Surge Current and Power in FETs

Obviously not. Why would I have needed to look?

Actually, all through. You could infer something from a sensor on it's surface, but you'd need to know a bit about it's inner structure to make a reliable inference.

It is important to keep time scales in mind. You can blow up a semiconductor laser in less than a microsecond, if you set your mind to it.

<snip>

Unlikely. They do all sorts of stuff before they get big.

And another killed the wife of a friend of mine. She started getting epileptic fits several years earlier, but this was in the days before brain-scanners. Eventually they opened up her skull to see what might be going on, then closed it up again rapidly. She died six days later. One of my father's mates had a similar kind of problem and lost his sense of smell early in the process

Odd. Brain disease would have been much more credible. But if his brain was getting screwed up he might not having thinking too clearly

Not always where it would do most good.

John Doe's judgement isn't all that good. The only sources he seems to trust are right-wing lunatics, or people touting for trade from right-wing lunatics

All that Phil is posting about is words that I've been posting in a user group. That's definitely a social construction. There's an identifiable human being behind it but Phil isn't interested in that and much prefers to witter on about some product of his own somewhat erratic imagination . <snip>

<snip>

A PTC thermistor doesn't start out hot or at a particularly high resistance.

Presumably you can estimate or measure it's thermal mass, and work out how long it will take to warm up enough to start limiting the current

<snip>

Not exactly.

<snip>

The idea of using a PTC is worse than the NTC. To protect the surge on connecting the battery the resistance has to be high from the start. The entire surge is over within less than a millisecond. The peak is right at the start of that time. By the time a PTC has cranked up resistance the peak surge has happened.

But even if it could be made to work, it requires more design effort still to assure it will not interfere with the surges of the motor start. I'm not sure it is even possible really. That depends on a lot of details and it seems far too messy an approach to worry with.

Even it it were true, not required. It's not hard to design in a resistor that won't be damaged in this situation.

I really think there is something seriously wrong with that guy. Maybe it's just the way he acts over the Internet, but if he is remotely like this in person he would likely have found himself in jail sooner or later. What a basket case.

Pity Phil can't actually work out how to land a punch.

That doesn't follow. Synchronous motors also generate more back emf as the rotor spins faster. That does tend to reduce the current flowing through the windings, whatever the switching arrangements.

Actually they don't. The relatively high initial resistance may resist current surges, but as they warm up there's less resistance to generate heat in the thermistor, so progressively less of the surge gets absorbed there. Because they dissipate less and less heat as they warm up, the amount of energy that ends up in them is limited, and they are less likely to get hot enough to sustain permanent damage, but they aren't absorbing all that much of the surge, which is free to damage everything else along the current path.

They protect themselves, rather than anything else. If they get enough energy to form a hot channel, that can end up with a very low resistance indeed. <snip>

Yes, absolute nonsense. But polyfuses have cold resistance, which can reduce a powerup surge a little.

This piqued my curiosity, so I did a patent search, finding EP0892410A2 (Improved thick film resistor assembly) assigned to Welwyn Components Ltd.

It's a very good survey of the field, and showing the evolution from porcelain-steel assemblies to glassivated anodized aluminum assemblies.

To obtain the patent, enter EP0892410A2 in the big search box of Google Advanced Patents screen.

Joe Gwinn

Of course he can't. That's because no-one here has a clue what the application actually is. And instead of explaining it (with a schematic and a clear explanation) you just get upset when possible solutions are offered which you hadn't considered yourself.

Mocking previous work isn't the same as making money and doesn't pay the bills intrinsically so I try not to engage in it, and it's hardly a job perk. When it comes to a competition between cash and ego cash tends to win every time. And here I was thinking engineers were supposed to be good at the capitalism.

This is the style I used:

but it was lower resistance in that case. They are thin and slightly cupped so hug the mounting surface when you bolt them down, and make good thermal contact. They really tolerate transient thermal abuse.

We made our own current shunts too. We bought sheet manganin, had it punched into exotic shapes, annealed it, and epoxied it to temperature-controlled anodized aluminum blocks. Our secret was a way to cancel eddy-current effects that created funny step response.

Our first NMR gradient amp was 70x better than what the customer had been using, which was a special constant-current Crown amp.

.....

Well if you're not sure what circuit is being discussed then you can be sure that no-one here does.

The biggest capacitor I ever used was about a quarter farad electrolytic.

I assume that the motor does not run immediately the battery is connected?

An NTC thermistor is a possible solution. Otherwise use a resistor which is shorted out by a FET as necessary. Or just a FET with slow turn on. Watch out for what happens when users, not simulators, use your product and switch it on and off in ways and circumstances you didn't expect or test for. Can they change the battery when AC power is on?

Thanks to advancements in software engineering we'll soon have a 100% Bitcoin-tied economy. Bitcoin can never be allowed to drop or the sun burns out. everyone dies. Kind of like that movie with the bus and Keanu Reeves I saw one time.

Since the part that needs to heat up is the same as the part that's conducting the current (and so dissipating the power), it'll heat up instantaneously. (I have data showing faster than 30 picoseconds response time for a very small bolometer.)

The issue is that you need to dump enough energy into its thermal mass to get it to switch.

Cooling down is a whole 'nuther story, because that requires heat conduction, which is slow. It does get quadratically faster as you shrink the distance over which the thermal diffusion occurs.

How about an inductor with a TVS in parallel? I do that a lot in small SMPSes to avoid de-Qing the input filter.

Cheers

Phil Hobbs

Dilbert is always about other people. ;)

Cheers

Phil Hobbs

But sometimes it's howlingly funny-bad.

That design looks good. The zigzags plus the mirror symmetry around the horizontal centerline should greatly reduce inductance.

Hmm. Low inductance implies low eddy-current effects.

Did Crown ever react?

Joe Gwinn

We were seeing about a 5% voltage overshoot on a current step, for maybe 100 us or so, from eddy currents in the aluminum.

Not that I know of. We sold Varian gradient amps for over 10 years, until Agilent bought Varian and killed the NMR and FTMS operations.

I think big superconducting magnets were bad economics.

You seem to be confused. I was referring to something posted some time ago. I believe I've mentioned in this thread originally I worked on a transistor based design but had trouble preventing current surges on switching between run and power down modes. "run" mode does not mean the motor is running, just that the entire system has power. Power down mode must retain power on the "Always_on_3.3V" power domain which is not shown in this schematic as it is on the controller board which is far from complete. It is regulated from V_main which is never turned off.

The motor driver is on this board, but controlled by the always on MCU.

No, it is controlled by the MCU and requires the user to turn it on thru multiple button presses.

Many things are possible. NTC looks like a PITA to find one that will do the job with margin. It all depends on the time constant. The 100 amp surge must not lower the resistance too quickly and the 10 amp motor start up must lower the surge quickly enough. Sounds dicey if even possible.

Yes, this has been discussed. Have you not been paying attention?

Again, this has been discussed.

Yeah, this isn't about users. This is about attaching the battery as part of service or initial fabrication.

No, the case has an interlock that prevents it like the old TV sets where the power cord is anchored to the rear panel.

The present power board design is not bad, but the designer just hand waved the surge prevention and has no data or analysis to support his claims that there is no problem.

You must live in a world where "instantaneously" meas something different from here. In this reality the time required for something to heat up is related to the the power input and the thermal mass, you know, like a car accelerating, the speed can not change instantaneously.

Better if it is already in the correct state limiting the initial surge.

Sorry, not sure how that works. The TVS limits the voltage across the coil which seems like that would minimize the benefit of the coil while still not allowing the rapid current increase to power the motor. I suppose the motor current build up is limited by that inductance, so you don't have to be faster than the bear, only faster than the guy with the hiking boots. Still, what does the TVS do other than reduce the effectiveness of the inductor?