LT3080 wierdness - dave's power supply (eev224) gone mad

[mnmlistmktg]

well, i've never met a lt3083 but that sounds truly weird, and a different problem from the Vset/Vout problem that this thread started with on the 3080. i guess you've done the obvious checking your made it right etc. you'd need pretty skinny or resistive wires to account for that voltage drop. assuming you followed dave's original circuit (or variant) i'd be wondering:
1. have you got sufficient load on the 3083 when its at rest? the lm334 bit. tho that wouldn't really account for a steady drop of Vout by current.

Yes, LM334 in place and functioning as expected... also, these issues only happen under load - same with either a power resistor or the EEV dummy load

2. does your power supply offer sufficient input voltage at higher current? is that rolling off? if you had a scope you could check its not loosing half the wave due to faulty or small input smoothing caps. (are they big enough?)

I believe so... I've been using a power adapter (12V 2A) for the time being, not going to set up the transformer / rectifier until I have the rest of the circuit worked out.  I have tried other adapters and even an ATX PSU, all with same results.

3. is the voltage at V-set (from the op amp) stable or does that too drop with current. that would be odd and suggest a supply problem.

The more I've investigated, the more odd things I've found.... can't recall if v-set has been stable, but am documenting everything and will post full report when done.

4.again with a scope are you sure you haven't got any oscillation with the current limit - disconnect that pull down transistor til you get this sorted.

I've found some sporadic oscillations, but I can never duplicate the conditions.  I'm going to start breaking the circuit down and do tests section by section as I build it back up.

5. if V-in is stable and the V-set/V-out increases (it shouldn't) with current then perhaps you've discovered an entirely new form of LT weirdness... who knows whats possible with these things? (LT certainly isn't going to tell us)

No, adjusting the current limit doesn't impact anything but the current limit, so at least that is working!

Thanks for the input, I'm starting to think I'll have to figure something out with the 317, but I want to try and nail down what's going on with the 3083 first.

[mcbmcb0]

for now i'd still remove the pull down current limit transistor. that's potential trouble you don't need.

re your power pack... has it got a smoothing capacitor? most wall warts don't. could you stick a 4700 or 10,000uF cap across it and test again?
still may not explain the weirdness at lower Vout, but worth a try

[mj21]

i had the same issues with lt3080 and even lm 317 seemed bit temperamental, so i decided to built my own regulator. after searching for few minutes and reading many posts i found this. i have tested this design and i must say it is very stable. i have already made 10 prototype with different op amps and they are all stable. give it a try. i am using dual rail supply. have not tested it with single supply as all my op-amps are dual rail supply ones.

all you have to do is remove the lm329dz and connect the adjust output of the daves circuit. it goes all the way to zero volt as well.

[mnmlistmktg]

for now i'd still remove the pull down current limit transistor. that's potential trouble you don't need.

Pulled out the transistor... didn't change anything, but I think I'll leave it out anyway until I figure this out.

re your power pack... has it got a smoothing capacitor? most wall warts don't. could you stick a 4700 or 10,000uF cap across it and test again?
still may not explain the weirdness at lower Vout, but worth a try

Added a 2200u on the input and tested in/Vctrl as well as Vout over loads... both were dropping as load increased, which leaves me to believe that load regulation isn't the issue, but rather line regulation.  I was a bit surprised that the 2200u input cap didn't seem to make a big difference, I'll need to order a bigger cap to see if that helps.  What I really don't understand is that the line regulation dropped even when I started loading at 5v out.  It's a 12v supply, so there should be plenty "in reserve"

[GiskardReventlov]

The conclusion here, for anyone not wanting to read the entire thread, is that LT never gave a final  response. From the lack of a response, (which really is a response) we can conclude that LT3080 is a problem.

Is it a documentation problem or a hardware/design problem? 
From reading this thread, I think it's a hardware/design problem and LT is never going to publicly admit that it's a problem.  But they have admitted so by providing no guidance.

Am I wrong? If so how so?

[AscendedDaniel]

I am also designing a power supply, and I found this thread. I decided to try out some of the test circuits on a breadboard, but I have been unable to replicate the failures mentioned here. Attached is my test circuit. (I least I hope it's attached. It's my first time on the forums.)

I set the output voltage to around 5 V with the pot. Then, I tried grounding test point T1 with a jumper wire to simulate entering constant current mode. That should put Vset below Vout. I also tried grounding Vout, to get Vout below Vset. After multiple attempts Vout-Vset was <3mV, within spec for the TO-220 package, according to the datasheet.

Am I not using the same circuit as in the posts before? The other possibilities I can think of are that LT silently fixed the issue in the LT3080 or there was a bad lot that eluded testing at the factory and I got a newer lot. The only codes I see on the chip are 1153860 and e3, but I'm not sure how to convert that into a date. 53rd week of 2011? I have a few LT3080s in TO-220 to mess around with. Let me know if you have any other ideas to try.

[David Hess]

If the LT3080 does have back to back diodes internally to OUT, then the really important spec becomes the +/- 10mA limit on the SET pin current. If you have, say, 10V on the SET pin and the output gets shorted, your circuit has to ensure that the maximum current into the SET pin is less then 10mA. Same story if the load actively drags the OUT pin up to 20V. Failure to ensure this in the design could lead to chip latchup.

Latchup also depends also depends both on the semiconductor process used and some aspects of the layout.  One of the big advantages of dielectrically isolated processes as oppose to junction isolated processes is their immunity to latchup.

Here is the circuit that LT meant to put into the LT3080 datasheet:



(Note to LT - you have my permission to this image directly into an updated datasheet.)

Their simplified schematic was a little too simplified.  :/

Texas Instruments tends now to not include a schematic at all and has been removing schematics as they rewrite data sheets.

[David Hess]

i've been playing similarly with the LM317 - had a hard time getting it stable when driven with an op amp (with protection diodes etc), current limiting and high current transistor. I'll be keen to see what you come up with

I have never had a problem driving an LM317 (or 7805) with an operational amplifier.  Depending on some details like the operational amplifier bandwidth, you may need to add frequency compensation and this will become more critical if an external PNP pass transistor is added in the typical configuration.

One of my favorite circuits is similar to this with a LT1007 or equivalent, precision shunt regulator, a lead network and frequency compensation around the amplifier, and a 2N3906 replacing the JFET if a 7805 or similar is used:



The last time I did a few of these, the line and load regulation was practically unmeasurable.