Constant current constant voltage linear supply problems/questions

[Sniper1]

So i am trying to include a small constant current constant voltage supply in one of my projects ( active load) and after searching the web i found Dave s design that uses just 1 power element (very important for me since i only have 1 heatsink slot that is not used, the heatsink is recovered from something old and it 2.5x1.5x10cm long i calculated online that it can dissipate 15W with no fan and that would be plenty for me atm) ,
to be more precise i am trying to make something similar to this:   https://eevblog.com/uploads/uSupplyBenchRevA.pdf   

But fists i thought it would be a good idea to simulate a bit in LTSpice and apparently what i have works well with ideal op amps but real ones dont work well ( i am also attempting to replace the NPN with a Pmos with no success.
On paper they should work unless i forgot to account for something ......
As for the specs I am aiming for they would e loosely this:

• 0V out ( i plan to use 2 Si diodes and have the output come from a floating GND that is 1-1.5V higher then true GND
• 0-12V output range but i can settle with less if it makes sense ( i think here i am sort of limited by the transistor and op amp input voltages + max dissipated power by the heat sink)
• low noise ( that i why i want ti to be linear)
• As for output current i want like 1A for 5-6V and under
• Stable , aka no oscillation and the CC limit to work well

NOTE: power supply kill be a 12-24V AC-DC adapter so i get isolated DC and sort of regulated input, or some battery)
I think/hope that my requirements are reasonable ( yea at super low voltages it would be harder to get large currents because of the large dissipation but at that point i will simply use a lower input voltage like 5V USB)
Regarding the DACs for control i plan to use I2C 8 bit ones for now ( there is a 12bit drop in replacement that i can se later or for a V2)
 

[coromonadalix]

overdesigned   with an lt3080

you could do it with multi turn pots

and for sur a linear psu will dissipate more than a smps or any switching psu

[Sniper1]

well i want to avoid pots and use a DAC since this way i can avoid some mistakes and can also ramp up nice from soft + atm the circuit can accept a pot in the DAC  net labels
Regarding the LT3080 tht seems excessive + not so much in stock ( also like 20$ last time i did check )  + if i use a more common Darlington NPN or Pmos it is easy to replace but wwith the LT3080 i am sort of stuck with only 1 component

PS SMPS preregulator is also added but i am confident there while on the analogue linear side i am not ahtat is why this blcok is designed to work as standalone so that i cn more easily test it / simulate it

[Kleinstein]

Going linear with up to some 12 W of power loss this may not work that well with a LT3080. Even if the chips can just handle the heat the temperature rise compromises drift and accuracy.

This is more like something for a seprate regulator and power transistors. This could be an LM723 based circuit (usually pretty poor constant current performance and going down to 0 V need a little care) or a more classical one around OP amps with diode OR. There are a threads around a LM324 "based one" that may work - though I would replace the 324 with at least an MC33174. The plans around are more for 3 or 5 A, but scaling down to 1 A should be pretty easy.

[Sniper1]

THX
regarding the op am i agree i will use something with under mV input offset
A and i also found the problem in my simulation it was the CC part and more precisely it was the op amp , it acted as a compactor not as a differential amplifier , i blame the model , the voltage regulation seems to be working well ......
I cant get the circuit to compensate 2 diode drops but that also may be because of the op amp model , ill search a bit on LCSC and see if i have that model in SPICE (or just use a similar model and hope it works....

Is the thread you are reffering to this one: https://www.eevblog.com/forum/projects/bench-cccv-psu-based-on-daves-usupply/?action=dlattach;attach=690201;image  ?

[PCB.Wiz]

As for the specs I am aiming for they would e loosely this:

• 0V out ( i plan to use 2 Si diodes and have the output come from a floating GND that is 1-1.5V higher then true GND
• 0-12V output range but i can settle with less if it makes sense ( i think here i am sort of limited by the transistor and op amp input voltages + max dissipated power by the heat sink)
• low noise ( that i why i want ti to be linear)
• As for output current i want like 1A for 5-6V and under
• Stable , aka no oscillation and the CC limit to work well

There are parts designed for CV/CC operation as Linear Controllers - 
eg dual opamps + ref like AP4310E etc, and more dedicated parts like MP26085.
Circuits using those have separate loop compensation for CV/CC pathways.

So i am trying to include a small constant current constant voltage supply in one of my projects ( active load) and after searching the web i found Dave s design that uses just 1 power element (very important for me since i only have 1 heatsink slot that is not used, the heatsink is recovered from something old and it 2.5x1.5x10cm long i calculated online that it can dissipate 15W with no fan and that would be plenty for me atm)

That's quite long and skinny.
What packages does that heatsink support ?

[Sniper1]

|| There are parts designed for CV/CC operation as Linear Controllers -
eg dual opamps + ref like AP4310E etc, and more dedicated parts like MP26085.
Circuits using those have separate loop compensation for CV/CC pathways.   ||

I will check their datasheets and think about them .

||   That's quite long and skinny.
What packages does that heatsink support ?  ||
Originally it had 2 TO247 + TO220-2   i plan to use the TO247 for my active load  and the TO220 for the linear power supply BUT since the heatsink is like super flat on that side replacing the TO220 with an TO247 would be easy.
BUT if i use a drill and a nut on the other side i think i can fit 2 TO220 ( but i need to find a drill ) also it wont make it dissipate more .
After a fast search it seems to be more or less like this : https://www.aliexpress.com/i/1005002309538018.html   ( this is rated for 20W , mine is slightly smaller + chinese W so that is why i said 15W ) 

[Konkedout]

When driving a MOSFET gate with an op amp there is the risk of instability.  The MOSFET gate is a sort of capacitive load.

Put 1K ohms or so between the op amp output and MOSFET gate.  The op amp inverting input must have a few K ohms or more in series with it.  Now add a capacitor (maybe 1-10 nF) between the op amp output and the inverting input.  You can add a resistor in series with the capacitor to improve the transient response.   This can all be simulated if you have good models for the op amp and MOSFET.

[deepfryed]

TSM108 is another step down regulator with CC/CV control but may not be considered "linear" in a traditional sense. If you want low ripple output, you will need a few low pass filter stages to clean up the output.

[deepfryed]

not sure if it's of any help, the Onsemi datasheet for LM317 has an example for CC/CV lab supply.

[Sniper1]

So i was designing with 2 LM317 and it should work on paper BUT somehow the LDOs get stuck and dissipate a lot and that breaks the simulation , is there a way to add a heat sink to them or ignore the heat ?

[Neomys Sapiens]

not sure if it's of any help, the Onsemi datasheet for LM317 has an example for CC/CV lab supply.

This is NOT a CV/CC power supply circuit. It is a CV circuit with a scabbed-on current limiter.
There is a LM317-based circuit for a CV/CC supply, but it does not extend  down to 0V.
(from the original NS datasheet, appears also in several appnotes pertaining to 3-terminal adjustable regulators)

[xavier60]

This is the design with diode ORed opamp CV/CC loops. My build has been working on my bench for 3 years.
It can be simplified somewhat.
https://www.eevblog.com/forum/beginners/lm324-power-supply-with-variable-voltage-and-current/msg3582664/#msg3582664

[Marco]

overdesigned   with an lt3080

Such a bizarre part to use too for an adjustable power supply, add a 1 to the device name and it has the entire control loop for CV/CC build in and you can get a reference design for a switching+post-regulator adjustable power supply.

[coppercone2]

I think its a fine part for your power supply because you are actually taking many hours to make it rather then buying one on ebay for $30

not exactly breaking the bank considering hours spent no matter what part you use

and if you built a nice one for yourself with expensive parts, you will find that when you have to make the cheap standard issue cheap crap at work, its interesting since its novel..

infact, you can put a bigger heatsink, and 2x lt3080, and nothing bad should happen IRL, at the cost of about 10$ extra.

each 3080 will respond to over temperature also, so you don't need any mechanical shenanigans with thermal coupling

[Marco]

But it's not much cheaper than LT3081, which is just plain better.

[coppercone2]

a even better choice. Or the LT3083

[Sniper1]

THX i will analyze it and replace the pots with my DACs

[Sniper1]

interesting i think we may have found a winner , LOL ( if i can get it to work with a DAC)

[coppercone2]

that will be a powerful project

[deepfryed]

I haven't read through the data sheet completely, but I don't think LT3083 has programmable current limit.

[Marco]

Yeah, it's the LT3081. Though if you want a digital programmable one instead of setting it with a pot you need to get a bit tricky. You need to do something like use a 390 Ohm resistor between ilim and the output and then push extra current through that resistor to set the limit.

[Sniper1]

True but apparently you can use a pot to change the set pin voltage so you can also use a DAC+ op amp magic BUT then again i am where i was before with the LM317s
BUT if i use a LM317 for current limiting and an LT3080/3083 then i may just make it work

[Marco]

If you use the LT3081 you don't need the LM317 and you get a stable and very fast CV/CC control loop which can work with no output capacitance.

[Sniper1]

I will also check that one, main problem with it is that it is not available on LCSC as of now , so i would need to move my order to mouser (  and cheek vs the shipped threshold since i want cheap/free shipment)
PS the LT308x series is in a way voltage controlled via a output current in the uA range so i need to be carefull when i convert it to DACs