regulators that actually regulate elude me still. Thanks.