VHF Laptop Power Supply (DART)

[scientist]

Just because they graduated from MIT doesn't mean they own an MIT patent.

[suicidaleggroll]

Just because they graduated from MIT doesn't mean they own an MIT patent.

http://www.google.com/patents/EP2606564A1?cl=en
Applicant: Finsix Corporation

[GiskardReventlov]

This isn't too exciting but maybe I'm missing some key point to this.  It seems to me that an innovative solution would be one that does away with the battery altogether.  How about a small butane (or whatever's in a cigarette lighter) powered bank of MEMS generators. Properly shielded and placed where the battery used to go. Go for a month or two on a single fill up. In a pinch a cigarette lighter could be used as a refill.

[calexanian]

WTF??!!!!!! smaller lighter electrons? Did anybody actually watch their video? Holy crap! And these are MIT students? What the hell is going on over there? They say they have put lots of time and effort into this project? Any competent engineer could knock out the basic design and have the gerbers off to the quick turn house in an afternoon!

Not to mention...


THIS IS A TERRIBLE IDEA!!!!!!

Why do you need to go that high in switching frequency? Just the distributive capacitance will kill you at 65 watts in that small of a package! Dielectric heating alone would be a safety hazard. Just the proper filtering alone at these frequencies to cover VHF would be larger and harder to build an tune than a traditional switcher. With new fets and high efficiency switcher designs I see no practical need to go that high in freq. Its all down side and no up. The ferrite size diff in that wattage range from low freq to VHF is not that much. You still need the same size bulk capacitance weather it be on the primary or secondary.

Young students with no practical product design experience should go to a grey beard once or twice to get this kind of thing ridiculed so they will think through this stuff before they broadcast their underdeveloped design to the masses. 

[miguelvp]

An old paper describing the technology they used (more detailed than their website) : "A Very High Frequency dc-dc Converter Based on a Class ?2 Resonant Inverter".

http://www.google.com/patents/EP2606564A1?cl=en
Applicant: Finsix Corporation

Other than name changing and terminology their dc-dc conversion part seems way too close to that paper by totally different authors.

instead of multi-stage call it multi-cell, buffer instead of storage, but it gets better, you can hit the prior art in the patent link and well the previous paper is there among others.

And look at the schematics too, yeah different they took the auto-transformer and replaced it with a capacitor. Of course a full rectified bridge so it's ac-dc converter instead of a dc-dc converter. Some pulse magic and voila new stuff, but the original paper has it covered with the multi-stage resonant driver and the on/off control.

But it doesn't matter if the patent holds I guess, most likely they won't get sued and if the original paper holder from 2008 makes a product I bet they won't try to sue them either.

I'm with GiskardReventlov but instead of a butane lighter just plug it in a cow, you get all the methane you need for you a hydrogen fuel cell to power your computer plus free milk! And tasty steaks once you need to replace the cow.

[scientist]

This isn't too exciting but maybe I'm missing some key point to this.  It seems to me that an innovative solution would be one that does away with the battery altogether.  How about a small butane (or whatever's in a cigarette lighter) powered bank of MEMS generators. Properly shielded and placed where the battery used to go. Go for a month or two on a single fill up. In a pinch a cigarette lighter could be used as a refill.

Butane is pretty flammable.

[calexanian]

But why has nobody picked up on the smaller lighter electrons thing?  am freaking bugged by that!

[wraper]

They say they have put lots of time and effort into this project? Any competent engineer could knock out the basic design and have the gerbers off to the quick turn house in an afternoon!
Not to mention...
THIS IS A TERRIBLE IDEA!!!!!!
Why do you need to go that high in switching frequency? Just the distributive capacitance will kill you at 65 watts in that small of a package! Dielectric heating alone would be a safety hazard. Just the proper filtering alone at these frequencies to cover VHF would be larger and harder to build an tune than a traditional switcher. With new fets and high efficiency switcher designs I see no practical need to go that high in freq. Its all down side and no up. The ferrite size diff in that wattage range from low freq to VHF is not that much. You still need the same size bulk capacitance weather it be on the primary or secondary.
Young students with no practical product design experience should go to a grey beard once or twice to get this kind of thing ridiculed so they will think through this stuff before they broadcast their underdeveloped design to the masses. 

Your arguments contradict one another. First you talk about the gerbers in a one day. If it is that simple why nobody sells them? Then you talk about realization problems. Are you serious at least a little bit?

[wraper]

WTF??!!!!!! smaller lighter electrons? Did anybody actually watch their video? Holy crap! And these are MIT students?

Specially watched that video. She says "smaller, lighter electronics" Clean your ears before shouting.

[PChi]

Some thoughts:-
I hope that it doesn't break the first rule in Power Supply design which is that it must be cheap.
It is small so needs to be high efficiency to keep the internal temperatures down which is another challenge. This could conflict with rule one.

How does it supply both the laptop and the USB which probably need different voltages?
The USB socket cables have no strain relief so doesn't look like a real product.

For off line switch mode power supplies the insulation required between the primary and secondary windings for safety cause leakage inductance which isn't helpful for high frequency switching.

I wasn't able to download the FINsix_Tech.pdf (internet too slow) but I don't see any new ideas. I'd be frightened of the patented technology claim. What new technology have they created?
The losses due to switching equal the loss per cycle times the frequency so at a higher frequency the claim to waste far less energy per cycle is a must.

Another problem is high voltage switches have high capacitance. The switch drive circuit is going to loose 1/2 CV squared every cycle. It's another reason off line supplies run at lower frequencies than low voltage switched mode units.

[calexanian]

WTF??!!!!!! smaller lighter electrons? Did anybody actually watch their video? Holy crap! And these are MIT students?

Specially watched that video. She says "smaller, lighter electronics" Clean your ears before shouting.

I know what I many not have actually heard. wait...   

I would really like to see what harmonics are coming off of that sort of design though.

[PChi]

I was able to access the  FINsix_Tech.pdf. So they intend to use air cored hence more turns for same inductance and with no core and more turns the leakage inductance will greater. I can envisage getting a non isolated supply to work (not easy and high efficiency difficult) but an isolated supply really challenging.

[wraper]

I was able to access the  FINsix_Tech.pdf. So they intend to use air cored hence more turns for same inductance and with no core and more turns the leakage inductance will greater. I can envisage getting a non isolated supply to work (not easy and high efficiency difficult) but an isolated supply really challenging.

They claim the inductance needs to be at 100nH level, so it might be only 1-2 turns.

[wraper]

At these levels small, air-core inductors and transformers that are printed in-PCB and surface-mount capacitors are
used throughout much of the system. With no magnetic materials and batch assembly, this
translates directly into cost and manufacturing bene?ts.

So it seems that transformer will be PCB itself. Also read somewhere that there will be multiple channels in parallel.

[calexanian]

air core at vhf is an extremely difficult thing to tune. leakages all oner. not to mention dielectric heating.

[PChi]

The only viable power planar transformers that I have seen were far from cheap. The turns were assembled from multiple layers seperate from the PCB. I have seen attempts to use the PCB but am not convinced they are viable for a good few watts output. They need lots of copper and little insulator, not like a PCB were there is lots of Epoxy and glass fibre but only thin copper.

[SeanB]

You can do a fairly cheap planar transformer using flex PCB, though it will be a massive stack of mylar and a huge spot for a multipole connector to bind the layers together. I have seen some though that wind a flex on the bobbin as the one low voltage winding that has a high current, making the coil, leadouts, sense winding and interwinding shield as a single strip that is wound on in a single operation. That at least used a standard core.

[Dago]

The only viable power planar transformers that I have seen were far from cheap. The turns were assembled from multiple layers seperate from the PCB. I have seen attempts to use the PCB but am not convinced they are viable for a good few watts output. They need lots of copper and little insulator, not like a PCB were there is lots of Epoxy and glass fibre but only thin copper.

Planar transformers using the PCB are very common in the industry. They usually use a stack of PCBs as the turns, like this:

I have some multiple hundred watt PSUs using transformers like this.

[nidlaX]

Has anyone gotten their hands on one of these for a tear down / analysis? Arrow Electronics was going to do a write-up / video, but I haven't seen anything posted.

[riccardo.pittini]

I'm working in a similar business and I have seen some teardown.

The unit is fully potted, with a very hard potting. Not that easy to reverse engineer. But also from measurements on EMI/EMC, it turns out that it is a conventional power supply (no VHF), operating at variable frequency in the 170-300kHz range.

So my feeling is that they struggled too much with the vhf technology and they had to fold-back. Here a good design allowed to match the size constrains

[Richard Head]

The figure of merit in this game is power density. The power density here is 21.7W/ cu. inch. That power density is not that unusual today at all, and routinely done at 500Khz. In fact Eltec energy does higher than this at 3.2kW with active PFC at around 100Khz and with EN61000 compliance.

[rsjsouza]

I am not qualified to know what the shortcomings of such a design are, but off the top of my head I wonder about switching losses, as well as thinking "if it was just as simple as cranking up the frequency, it would have been done lots of times before".

I agree. Despite not AC/DC, there's a relatively new (April 2016) DC-DC Buck converter that does the conversion at 10MHz. Looking at its footprint it is very bizarre, but they claim a reasonable efficiency.

http://www.ti.com/product/TPS54A20

[f4eru]

I think a big potential space-saving trick has been missed for decades.
A significant space-hog in any SMPS is the input reservoir capacitor, however for a device that is used for charging a battery, there is no fundamental need to supply continuous DC.
If the charging circuitry was designed to accept a 100/120Hz pulsed supply, then it would be possible to use a PSU with minimal smoothing, and eliminate the capacitor.
It would also eliminate the need for a seperate power-factor correction converter.

yep. I did that in a commercial product.
80w psu for a product with a  motor. I used a typical pfc chip as a flyback controller. No electrolytic at the primary, the secondary electrolytic took all 100hz voltage ripple.
advantages: less  cost by avoiding pfc hw, prim cap. Emc is a little spread. Smaller design .
disadventages: ripple at the output, higher peak current through components, slow regulating response

[Marco]

Induction hobs do the same thing.

[f4eru]

And  also led psus, at least the good ones (avoiding chemical caps is vital for hot running led psus)