Power Supply Circuits

Inventor

Just last week I got a call from the patent attorney informing me that the deadline for my power supply patent is approaching. Unfortunately I had to tell her that the funding source dried up and we cannot proceed. So now I have this collection of power supply circuits that are possibly new and possibly of some benefit to the world, and nothing to do with them. So I thought I would post here about them.

The fundamental idea is to replace the diode of a linear or switching power supply with a MOSFET. That part isn't new, in fact the patent search revealed several patents where people were trying to figure out ways to do this. But the existing patents all had stuff like funky extra windings on the transformer or other custom specially tweaked stuff. My solution is simply a drop-in replacement for the diode.

I made several variations: one with transistors, one with a comparator, positive and negative outputs, and a bridge rectifier topology.

If you're interested, I could post some figures or the provisional patent application text, or maybe I should create a web page of it? I hesitate to do so because I posted some of my electric circuit creativity on Usenet and received harsh criticism and personal insult. It was very disheartening. My creations, both tangible and intangible, are my gifts to family, friends, and to the world so getting verbally attacked in response to one was quite a downer. This forum, however, had no such criticism so that says something about the people here.

I welcome your comments...

Inventor · Posted: Sun Mar 30, 2008 12:12 am Post subject:

Here is a photo of the prototype that I made. I took a piece of wood composite board, I forget what it's called, but it's natual medium brown in color as in the photo. I made a briefcase insert out of it with some standoffs, then I drilled it for mounting all the parts.

To the left is the Wattmeter for measuring input power from the wall socket and the voltmeter that measures output voltage. To the lower left is the input transformer, the invention in a plug module, and a filter cap and load resistor on plug modules. I did the modular connections with banana jacks for a retro style and good connections. Along the top are some other modules that plug into the circuit.

I did a basic test where I put in a power supply diode and calculate its efficiency and then swap it out for the new circuit and again calculate. The new circuit is 7% more efficient than the old diode, if I recall correctly.

Actually the prototype is not ideal, it's just a demo and there's loss in the wires and the connections and such the way i made it, so the total efficiency is not that great, but what matters is the delta between the two rectifiers as far as I know.

Although the circuits may not be novel, I had fun thinking them up and proving that they work so I guess it was worth doing. I may post some schematics soon, so watch for new posts...

Power_Supply_Prototype.jpg

Description:

The briefcase-based power supply prototype that I made

Filesize:

316.16 KB

Viewed:

376 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Inventor · Posted: Sun Mar 30, 2008 6:18 am Post subject:

We can define the problem and get all cozy and familiar with it by looking at the familiar diagrams shown in the attached image. On top is Figure 8, the Half-Wave rectified linear power supply, and below is Figure 9, the Full-Wave rectified version.

There is one slight twist, however, the diodes have been replaced by a three-terminal block diagram symbol. That's the thing that has always been a power diode in the past. The difference is that I added a ground connection and called that three-terminal thingie a "diode replacement".

Studying the circuit for losses, we know there will be some losses due to transformer leakage and parasitics and some loss in the capacitor series resistance, but most of it is in the diode (besides the load, of course). Wouldn't it be nice if we could just plug in something and magically erase most of the diode power dissipation?

Well, we can. And I'll show you the circuit in the next post.

PowerSupplyFigure08.jpg

Description:

Half-Wave and Full-Wave circuit topologies with ideal diodes in block diagram form

Filesize:

41.42 KB

Viewed:

389 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Inventor · Posted: Sun Mar 30, 2008 6:54 am Post subject:

OK, here's where I show you the fun stuff. Take a look at the schematic images below in Figure 7 (a) and Figure 7 (b). The transistor-based circuit is in (a) and the comparator-based circuit is in (b). You plug either circuit in for the ideal diode block diagram of the previous post and it performs the diode's rectification function with much lower loss.

Of course, it's a different beast so it has different requirements, characteristics, and performance. The MOSFET has to be one of those "new" ultra-low on-resistance MOSFETs such as a good HexFET from International Rectifier. Also it's not too happy delivering anything below a +/- 5 V output, and it's really at its best with +/- 12 V. Then again, manufacturers are making power MOSFETs with lower gate voltages nowadays so I don't know but that may help this situation.

The way the transistor circuit works is the left transistor provides a diode drop to balance the diode drop of the right transistor, so when the input and output voltages are equal there is no current flow in the base resistor so the right transistor is shut off. This permits the pulldown resistor to pull the gate voltage down to zero, turning the MOSFET on. If the input voltage is greater than the output voltage, the base current of the right transistor flows into the transistor, shutting it off and allowing current to flow from left to right, pulling up the load voltage. If the input voltage is less than the output voltage, we want the MOSFET off, and we get that because a reverse current forms in the base resistor, turning on the right transistor which then shuts off the MOSFET. Alternatively one can think of that state as having the MOSFET be diode-connected in the reverse direction.

The way the comparator circuit works is more obvious to me at least. You want to compare the input and output voltages, so just compare them. The trick is, the way other patents did this was to create a separate little DC supply rail for the comparator. That's crude yet effective, but even better is the circuit of Figure 7 (b) since the comparator simply runs off of the output DC signal. The only trick is we have to divide down the compared signals to get them within the input range of the comparator, but that's easily done with resistors. Also, not shown is the hysteresis that I found was necessary when building the circuit.

These circuits will work in the steady-state condition but what about during start-up? At start up, the circuit in (b) has 0 V supply voltage on the comparator, so the comparator cannot drive the gate voltage reliably. We need some start-up circuitry to get things going. To do that, I just put a resistor to ground from the gate of the MOSFET in (b), pulling the gate down so the MOSFET is on when the comparator is out of commission. In both cases, the substrate diode exists to act as a standard diode during low-voltage start up.

The circuits aren't perfect, they do have resistors so there is some slight loss there. The MOSFETs are typically 12 V gate-drive parts, or at least the ones I was able to get a year ago, so they have greater on resistance when driven at 5V. But counterbalancing that is some good news: we can parallel these MOSFETs. They don't take much space so we can put a bunch of them on the circuit board if we want. They work just fine in parallel and their resistance will reduce to Req = Ron / N, where N is the number in parallel. Later I will discuss a situation where we put two MOSFETs in series and take a 2X hit on dissipation in order to gain functionality.

I don't know if the circuit described in this post is novel, but I created it about two or three years ago and it might have been novel at that point, not sure. Anyways it's probably been discovered by industry because it's so beneficial to power supply efficiency, and the trend of replacing diodes with MOSFETs is well known. Whatever.

PowerSupplyFigure07.jpg

Description:

The MOSFET-based diode replacement circuit, two example topologies

Filesize:

46.37 KB

Viewed:

397 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Inventor · Posted: Sun Mar 30, 2008 7:28 am Post subject:

Next I'd like to show you the full-wave bridge that I came up with. The closest thing the patent search found to this thing was a patent by HP for a rectifier that had four MOSFET switching elements, but it was a brute-force approach with a little bitty power supply for each MOSFET driver and four control circuits. Anyway, the near-zero loss bridge is coming, and this is one of the designs.

Take a look at the image below and you will see a bridge rectifier where the four diodes have been replaced by these MOSFET diode replacement circuits. The only trick was I had to figure out where to put the third connection. I tried a few ways and the way shown below worked.

But it's not quite that simple. To make the bridge work we need both positive and negative versions of or MOSFET circuit. If we take the circuits of Figure 7 and flip them upside down, replace the P-channel transistors with N-channel ones, and replace the pnp transistors with npn's, then we have the dual form of the diode. This dual form works on the bottom, or return path. So in Figure 10, the two diodes to the upper right are positive diode replacement circuits, and the two to the lower left are negative diode replacement circuits.

There is one gotcha of this circuit. During the brief switching interval that an input sine wave is below the output voltage, the bridge oscillates. Oh my! That's because when the voltages are reversed like that, the bridge acts like two inverters connected together and they create an oscillator. The only fix I found for that was to replace one of the MOSFETs - any one of them, really, with a diode. So you end up with 3/4 of the possible power savings which is still pretty good, and it's simple, easy to build, and stable. There may be other ways to stop the oscillation such as changing the size ratios of the MOSFETs as is done in SRAM cells to prevent oscillation.

So that is my bridge rectifier. I think it is new but I'm not sure. Hope you get a kick out of it.

PowerSupplyFigure10.jpg

Description:

An efficient full-wave bridge rectifier

Filesize:

41.99 KB

Viewed:

366 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Inventor · Posted: Sun Mar 30, 2008 8:01 am Post subject:

The previous circuits are useful enough, but what if we want regulation of the output voltage? Well, for that we need a reference, an adjustment, the output voltage and the input voltage. The circuit topology for one such example is shown in the attached image, Figure 11.

The four inputs are connected to open-collector comparators appropriately and the circuit works well. That is, including a few mods which I forget what they are, but they involve start up and hysteresis. Anyway, you can tinker with the circuit in Figure 11, but there is one limitation. The cannot drop lower than one diode drop below the input voltage. This is because of the substrate diode conduction in that situation.

To solve this problem, we can just connect two MOSFETs back-to-back as shown in Figure 12. Just put those two in place of the one shown in Figure 11 and the regulation voltage can drop below the input voltage.

PowerSupplyFigure11.jpg

Description:

A regulated power supply

Filesize:

92.49 KB

Viewed:

382 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Inventor · Posted: Sun Mar 30, 2008 8:04 am Post subject:

Shown in Figure 13 below is a full-wave rectified version of the previous circuit (Figure 11). Note that, though drawn upside-down, the bottom circuit is just a duplicate of the top circuit but connected to the other side of the center-tapped transformer.

PowerSupplyFigure13.jpg

Description:

A full-wave regulated supply

Filesize:

124.32 KB

Viewed:

376 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Inventor · Posted: Sun Mar 30, 2008 8:09 am Post subject:

Um, correction, the previously posted bridge rectifier does not oscillate - this one does! lol, sorry for the confusion. Figure 14 below shows a bridge rectifier made from MOSFETs only but it has no directional control circuitry. It's just an interesting curiosity since it oscillates, though as stated before you can replace any one MOSFET with a diode to stop the oscillation, and resizing may help. But anyway it is kind of kewl to make a bridge rectifier out of something other than diodes, and it is lower power loss as well.

PowerSupplyFigure14.jpg

Description:

Simplified MOSFET-only version of full-wave bridge rectifier

Filesize:

48.05 KB

Viewed:

386 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Inventor · Posted: Sun Mar 30, 2008 8:16 am Post subject:

And just for fun, I'll finish up with Figure 16, the 555-timer based regulated linear power supply. Why did I make this? I just like 555's and find them to be useful on occasion. I wondered if it was possible and it worked in simulation but I never built it. It is a bizarre irony to me that I made a rectified power supply with a 555, a MOSFET, and no diode, but there it is. What fun!

PowerSupplyFigure16.jpg

Description:

555 version of the regulated efficient power supply circuit. Why? Because!

Filesize:

33.55 KB

Viewed:

390 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

DrJustice · Posted: Sun Mar 30, 2008 9:00 am Post subject:

Apart from observing that we'd get a bit higher part count to weigh against the reduced heat loss, I'm afraid I'm not qualified to comment on the actual circuitry, but it is an intriguing proposition. And a 'green' one too Smile

In any case, thank you for sharing your ideas!

DJ
--

DrJustice · Posted: Sun Mar 30, 2008 9:09 am Post subject:

BTW, I like the modular briefcase prototype - it's well cool Cool

DJ
--

Inventor · Posted: Sun Mar 30, 2008 11:00 am Post subject:

Oh, that reminds me. A switching power supply is typically 85% efficient, leaving us with 15% in losses that can be saved. So that 7% savings with the diode would presumably be about half of what there is to be saved. So really it's a 50% savings overall. So yes, very "Green".

Also, once semiconductor manufacturers integrate the circuitry, as HP's patent proposed, then they'll just be simple parts no more complex than existing parts. The bridge would have four terminals and the diode replacement would have three terminals. So that reduces circuit complexity a lot.

Inventor · Posted: Mon Mar 31, 2008 2:41 am Post subject:

Thanks for your comments, DJ. Yes I kind of like the briefcase too. It's very DIY. The banana jacks provide a really strong physical connection, which I like. I also did the retro styling with wood and simple, old-fashioned electrical parts to make a point. This efficient circuit was discoverable a long time ago, like any time from the 50's til today. If we had known about it in, say, the 60's, market pressures would have forced the development of the low Ron MOSFET a lot sooner, and we'd all have cooler power supplies by now for decades. It's a very "retro" discovery in a way.

Inventor · Posted: Thu Feb 12, 2009 4:06 am Post subject:

I just thought I'd tickle this thread to alert you to some discussion that's going on in another forum regarding this invention. The thread is here:

http://www.littleblackdog.com/topic49893.html

In that thread I describe the basic circuit and another engineer mentions a very interesting super high power application for the circuit. Some ideas are exchanged, and I also mention finding some Department of Energy resources for commercializing the invention, which I am currently pursuing.