Building a simple high current 1.5V DC-DC power supply - how do I stabilize it?

[momaka]

I'm trying to build a high-current (possibly up to 10A) 1.5V DC-DC power supply. The bad OST caps in my Shuttle SB83G5A took out the Intel 915G Northbridge power supply (upper MOSFET and its PWM controller), so I thought maybe I can build my own power supply for the NB with junk parts I have around.

Attached below is a design I have already tried on my breadboard. It works fine with low loads (up to 0.5A) but as I increase the load (to about 1.5A), the voltage begins to overshoot to about 1.6V. As I further increase the load (tried up to about 4.5A), the voltage then sags down to 1.42-1.48V and the power supply seems to be oscillating (there's a rapid fluctuation between above voltages).

So my question is (or rather questions):
- Is the power supply oscillating?
-- If yes, how do I stabilize it? I read a bit on compensation just to understand that there are 3 basic types and that I am using type 1 (since it's the simplest). The math stuff / calculations is above my level, though.
-- If no, could the lousy regulation be because I'm using an LM358 op-amp and not something else (and if so, what should I use)?

I tried playing around with the compensation capacitor, Ccomp. Initially, I had it at 10 nF. Doesn't seem to make a big difference when I tried 68 nF and 100 nF. It does make a big difference if there's nothing in it's place, though - the PSU whines loudly even with a small load. As for compensation resistor, Rcomp, I have 1 KOhm. Perhaps I need to have a different value here?

As for the output filter capacitor Co, initially I was using a 4V 820 uF Fujitsu FPCAP RE series polymer in parallel with a 16V 1500 uF Rubycon MCZ. I thought maybe the low ESR of the Fujitsu was messing with the circuit so I tried a general purpose 10V 2200 uF cap in its place but there isn't any notable change in the behavior of the power supply with the same 3 loads. The MOSFET I'm using for Q1 is a FQP19N20C (19A, 200V, N-channel). And for Ci, I have nothing - just wires from the 3.3V rail on the PSU going straight to the MOSFET drain. Perhaps this can cause problems too?

I also tried to add a diode and inductor (taken from a PSU choke) on the output to make it like a buck power supply. But that didn't make a difference in outputs either. And from what I read on compensation, I would need a type 3. But looking at the formula for that, I think I couldn't do it even if my life depended on it.

Ideas? Questions? Comments?

I really like circuits, but I suck at math (and even more in school in general). Would greatly appreciate any and all help. Thanks.

[Th3_uN1Qu3]

Re: Building a simple high current 1.5V DC-DC power supply - how do I stabilize it?

Since you're going with a linear regulator (not a good idea, low efficiency -> will need big heatsink), you could just go with a classic circuit which has the compensation built-in, like the LM317 circuit with external pass transistor. I would use a PNP power transistor as they have a slightly lower voltage drop.

[randtek]

Re: Building a simple high current 1.5V DC-DC power supply - how do I stabilize it?

A technician friend and I spent quite a bit of time some years back trying to do what you are doing-using an OP Amp to control a regulator. We found the same issue-it is not stable under varied current situations, especially under high current. For an application like this, you need something more suited to regulating. I agree with Th3_uN1Qu3. A LM317 and external pass transistor, possibly with a high wattage dropping resistor as well, to help keep the transistor from having to dissipate ALL of the wasted energy.

[budm]

Re: Building a simple high current 1.5V DC-DC power supply - how do I stabilize it?


https://cdn.badcaps-static.com/pdfs/...52b97b4ce8.pdf
http://www.analog.com/library/analog...41-05/ldo.html

https://cdn.badcaps-static.com/pdfs/...56c96dea71.pdf

[momaka]

Re: Building a simple high current 1.5V DC-DC power supply - how do I stabilize it?

Well see, the idea for my circuit above came from the way some PSUs regulate their 3.3V using a MOSFET and a 431 shunt. In fact, my circuit is exactly the same as the 3.3V rail circuit used in a Raidmax RX-380K (Sun Pro KY-480ATX) PSU, except for the fact that I am using a 358 op-amp instead of a 431 shunt (and also much better caps - Teapo SC vs CS/GL "LOWESR" caps) since common 431 shunts can't regulate to lower than 2.5V. Even the compensation in my circuit is the same: a 10 nF cap and 1 KOhm resistor.

So my logic is, if you can pull several amps from the 3.3V rail of those PSUs, then wouldn't it be possible to scale down this circuit for a different voltage?
The only thing I can think of is if those power supplies that use this circuit were never actually capable of supplying more than a few amps on the 3.3V rail. And I'll be honest that I have seen this circuit only in extremely cheap power supplies (which the Raidmax I mentioned above sure is).

Either way, it looks like I'm beating a with this circuit, though . The good thing, however, is that I read quite a bit of material on compensation so I did learn a few things along the way.
After reading this TI application note for the 3rd time, I finally began to understand what those capacitors and resistors do for the different compensation types - it's more or less just calculating how the impedances affect the gain in the op-amp circuit. However, I haven't done any complex calculations in over a year now so I didn't and won't try attempting to do so. Also, that TI app note is for a buck design rather than linear, so I'm pretty sure that also has to do with the fact why my experiment above couldn't achieve the desired results.

Regardless, I did try playing around with a type 2 compensation with my circuit above just for the heck of it. Of course, there was hardly any change in the performance. In fact, some of the resistor/capacitor combos I tried made the circuit perform even worse (1.8V under 1A of load and 1.35V under 4A of load).

So time to ditch the circuit, I guess....
Or maybe it will work if I try driving the MOSFET with a 555-based PWM controller (with fixed duty cycle) and have the op-amp disable/enable the 555 timer to implement pulse skipping?

Since you're going with a linear regulator (not a good idea, low efficiency -> will need big heatsink)

Actually, even when I was drawing 4.5A with the original circuit in my first post, the heat sink I have on the FQP19N20C never got more than just slightly warm. The heat sink is about 20 cm^2 and 1.5 mm thick. Of course I know that doesn't mean much.

you could just go with a classic circuit which has the compensation built-in, like the LM317 circuit with external pass transistor. I would use a PNP power transistor as they have a slightly lower voltage drop.

Sounds good. I do have a LM317 regulator and TIP42 PNP transistor, both in TO-220 cases. The only reason I wanted to make the op-amp/FET circuit above is because I have A LOT of MOSFETs and quite a few 358 op-amps. I also have a good source of fixed and adjustable 1117 LDOs. Can those work in place of the LM317?
Perhaps my title should have been more clear - I'm trying to build a circuit AND clean up junk I have

@ budm: thanks for the links. Some of them are a bit above my knowledge level, though.
And if I have to go as far as getting a linear controller, I might as well just pull a buck controller from one of my dead motherboards and play with that instead - might be just as easy. Actually, I think that's one of the next few things I might try.

I was also thinking maybe mod a Deer PSU Th3_uN1Qu3 style by playing with the DBL/TL494 reference voltage.

[tom66]

Re: Building a simple high current 1.5V DC-DC power supply - how do I stabilize it?

You misunderstand the 3.3V supply in a modern group-regulated. In the majority of PSUs, it uses a mag-amp type circuit. The current through the inductor can be controlled by making it controllably saturate, which makes it behave a lot like a short circuit. It's a self-oscillating switching supply. It's a very neat design; sort of like a buck converter, and reasonably efficient. This explains it more:
https://cdn.badcaps-static.com/pdfs/...dca16b7c32.pdf

[momaka]

Re: Building a simple high current 1.5V DC-DC power supply - how do I stabilize it?

You misunderstand the 3.3V supply in a modern group-regulated. In the majority of PSUs, it uses a mag-amp type circuit. The current through the inductor...

I know the PSUs that use the magamp circuit - those are easy to distinguish with the 2 inductors for the 3.3V rail (a small one for the magamp circuit usually near the main traffo and another for filtering next to the 5/12V common-mode choke).
But in the PSU I was referring to, there is no inductor. Instead, the rectified 5V rail is sent to the drain of the 3.3V MOSFET and its source is connected to the 3.3V output. Gate is pulled up with a 1 or 2 KOhm resistor connected to the rectified 12V rail, and the switching of the gate is done via the 431 shunt.

Thanks for the link, though. You are right in saying that I don't understand how those magamp circuits work . Up to this point, all I know is just how to identify them (read: monkey technician ).

[tom66]

Re: Building a simple high current 1.5V DC-DC power supply - how do I stabilize it?

I know the PSUs that use the magamp circuit - those are easy to distinguish with the 2 inductors for the 3.3V rail (a small one for the magamp circuit usually near the main traffo and another for filtering next to the 5/12V common-mode choke).
But in the PSU I was referring to, there is no inductor. Instead, the rectified 5V rail is sent to the drain of the 3.3V MOSFET and its source is connected to the 3.3V output. Gate is pulled up with a 1 or 2 KOhm resistor connected to the rectified 12V rail, and the switching of the gate is done via the 431 shunt.

Hmm, with a good heatsink, plus fan cooling, maybe possible up to 10A... but any more, you're pushing it...

Thanks for the link, though. You are right in saying that I don't understand how those magamp circuits work . Up to this point, all I know is just how to identify them (read: monkey technician ).

I have no idea how they work either, other than by controllably saturating the inductor and using it a lot like a MOSFET.