PWM shunt regulator

What made you choose the R6 gate resistor value, guessing edge rate control?

Here that would trigger lots of extra regulations requiring all sorts of anti-islanding algorithms (AS4777.3), lock-out switches, regular anti-islanding testing with certificates, permission in advance from the government and separately from the electricity distributor, and consequently lots of expense and delay. Nominally it's to prevent linesmen being electrocuted when they think they have turned things off, but more plausibly it was to frustrate renewable energy installations in the vain hope of prolonging dependence on fossil fuels.

The board is plugged into a 48V buss, powered by a kilowatt MeanWell power supply. I can't back-drive that.

And besides, it's only about 20 watts that I have to dump.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Gate resistance reduces switching spikes. I have two sections of a 1K resistor quad pack available. This thing will switch around 2-3 KHz so switching losses will be small even with slow edges.

Spice says drain rise and fall are around one microsecond and it'll dissipate 50 milliwatts in the fet.

So why not go to 1K in the gate? That only adds 10 mW.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

People do sell dummy load boxes that feed back into the line, but they are multi-kilowatt things that you install with a forklift.

I'm less ambitious.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

It's an 8-channel programmable dummy load module that can simulate a resistor in series with an inductor. It usually absorbs power and can do that for years, and occasionally returns power to the user.

Actually, I don't know much about lithium batteries except that they can catch fire.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

More technically, they can go into thermal runaway - they have to get above 120C before they can do it, so any sensible battery condition monitoring scheme can warn the user long before there much of a risk.

Even more technically, all batteries self-discharge, and they do it more when they get older and more when they get hotter.If your battery condition monitor keeps track of the self-discharge rate - which it can do easily enough in periods when the batter is neither being charged or discharged - and less accurately when it is active - you can keep track of the risk that it will go into thermal runaway.

This has been spelled out here from time to time, but its one of those problems John Larkin seems to file in the "too hard" basket.

If only everyone would listen to you, there would be no more lithium battery fires.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Excellent exposition on the law's obfuscated intent to foster a fossil fuel status quo beneficial to the trust babies who inherited the cash flow from their daddy's old school energy monopoly. A little later in this thread John discloses how a "wall wart" (or whatever) is used to provide 48VDC. John mostly adheres to the Anglo admonition against custom circuits connected to the grid power. It's a "here there be dragons" mindset. So, naturally, only China can create cheap dropper power supplies for Christmas lights found in many USA homes. It's simply above the paygrade of a Harvard student. </rant>

John's circuit is competent enough for me. Thank you for sharing, John. And there's nothing wrong with using old ideas in new products.

Ok, I admit to having high albedo. It's not my fault.

My power supply is a MeanWell 1KW switcher. It comes with all the pretty UL and CE and CSA stickers, so we don't have to re-do all that stuff in a test lab. And it costs a fraction of what we could build it for.

Warts have the same benefits.

Harvard students are artsy and business types, or the occasional coder.

The roughly 2 KHz triangle on the power rail does get into our output a little, but the product is a dummy load so few people will notice. The trick here is to use the biggest polymer aluminum cap that I can fit on the board. Looks like 150 or 180uF 63 volts.

I simulated a linear shunt reg version, but it's a big deal.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

How does the sensor monitor every part of the battery? A small defect in a separator could easily heat up to 120C before the temperature of the mass of the cell rose noticeably.

Warning the user isn't much good, the battery technology needs to be fail-safe not impending-fail-evident to the user.

A microscopic dendrite defect can start an ignition that propagates in all directions at centimers per second.

I once worked with a fire alarm company in Freehold, New Jersey.

They had a sign on the wall

IN CASE OF FIRE RUN, YELL "FIRE"

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

I once applied for a job at a company that

I once applied for a job at a company that made fire extinguishers for putting out explosions. This involved making a bigger explosion to propel the chemicals that inhibited combustion. Their products were used in the fuel tanks of various military vehicles and some aircraft. I was put off accepting the post when I discovered how the vacancy had arisen. John

If people paid more attention to what they were doing, there might be fewer lithium battery fire. Listening should be an active process, and if you paid enough attention to what was being said, as opposed to monitoring it for flattering and getting cross when you didn't find any, you might develop a better understanding of what was actually going on.

A single temperature sensor in the middle of battery monitors for increased heat generation in every part of the battery. The components have a much higher thermal conductivity than the outside air.

I've argued for a tetrahedral distribution of sensors around the central core. They could detect an area of localised heating. the idea didn't go down well.

Fail safe would involve a big resistor into which you could start discharging the battery if you detected worrying warming. You'd have to design the system to cope with that, and it would make the designers job more difficult.

Bad designers make their lives easier by ignoring that kind of idea.

The dendrite has to grow before it can initiate an ignition.

You've got to abuse a battery for quite a while before you can get it to ignition, and presumably the area around the dentrite will be warm for quite a while before it gets hot enough to initiate thermal runaway.

Hot spots in batteries warm up the whole battery. The component cells have a much higher thermal conductivity than the outside air.

People who design fire alarms are more aware than most that they can fail.

Triangular waves have an appreciable higher harmonic content. A Baxandall inverter has less, and there a Cuk-type variant with extra windings on the feed inductor that can have even less. Bit above your pay grade, but possible.

Getting transistors to dump lots of heat is a pain.

No Why?

Valves are much better at it.