Linear Regulator using FETs for pass device

[BrianG]

OK, let me start by saying that I know this is NOT the forum for this type of question, but I hope there is enough brain power here to help me out.

Here is an excerpt from a thread I started a while back about a couple of transformers I got:

Originally Posted by BrianG

The second transformer (1,000w) was a little larger with a little bigger windings. The output of this one was 15.9v AC unloaded with no center tap. With a 0.75 ohm load, the output dropped to 15.1v AC, which equates to 20.1A. Calculated output impedance is 0.04 ohms. With a 0.46 ohm load, the output was 14.6v (calculated 31.8A). Even though this one is capable of higher power and has lower output Z, the output V starts out too low (unloaded) to be useable for a PS at heavy load unless I make it a strict 12v PS. Then, I could use this to its full potential up to 65-70A.

Since then, I got another one of these transformers. The really low output impedance is good for a fairly stable voltage output. I want to build a regulator out of these, but I have two issues:

1: Loaded to ~30A, the AC output drops to 14.6v AC. So, if the regulator is built to supply ~13vDC, I think that the output will drop too far for the pass transistors to work, even after rectification and a lot of filtering. So, to minimize the drop-out effect, I need to set the PS to regulate down to ~12v. So, my choices are: be happy with ~12v, or find a solution with a lower drop-out value. If I can find a low drop-out solution, I can increase the load to get more current out of it.

2: Coming up with a pass transistor solution to pass at least 40A of current AND have the power dissipation value is problematic. Paralleling bipolar transistors is a no-no unless you do all kinds of magic to prevent individual thermal runaway.

So, I'd like to use MOSFETs. Their low rdson value would make for a low drop-out regulator. And they are easily paralleled (not to mention paralleling would reduce the total rdson). However, I have no idea on how to bias them for use in linear mode. Can anyone help get me pointed in the right direction? Artur maybe?

[BrianG]

Anyone?

[BrianG]

I know this is an old thread, but I recently dragged out the transformers in post 1 and want to start building power supplies with them. So, I'd really appreciate it if someone chimed in, especially about question #2.

[Metallover]

I'd love to answer your questions, but I don't have the knowlege to do so. Please check back when I get an electronic engineering degree...

I say look on another forum?

[BrianG]

I was gonna do that, but I assumed there would be enough intelligence here...

[Metallover]

http://www.edaboard.com/ftopic147431.html

http://powerelectronics.com/power_se...hstand_stress/

[BrianG]

The second link requires login to see any schematics, but it just talks about paralleling bipolar transistors. You have to be very careful with regular transistors because as they heat up, they conduct more. This leads to "current hogging" and the best transistor of the parallel array (best meaning highest current gain and lowest C-E saturation voltage) will hog most of the load and burn out. Then, the next best transistor will do the same, and so on until they all burn out. There are ways to safeguard against this (using small value, but large power resistors on the emitter), but it is a PITA IMO. Not to mention the voltage drop on these resistors creates heat and power loss (lower efficiency) and means you need a higher input voltage than normal to ensure it stays high enough above the desired output for proper operation (also known as drop out voltage). Much easier to parallel FETs because they conduct less as they heat up, which results in a sort of "self-regulation".

The second page is probably closer to what I am looking for, but some important diagrams are missing.

Thanks for the effort though. I guess I will either keep searching, or use a bazillion PNP pass transistors in parallel with emitter resistors. Ugg.

[Metallover]

Now all this is to make a power supply?

[BrianG]

NVM about the missing diagrams in the second link - they are in the PDF...

[BrianG]

Originally Posted by Metallover

Now all this is to make a power supply?

lol, yeah. These transformers I got for free are very beefy, able to supply at least 50A @ 14.5v, and I have two of them. If I can come up with a fairly simple regulation circuit, they will make for inexpensive and high current supplies.

[BrianG]

OK, after reading a bit more, I have decided. It IS possible to do what I want, but the cost of just the FETs would be well over what I consider "inexpensive". As such, I will just get a bunch of bipolar transistors (TIP33 looks like a good candidate) and design in some emitter resistor compensation. At BGMicro, I can get all the parts needed (minus what I already have) for TWO high-amperage supplies for well under $100.

[Metallover]

Is that your store, BGmicro? Or did they name it after you?

[What's_nitro?]

Hmmm?? IRF540N

@25*C - 33A, 100V, 0.04ohm RDSon, 120W.

How did I miss this thread?

[What's_nitro?]

Or these... TIP35

@25*C - 40-100V, 25-40A, 125W.

I'm just scrolling through my PDF datasheet database...........

[BrianG]

I didn't look at the specs of your linked FET, but it looks like a typical FET and those are meant to be used in switching applications. They typically don't have good performance when used in linear applications due to the hot spots on the die (despite their high rated power dissipation). And the proper FETs are expensive, so the FET idea is dead. Although, I might be able to wring a little more efficiency out of the supply at lower currents by switching the recified output of the transformer by using FETs at the switches, but not sure if I want to go there. I got these transformers for free so I don't want to spend a lot of cash.

As far as bipolar transistors go, I aloso have to have a decent hfe (current gain) value. At high currents, hfe goes down, so I need to make sure the gate current will be fairly low. Otherwise, I would need another transistor in a darlington configuration to get more gain. But each stage introduces more drop out voltage. so I'd rather not do that. I plan on using an LM317T at the actual regulator, so I should only need one stage.

Anyway, the BGMicro store is not my store (I wish it was!), but they have components at FAR better prices than RadioShack.

So, I'm looking at using 10 PNP transistors, each with the following specs:
- 40V Collector-Emitter voltage min
- 10A min collector current (like to have a little headroom)
- Current gain of at least 20 @ 10A (will be higher at lower currents, but I like to spec for worst case)
- 50w power dissipation or better.