Correct design for power supply ?

[Nachesko]

Hi guys, whats up, I'm new in the forum, I'm designing a power supply, its just for fun, and i want to know if the design at so far is OK. Here are the specs

Vin: 17 V (from a switch mode power supply, but replaced probably with a 12 v transformer+rectifier+2* 4700 uF caps)
Vout: Variable from 1.2 to 17 V approx 3 amp max
Soft latch switch for regulator

It's simple, but the idea it's to make it compatible with a uC, it's based on the uPower Supply from Dave, adding complete feedback from the output V and I, butt, go slowly. I send attached the design, i can't simulate it correctly (or i don't trust the simulation) in multisim, but it's easier than ltspice for me

(sorry for my English, i speak Spanish)

[Shuggsy]

With R7 being so low, I feel that your circuit will oscillate rather quickly. In Dave's video on his soft-latch circuit (), your R7 and C2 were put in specifically to prevent this. What you want to do is vary the RC time constant such that it will not switch Q1 and Q2 on and off very quickly. Those 2N3904's will turn on when the voltage on C2 is about 0.6V, so it's a rather small amount of time to charge that cap up to 0.6V with the components you have. The time C2 takes to charge up is a function of both R7 and your input voltage when the MOSFET is on.

[Nachesko]

Thanks for the response, i knew the function of C2 and R7 but no the voltage required to active the circuit. That's the problem with the simulation, i don't understand completely the time of simulation, and having in account the errors, generate me that problems.
 What do you think of the rest of the circuit ? I'm going OK ? do i need some class of pre regulator before the main regulator for avoid a voltage drop on the soft latch circuit on big charges ?
Again thanks for the response

[Shuggsy]

Not sure what the reason for R8 is except to maybe load down the input to the LT1084. Be aware that at 17V you'll need at least a 1/2 watt resistor there if you keep it as 1K.

The only voltage drop you should see from the battery/supply input should be whatever is dropped across Q3 due mainly to its Rds_on resistance... also be aware that Q3 needs to be able to handle passing as many amps (really, slightly more as the LT1084 is not 100% efficient and I like to always have some headroom). I'd suggest a MOSFET with as low on resistance as possible, able to pass at least 3.5-4Amps from drain to source, and able to have at least 20V on its gate to avoid getting blown out by that battery/supply voltage. Whatever your input, make sure that the peak voltage on the gate of the MOSFET isn't enough to harm the device. The 2SJ517 seems to have 2A of continuous drain current and a gate-to-source voltage of +/-10V. Since you're pulling the gate of Q3 up to your input voltage with R3 when Q3 is off, that's too high for the device. One solution is to make a voltage divider from your input down to ground in place of R3 to drop the voltage down... but this will also draw current when Q3 is off which defeats one of the advantages of Dave's circuit. Another solution is find a MOSFET which, as I suggested before, has a high gate-to-source max voltage. One example I found was this: http://www.digikey.com/product-detail/en/DMP4051LK3-13/DMP4051LK3-13DICT-ND/2179803, but anything that meets the specs I stated before that you aren't operating it outside its safe operating region (usually shown on the datasheet) should be alright.

I don't think you'll need a pre-regulator unless you want to boost the voltage up to 20V or so to make sure that you can get 17V out. The LT1084 has a dropout voltage of 1.5V max according to its datasheet, so with a 17V input, the most you could get out (without a voltage drop across the power MOSFET, Q3 in your case) is ~15.5V. In other words, with your given input voltage of 17V, I would actually spec your supply to go up to about 15V rather than the full 17 of the supply.

Otherwise, things look alright to me. If you want the circuit to handle load steps gracefully, the LT1084 should do a good job of this but more output capacitance can help too.

[ptricks]

looks ok but is way more than I would do for a power supply. You can cut about 95% of the parts you used by just using a toggle switch

[Nachesko]

I know ptricks, but the idea for this project is to make it compatible with a uC, and i need all that electronics before. The power supply for now it's only a practice, no specific function or needed specifically for something