New experimental topology â€“ the tricky way to use multi-phase bridge~ / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

New experimental topology â€“ the tricky way to use multi-phase bridge~

1 2 next

Move Thread

LAN_403

DrRMR

Wed Jul 18 2012, 11:48AM

Registered Member #4123 Joined: Wed Oct 05 2011, 07:47PM
Location: Shawnigan Lake, B.C. ,Canada
Posts: 5

New experimental topology â€“ the tricky way to use multi-phase bridge~

Hey guys,
This is my first time posting thread on 4hv. I was told that there are a bunch of very experienced electronics fans here, and it seems pretty true so farïŠ. I posted this topology that I came up with a few weeks ago on another electronics forum, but I thought nobody could really understand the significance of my invention over there. So I hope you guys will enjoy my project and give me some critical feedback if possible. That would be great!
Okay, back on our topic. I was thinking about the controlling a 4 phase motor with a 4 phase bridge. Usually multiple sets of windings are used in multi-phase motors, so I thought there might be a way that we can connect the output terminals of the bridges in contrary.
This is what the now topology looks like

As you can see, the output function of the whole system is very similar to the simple full bridge topology, objects that requires alternating currents such as a transformer could be loaded in the system. You might want to ask the reason of using a system which is way more complicated rather than a simple system that can do the same job. And Iâ€™ll just list the advantages and the disadvantages of the new topology as following so you can compare and see how itâ€™s gonna work.

Advantages of the new system:
1\ Each MOSFET or IGBT (simplified as MI in the following content) will work at a duty cycle of 25% or lower. Although the total power dissipation of the system does not change, but the dissipation on each of them will be smaller. In this way the MIs will be capable to exceed the power limit of their package (normally to-247 have a maximum power dissipation of few hundred watts ) due to the smaller thermal resistance.
2\Because the Irms passing through the each MIs have been halved. That makes the whole system capable for some extreme conditions such as Tesla Coils and Electric fishing machines. I believe the Ipeak that the system can handle will increase by a lot compare to the normal full-bridge.
3\Comparing to a paralleled full-bridge system, putting MI in parallel in order to increase its capability of high current, the new system can avoid the unbalanced current in the MIs which is the biggest problem in the paralleled full-bridge system. When the MIs are running in a paralleled full-bridge system high current, the slightly difference of the characteristics can cause one of them get a little bit hotter and decrease the capability of high current which will make is dissipate more and finally this poor MI will get into a vicious circle.
Disadvantages:
1\ itâ€™s really complicated. Take more efforts to make.
2\Twice as much materials required.
3\ More expensive.
According what I listed, I found it actually able to benefit a lot of projects. For ex we can push the to247 MIs to their extreme and get rid of the expensive IGBT modules in some of the application.
In commercial applications, there is a line between quality and price. As soon as the product reaches this line, increasing a little bit of quality will cost a huge amount of money. I think spending money on the 4 more MIs is a wise deal to Increase productâ€™s quality.

That basically all I wanna say about the Theory. And I do have a real test for the new topology.
The schematic

So here is the Experimental project. I used a D flip-flop to divide the input frequency in to half, and then send the signal into a AND gate with the original signal that have a maximum duty cycle of 25%. Afterward, I sent the output of the logic core into a pair of MOSFET drivers (UCC27423 in this case) to drive two pairs of MOSFET Totem Pole in order to drive the Gate Driving Transformer. Because I made the GDT driver part separately so I couldnâ€™t put them all together nicely on one hole board. In this experiment I loaded the system with a Flyback Transformer.
Here are some pics.
This is the driver

And the GDTs

The actual bridge
The wave form of one of the Gate

The output waveform while not arcing

The output waveform while arcing

And the arc @12v
The input current while arcing

The temp of one MOSFET after 1min

Compare to the temp of the room.

And that all I have~
Let me know if you have any questions and concerns so we can improve it ~
Thanks a lot! Have a great day~

[/link]

Steve Conner

Wed Jul 18 2012, 01:10PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

There is one flaw in your idea: When two devices carry a current alternately as you propose, the total losses are unchanged, because the whole current always flows through one device. The dissipation hasn't been reduced, just spread over two devices.

When two devices are used in parallel, the total losses are halved. Each device carries half the current, so has one-quarter the I2R losses, and two quarters make a half.

Therefore, your idea needs twice the silicon area of parallel devices.

I made a number of simplifications: I assumed ohmic devices (MOSFETs, not IGBTs) and perfect current sharing of paralleled devices. So, the real-world advantage due to paralleled devices will be less than 2:1. In particular, if the devices had a voltage drop that didn't vary with current, there would be no advantage at all. But even IGBTs have some ohmic component to their voltage drop, so I think parallel devices will still show an advantage.

One application where this idea makes sense is the multi-phase buck converter, as used in CPU core voltage supplies. 3 buck converters with their outputs connected together performs better than one big buck converter made with 3 mosfets in parallel. They are made up to 12 phases to handle modern CPUs that require over 100A at 1.5V.

Mr.Black

Wed Jul 18 2012, 01:16PM

Registered Member #3960 Joined: Mon Jun 20 2011, 06:27AM
Location: Wuhan Hubei China
Posts: 33

pu~it is equivalent of two full-bridge power combiner

Steve Conner

Wed Jul 18 2012, 01:18PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Yes, in his QCW Steve Ward uses two full-bridges with ferrite balancing transformers to combine the outputs. It behaves like parallel devices but the transformers force almost perfect current sharing.

You can combine any number of bridges in this way: the Harris 3DX50 AM transmitter uses something like 48 full bridges of IRFP460s to make 50kW of medium-wave goodness.

Jinkela

Wed Jul 18 2012, 01:23PM

Registered Member #5788 Joined: Tue Jul 17 2012, 11:25PM
Location:
Posts: 9

I told you,your circuit is equal to two MOSFET parallel full bridgeâ€¦

Steve Conner

Wed Jul 18 2012, 01:26PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

No it's not, it is worse (half as efficient for a given silicon area)

Mr.Black

Wed Jul 18 2012, 01:35PM

Registered Member #3960 Joined: Mon Jun 20 2011, 06:27AM
Location: Wuhan Hubei China
Posts: 33

So it seems

Mr.Black

Wed Jul 18 2012, 01:44PM

Registered Member #3960 Joined: Mon Jun 20 2011, 06:27AM
Location: Wuhan Hubei China
Posts: 33

Transformer power combiner?
but i consider the large current may lead to transformer magnetic saturation

DrRMR

Wed Jul 18 2012, 03:17PM

Registered Member #4123 Joined: Wed Oct 05 2011, 07:47PM
Location: Shawnigan Lake, B.C. ,Canada
Posts: 5

Thanks for your comments Steve
Concluding what you said, the efficiency of my design is truly a big problem. I didn't know much about the way people work with paralleled power devices when I had this idea, all I saw was a application note from IR which was saying that devices for paralleling need to be carefully selected. So I thought putting devices in parallel within the full-bridge application could cause some trouble.
Another point of this idea that I'm talking about is the decreased thermal resistance. Although I found It perform better than a normal full bridge, but it actually might be worse than the paralleled bridge you mentioned. The only problem is I've never compared it experimentally with a paralleled bridge. My design lost in theory, but I would prefer to make it solid by a experiment.
Thanks a lot

DrRMR

Wed Jul 18 2012, 03:20PM

Registered Member #4123 Joined: Wed Oct 05 2011, 07:47PM
Location: Shawnigan Lake, B.C. ,Canada
Posts: 5

And by the way, do you mind share a little more information about Steve Ward's combiner? I'm really fascinated about that.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob