Facing a problem with Dave's power supply design

[Asim]

Hi guys, so i bread boarded the circuit Dave designed,I have a problem with the first op amp (1a in the schematic). The voltage across the 1 ohms sense resistor is 1.5mV which corresponds to 1.5mA ( expected because of the lm334+ some losses). the problem is with the output of that op amp, it is giving me 70mv while it should give me 1.5mv( is it a noise problem?), the output starts to change when the voltage across the sense resistor increases beyond 100mv( i don't have an exact value) I double checked the 10K ohms resistors and they are fine. while trouble shooting i disconnected the sense resistor and feeded 5V I was getting 5V at the output of the op amp as expected.
is this normal? if not where do you think the problem is ?

THANKS.


 

[facumedica]

Well, what you got there is called "the Rail-to-Rail limits"  I've faced the same problems, if you put -2v on the VSS pin of the Op-Amp you will see that it will reach 0v without any trouble. I think this is your problem, If I'm wrong, excuse me, english is not my native language   

[Asim]

Well, what you got there is called "the Rail-to-Rail limits"  I've faced the same problems, if you put -2v on the VSS pin of the Op-Amp you will see that it will reach 0v without any trouble. I think this is your problem, If I'm wrong, excuse me, english is not my native language   

OUCH, I totally forgot that  , yea that must be it. i will test it later
 

[edavid]

Well, what you got there is called "the Rail-to-Rail limits"  I've faced the same problems, if you put -2v on the VSS pin of the Op-Amp you will see that it will reach 0v without any trouble. I think this is your problem, If I'm wrong, excuse me, english is not my native language   

No, that's not it, the TLC2272 is used specifically because it has rail-to-rail output.

[facumedica]

Did you use that Op-Amp??? Or did you replace it?

[Asim]

I did use it

[Asim]

i was trouble shooting it again, the current limiting works fine when the voltage across the sense resistor is greater than 300mV. I tested it with shorting the output of the power supply to the ground and by increasing the current limit pot from minimum(0V to 500mV). something is wrong with the differential amplifier when the input voltage is low !

I am using TLC2272CP

[mrkev]

I believe that facumedica is right... Even tho it is rail-to-rail, you can't expect it to go to 0V at the output. Keep in mind that inside, there are two mos-fets at the output. Even tho they are fully open, they will have some resistance and D-S voltage...
Try to implement low negative voltage, one diode drop should be enoug...

[Asim]

I believe that facumedica is right... Even tho it is rail-to-rail, you can't expect it to go to 0V at the output. Keep in mind that inside, there are two mos-fets at the output. Even tho they are fully open, they will have some resistance and D-S voltage...
Try to implement low negative voltage, one diode drop should be enoug...

I have no other choice but test it, the thing is i don't have a negative power supply. so i will build a simple one with lm337 / 7905 somewhere , I alse have a center tapped transformer already.

[Kevin.D]

Yes facumedica was right ,it's the output limits of this opamp .There's no true rail to rail opamps . looking at the data sheet for this particular opamp shows you minumum OL (output low with 500uA load)   typicaly 90mV -150 mV ,so it's  even poorer than a cheap lm324 (OL min ~ 20mV) . You got a few options  ,you could supply a small neg rail for the opamp  ,or add a small offset to the diff output .

[mrkev]

I believe that facumedica is right... Even tho it is rail-to-rail, you can't expect it to go to 0V at the output. Keep in mind that inside, there are two mos-fets at the output. Even tho they are fully open, they will have some resistance and D-S voltage...
Try to implement low negative voltage, one diode drop should be enoug...

I have no other choice but test it, the thing is i don't have a negative power supply. so i will build a simple one with lm337 / 7905 somewhere , I alse have a center tapped transformer already.

Actually, you don't really need that if you put one diode at ground and connect all opa directly to that gnd, rest of schematic directly after that diode, it should be enough. Even shottky should do.

[Asim]

That's a neat trick mrkev ,I will try to remember it next time 
 I tested it and it is behaving as expected.
 
that's strange anyways, Dave didn't use a negative supply or a diode to lower the reference of the op amps grounds even though i am using the same op amp ?!

[mrkev]

That's a neat trick mrkev ,I will try to remember it next time 
 I tested it and it is behaving as expected.
 
that's strange anyways, Dave didn't use a negative supply or a diode to lower the reference of the op amps grounds even though i am using the same op amp ?!

When you design things, it never works as expected at first go Anyway, i don't think he missed that, he was probably sattisfied with current limitation starting at about 100mA...

[Bored@Work]

that's strange anyways, Dave didn't use a negative supply or a diode to lower the reference of the op amps grounds even though i am using the same op amp ?!

You are aware that Dave never finished the design because it has at least one fatal flaw? And the diode trick? Well, it creates all sorts of funny problems, e.g. not having a common ground any more.

[Asim]

Ummmmm, no I am not aware. What fatal flows ?!!

[mrkev]

that's strange anyways, Dave didn't use a negative supply or a diode to lower the reference of the op amps grounds even though i am using the same op amp ?!

You are aware that Dave never finished the design because it has at least one fatal flaw?

No, actually he builded exactly this design with few changes (so it could be controled from uC)... I don't know what you're talking about... http://youtu.be/Lg6oYFerUlA

And the diode trick? Well, it creates all sorts of funny problems, e.g. not having a common ground any more.

Nah, not really; even for bigger current than 1A... I can't think of any funny problems you may see there, nor how it changed "common ground".
Simple negative source is better, sure, but that diode is't usually a problem, more so if Asim just wanted to find the problem on bench supply while considering negative supply later...

[Bored@Work]

I am talking about the LT3080 having a tendency to blow up. Just because Dave once build something doesn't make it correct or perfect.

And you can argue all the way you want, you didn't think the rubbish with the diode through.

[Rerouter]

Bored is right there, you essentially need to diode clamp the set pin to the output pin, exceeding one from the other by (i believe) 1V can cause it to die internally, or just act funny,

This was only later found out after harassing linear for some weeks, as they where unwilling to admit to it, they did make a newer alternative with a 10V differential which makes things more manageable,

[Asim]

after reading Rerouter comment, I searched the forum until i found this https://www.eevblog.com/forum/projects/lt3080-wierdness-dave's-power-supply-(eev224)-gone-mad/


LM317 here I come , though i will need to generate a negative rail from the battery to get to 0V output.

any idea of an easy way to do it ?

[Rerouter]

generic 3.3/5V isolated switcher, you hook the positive side of the output to ground and away you go with -5V,

[mrkev]

I am talking about the LT3080 having a tendency to blow up. Just because Dave once build something doesn't make it correct or perfect.

Ok, I didn't know that. Btw. you were the one who told us, that Dave never build it. Next time, just say what the problem is right away and safe yourself (and everybody else) some time, more so when it's not an obvious design flaw...

And you can argue all the way you want, you didn't think the rubbish with the diode through.

As I wrote, I wouldn't recomend it in final version. However it could be usefull in some cases.
Anyway, I can't really "argue", since you didn't use any arguments ("it's rubbish" doesn't count even if you write it twice...).

[rdl]

LM317 here I come , though i will need to generate a negative rail from the battery to get to 0V output.

It costs a couple of bucks, but I once used an LM385-1.2 and a 7660. I think used a voltage regulator too, I'd have to dig up the schematic to be sure. It might not be a great solution for battery power though. A CMOS 555 inverter may be better. The LM385 is pretty low power though.

[gxti]

LM7705 is a nice, inexpensive switched-capacitor regulator that puts out a stable -0.23V, just enough to make any "rail to rail" op-amp actually go rail-to-rail. Will work for some non-RR opamps too but not the older designs like TL072 with overhead big enough to kick a football through.

[GiskardReventlov]

I am talking about the LT3080 having a tendency to blow up. Just because Dave once build something doesn't make it correct or perfect.

In my search here I found it was the entire LT308x series that was flawed. But what I don't understand is why the issue seemed to die a silent death. Is LT held in some venerable position?

And yes a big clue was the fact that Dave abandonded the project. From my naive point of view it was a good learning experience.  i.e. Don't put all your eggs into a single-chip-basket and stick with what you know works.

[Asim]

guys, I have an idea feel free to correct me if i am wrong. I want the power supply to be portable that's why I want it to be battery powered. I will be using portable(3 or 4)  5V phone chargers( it has a lithium ion with a charging  and step up circuit ).

here is the idea: connect two batteries in series to get 10V and the 3rd one connect it backward so I will have 10V_ground_ -5V . i will be using the -5V to power the op amps. I will generate a -1.25V from the +2.5V reference with a inverting amplifier  and use that as a reference voltage to the set pin, when the current limit kicks in i will short the set pin with a transistor to -1.25V instead of the ground.

 any problems with this ?