AC-line supplies - under 1% loss

They could have used four STY105NM50N, for $80 total, and spent more on the inductor. The two serving as 60Hz active rectifiers are driven by 5k gate resistors, turn-on time constant Rg Ciss = 5k 10nF = 50us.

Forgetting the PFC stage for the moment, it'd be easy to make a 99.9% efficient 1.2 kW active full-bridge rectifier, i.e., with loss < 0.1%.

Active rectifiers for mains bridge replacement is easy

What is hard is to construct one that handles 4kV surge and burst and keeping a price point not orders of magnitude higher than a standard bridge

Cheers

Klaus

A standard bridge is dirt cheap, so a few orders of magnitude higher isn't so bad. I have no experience with MOSFETs as bridges, but the PFC guys don't seem worried by it. It's not very hard to protect the gates, and these big MOSFETs can handle a lot of energy while in avalanche.

I've studied avalanche breakdown in many types of silicon devices, read papers, and have made many measurements of MOSFET avalanche breakdown, and would like to assert that it's not intrinsically more vulnerable than a rectifier diode.

Anyway, the central idea here is that if the HV power-transmission guys can keep their losses well under 1%, right up to our property boundary, why can't we try to do the same in our consumption?

On the subject of cost, the 19 m-ohm STY105NM50N is the best 500V part in my table, and is much better than needed, even for 1.6kW. Alternates might be the 28 m-ohm STY100NM60N, which Newark has on sale for $7, or the 65 m-ohm FCH072N60F, which is going for as little as $2.50. That one would create about 0.06% loss at 1.6kW, or about 14x lower than a bridge rectifier. The $7 part costs $28, but would save $70 on electricity, after 15 years of service with a 20% duty cycle.

I wonder what Tesla does? With high efficiencies in the motor and battery, I'd hate to think they are tossing much in poor efficiency designs.

not many details but still interesting, 97% @ 25kW

Not so bad with transistors that big, I would guess. You can keep it switched on during the surge -- dumping the surge current into the filter cap (perhaps use a nice PN diode to bypass from PFC input to bulk cap, to handle both inrush and surge).

Tim

"Tim Williams" wrote in news:qleh4d$u15$ snipped-for-privacy@dont-email.me:

HV diodes are often a number of PN junction elements in series.

Not going to get the efficiency number, but not needed anyway to that degree.

To get the surge protection, series together diodes. Yes the junction potential also sums.

That is relative. If one is only trying to rectify ten volts, the losses would be greater than if one were trying to do 100 volts.

The same reason we use HV for distribution. The losses are proportionately less.

It's not so much about energy, but dv/dt over drain source during burst. EN61000-4 test define 2kV/5ns, which is far higher than most MOSFET dv/dt can handle.

Cheers

Klaus

If you switch it on during the surge, how do you do than with confidence in sub us. Surge is up to 4kV at 1.2us

And when do you turn off? If you do not turn off in time, you connect the cap to the line and bad things happen

Cheers

Klaus

Page 11

On a sunny day (Fri, 13 Sep 2019 02:01:00 -0700 (PDT)) it happened snipped-for-privacy@gmail.com wrote in :

I would not expact 5nS to pass the mains filter L C. That is 200 MHz.

At 200MHz and saturated current, the CM inductor is nothing but a capacitor. But you are somewhat right, might not be as bad as I wrote, but something to dig deeper into

Cheers

Klaus

As stated, it can handle avalanche.

TI uses a 50us filter for turnon, but 10ns for turn off.

Where'd you get that number?

I've heard of transmission system losses nearing 20% (2009). Perhaps that was a worst-case scenario for places that don't generate much of their own.

RL

A number like that isn't surprising, for the entire system input-to-output. It's individual elements, xfmr, etc., that are under 1%, have to be I guess, to allow a bit more for long transmission lines and still keep the total loss to under 20%.

More important is that the pulse is tall enough to saturate CMCs. It won't be saturated due to mains current, because that's still balanced.

Diff mode, the 50 ohm transient just bounces off the input X1 cap. What's left isn't all that much, and the CMC will not be saturated in leakage mode (for obvious reasons).

Tim

Fast switching is half of all we do..?

No need for fast switching anyway, the body diodes are already pointing in the right direction.

Unless you read that as, short out mains during a surge, which would be interesting, and risky, but the transistors could be dimensioned to handle it. 2.5kV /diff/ surge at 2 ohms is 1.25kA. We're talking big transistors already, so this isn't too too ludicrous, at least in the kW range.

Also the possibility to use 1800V SiC FETs to ride through it, using MOVs to clamp the brunt of it.

CM surge doesn't really matter, which is nice.

Tim