4 Wire Kelvin Correction with programmable limit

[Rerouter]

In order to correct my biggest nit pick about the usupply (no negative sense line / no correction) i set myself to make a 4 wire sense correction amplifier,

first i began with only a standard error amplifier and a differential amplifier amplifying the difference, but noticed there was one massive flaw to this design, should one of your sense lines become disconnected your output shoots off to maximum which generally is very, very bad for whatever you are powering,

so i added a second differential amplifier to measure the difference between what was already being measured and what was being put out, in other words how much correction was being applied, that  was then fed into an op amp acting as a comparator, which then drives a mosfet and clamps the original error amplifier into a simple buffer should it become greater than the programmed limit, and act in the linear region to stabilise

the 2.2V input is a voltage i am aiming for as my lowest voltage and most likely to cause me grief, mainly as most op amp common mode ranges up to vss-0.9 to vss-1V  or much worse (part of the reasoning to keep the signal levels low)

so i offer up my design for all the scrutiny it is worth, and am happy to take on any constructive criticism, mainly i would like to know if there are any major issues with this design i have overlooked, or if there are any very simple improvements you can think of,

in my own mind i hold doubt about my differential gain resistors, but would like a second opinion

and if you want to play with it, here it is albeit a little messy http://tinyurl.com/9mrjbls (falstads)

[Rerouter]

Still working on this project, so to try and spur some opinion here is my next module

another poster in one of the usupply threads mentioned downprogramming an output, effectively pulling more current to drop the output quickly, and i saw in this a way to both fix the issue of high impedence load being slow to respond and the issue of the output capacitor having to first dump all of its charge before reaching a stable constant current,

so what i have done here is combined the constant current buffer amplifier with a differential amplifier that measures the difference between the set voltage for the regulator and the output voltage, (output divided to the same signal level "Vdiv")

this means when the set voltage is lower than the divider voltage it will then amplify that difference and begin pulling much more current through the transistor, (design limit at 400mA)

and while the output is the same as the set voltage the differential part makes no difference to it, and thanks to the 390K pull up resistor slightly off-setting the non-inverting input will cause the configuration to consistantly sink 0.4-1mA over the full supply range,

and for safety i added the 2 mosfets to disable the downprogramming, as should an external capacitance be connected to the output, it would start sinking from it, and this could be un-ideal for some,

i would love to hear your opinions and critisims on these modules or even a suggestion for other features that are commonly only seen in high end supplies,

[ejeffrey]

In order to correct my biggest nit pick about the usupply (no negative sense line / no correction) i set myself to make a 4 wire sense correction amplifier,

first i began with only a standard error amplifier and a differential amplifier amplifying the difference, but noticed there was one massive flaw to this design, should one of your sense lines become disconnected your output shoots off to maximum which generally is very, very bad for whatever you are powering,

An easy way to solve this is to add a ~10k resistor between the output and the sense leads.  If the sense leads are disconnected it just reverts to 2-wire mode.  This does generate some theoretical error, but as long as the lead resistances are <<10 kohm the error will be negligible.

[Rerouter]

And here is a simulation for the entire feedback loop for any of you to play with, the values for the regulator are a guess, however the rest of the circuit is effectivly what it shall be, with only the voltage set op amp powered from the same rail, all of the others powered off 5V (current sense amplifier is an equivilent to what i will be using, and the actual device has no common mode input issues)

have fun http://tinyurl.com/cb2rq7k

and yes i have tweaked some values off of my previous modules, and have yet to simulate in spice, but it definatly gets me in the ball park

to ejeffrey, what i prefered over my implementation was you could set a fixed limit for how far it can correct, also it allows you to measure low value resistances while in current liming mode,
and when powered off 5V the kelvin sense will be able to survive up to 105V on either input,