Power supply project

[Rerouter]

you face an issue with the INA138, its input range starts at 2.7V, that means to measure with any level of accuracy, you need your input voltage at 2.7V when your supply is outputting 0V, though a bit more expensive what about a LT6105?

equally interested in your approach, my own design is facing some limitations and hoping others may have though up a unique approach,

[shebu18]

I go down to 1,25 with the LM338, the LM25085 can not go more than 1.25V down, so i will use 3,4 or even 4V when i want 1.25V on the output. I do not need less than 1,25V on the output. For me this is ok.

[shebu18]

So these is the schematic and the buck converter. I also have the uCurrent but it is not displayed.
If you have any comments please feel free to comment.

[shebu18]

I also finished the PCB layout. Now i need to order it at seeedstudio.


Yellow-TOP
Orange-BOTTOM


edit: What are good cap values for filtering spikes? 100n or 10u or both?

[T4P]

you can try both but 100nF is for higher speed spikes and 10u for slower spikes ( but obviously it can take more )
So yeah

[shebu18]

I will use both in an 0805 package.


What can i use instead of a SPST NC relay? I want to connect a pin to GND when the PSU starts, then using the MCU "free" the pin. It is about the RT pin on the LM25085. Tie it to gnd to enter shutdown and leve it unconnected for normal operation. So i want to start in shutdown mode and after i set the values the chip should exit that state.

[ptricks]

You can use a transistor latchable switch .

[shebu18]

i thought about that but no, i need it to be pulled to gnd right at the moment when i push ON. I could tie the base of the transistor to 5V, than after i turned the MCU on, set the rheostat and the output to pull the base to gnd, turn the transistor off and start the LM25085. Right now i had a relay with one NC and one NO contact, so i used this for the solving of the problem.


I will test the transistor method on the breadboard, maybe it would be better.

[shebu18]

I could use something like that in the picture. Q1 is connected to RT pin and GND and the base from Q2 is connected to the MCU pin. If i turn the psu on Q1 pulls RT pin low and the LM is in shutdown mode, is i saturate Q2 then he pulls the base of Q1 low and RT pin works normaly. So the LM is in the working state. I will try to add this method to the one with the relay.

[NewBeginner]

So these is the schematic and the buck converter. I also have the uCurrent but it is not displayed.
If you have any comments please feel free to comment.

@shebu18
I may be wrong but I don't think that the current limiting method used in Dave's design will work well with the LM338 (or other regulators that don't go down to 0V). Especially at low set voltages/set currents.
More details: https://www.eevblog.com/forum/projects-designs-and-technical-stuff/general-purpose-power-supply-design-7488/msg143536/#msg143536

Maybe Dave can shed some light on this .

For a higher current you could use the LT3080 with an external pass transistor (as shown in the attached image - extracted from the datasheet).
But in that case I think that care must be taken when selecting the filtering capacitors. A very low ESR capacitor can cause the output to oscillate (at least that's what my LTSpice simulation shows ).
That way you can use the LT3080 to set the output voltage and the external pass transistor to carry the large current, leaving only a few 10s of mAmps for the LT3080 to handle.

I did not test this setup so I may be wrong. I only did some simulations in LTSPice for this.

Thank you



Edit: I have attached the LTSpice simulation. The TIP147 model used is from: http://www.onsemi.com/pub_link/Collateral/TIP147.LIB

[shebu18]

Than i change the LM338 with a power Transistor and do the powersupply the old way. i will test the LM338 theory.

[NewBeginner]

@shebu18 Have you tested the LM338 theory? What results did you get?

Thank you

[shebu18]

Still not, i didn't found the time. i hope i will have the time this weekend.

[shebu18]

I tested the LM338. I had 5V on Vin, 1k1 on Vout to GND and a 100R on the adjust pin. The output was 3,75 with adj pin in the air. If i tied adj low the output fell to 1,25V and the current did not increase. As a test i tied for a short tiem vout to gnd and got to 3amps. I do not know what happens if you tie adj directly to gnd, but i think the lm will go to 1,25V

[ptricks]

The way to make these type regulators go lower is to feed them a -V on the ground pin. Of course you need a dual supply voltage, but those are pretty trivial for the load the regulator will present and designs are everywhere for how to convert a positive to a negative.

[NewBeginner]

So basically this current limiting method won't work properly with the LM338 (or other voltage regulators) if it won't go down to 0V.

I may be wrong but consider this case:
If you set the output voltage to be ~1.25V (Vout=~1.25V) and, say, you limit the current to 100mA, then when you short the output to ground you will get a current which is much higher than 100mA.
So the current is not limited in this case.
If I understand this correctly this is because the current limiting circuitry won't be able to lower the output voltage (by pulling the ADJ pin low) because the output is already at its lowest possible value (~1.25V).

Thank you

[shebu18]

Theoretically yes, i will need to test this particularly case when i get the boards.