Surge Current and Power in FETs

You are right to want to consider connection inrush - it can be a real problem. The way I have addressed this when it becomes a concern is to very simply have a series resistor right up front in the power chain. This resistor is shunted by an appropriately beefy FET that is turned on by normal operation.

piglet

Sounds like a nightmare. Things like that tend to never be done, partly because the good people leave. At some turnover level, the time to complete approaches infinite.

One of my customers is working to do their own design and cut us out. Two "well managed" teams in two countries have been at it for three years now. We're not too concerned.

LOL "We're open to a full re-design if the new hire deems it necessary" means "We know we need a full re-design" they're just not keen on putting a full-redesign fee on offer, up front.

You are right. I was thinking of something that would be bypassed with a control based on attaching the battery. This only needs a 1 ohm resistor which would be near invisible in all operation other than having the motor on. So a bypass based on the motor controls. Very simple, very effective. Great idea!

...

That reminds me of a power supply we bought in, many decades ago. Rather than a thermistor it had a surge limiter resistor which was short circuited a few seconds after power on by a semiconductor device (which may have been a triac but not sure why they used a triac). This was fine when the power supply was turned on and left on for minutes or longer, then turned off and left off for minutes or longer, then turned on and left on for minutes or longer. When under load and turned off for only a few seconds and then on again it failed to fire the triac and...

<snip>

FWIW, my SMPS tear downs inevitably reveal thermistors in use as inrush limiters.

Danke,

Cheers

Phil Hobbs

Jack up the price.

Cheers

Phil Hobbs

-----------------------------------

** Same issue with the Bose 1800 stereo power amplifier.

A 4.7 ohm, 10W WW resistor was bridged out by a SPDT relay about 100ms after switch on. The relay was energised via one of the 90V DC rails as it neared full voltage.

Worked fine *unless* you switched on the amp under load and drive or there was an outage during use. The high load current increased the AC voltage drop across the 4.7 ohm so much the relay could no longer close.

Result - one dead resistor and no sound until the amp was repaired. Major issue in a live sound system.

...... Phil

I did that on the AC line input of a big NMR gradient amp, a resistor and a triac. I actually used it as a pre-regulator too.

It eventually broke the resistor, a big 250 watt metal-case MIL type wirewound. Thermal fatigue.

A Welwyn porcelainized steel thick-film worked.

My concern with the thermistor idea is the time factor. There is already a poly fuse (either 10 amps or 16 amps, can't recall) that is specified for 0.7s even at 75 amps. Maybe that's not the same thing. But I don't have a lot of confidence in thermistors acting in the time frame required, sub millisecond or 1,000 times faster than this fuse.

I guess the idea is the thermistor starts high resistance and drops as the current flows, but again, how quickly will the resistance drop? In this case it needs to *stay* high long enough but get out of way once in use. This sounds like a delicate balance and hard to achieve. I don't have the option of building something to test on the bench under lots of conditions only to find the vendor is changed when in manufacturing.

So the direct control option seems the best. This could also be used potentially to mitigate the current surge on switching to battery when the power fails if wired into the power fail detection.

But first, I have to convince someone else this is even needed. It gets old trying to motivate people to take the extra step. Heck, I couldn't even get the power board designer to use a polarized connector for the battery connection because, "They can always wire wrong to the battery".

I'm fed up actually, but I feel I would be abandoning the project to leave now.

============================

<

** FFS you imbecile - explain the application properly !!!

Someone plugs in a battery and the DC supply must be perfect in under 1 ms ?????

What f****ng BS is this ???

NOTHING known will do that.

...... Phil

Wow. Easy boy! Calm down Trigger. I never said what you think I said. Maybe my wording wasn't perfect, but your response is always so over the top.

Sub millisecond is the time to act to protect against the surge. The Polyfuse is that way, it has to react to the over current and can't respond that fast.

If you bothered to read further before exploding, you would find that my concern is the thermistor would not see enough current to stay out of the way when *needing* to pass high currents.

I did a quick sim and 250 mohm will knock the surge enough to not kill the FETs if held for some 5 ms. If a thermistor has 250 mohm at say 30°C how much does it need to heat up to be below say 40 mohm? How quickly does it heat up and cool down? The motor is run intermittently and needs full power very quickly when turning on, cycles of 2-4 second ballpark. They are trying to reach full pressure in 100 ms which is not easy. Limiting the power of the motor for even 10 ms likely would impact this goal.

The easy solution is the anti-spark connector, VERY easy. More reliable since it can't be bypassed is to add a FET and resistor to the board. The thermistor seems to have a lot of thermal mass issues that make it complicated to make work. It has to respond quickly to the normal current but not so fast it lets the sub-millisecond surge through. That's a window of 1 to 10 ms. It's possible, but do they specify the thermal behavior that way? I'm thinking it would work fine to limit the surge, but not respond quickly enough to allow the full 10 amps through to the motor at the outset.

I'm thinking this issue is going to be the straw that breaks the camels back and I will give up trying to help. We'll see. I'm going to send the simulation to the board designer that shows a 250 milliohm resistor is required to bring the surge barely within limits. This same 250 milliohm resistance is too much to allow the motor to draw 10 amps without excessive droop, so it must be bypassed when running. Either that or use the connector that charges the caps when the battery is plugged in.

snipped-for-privacy@gmail.com wrote: ==============================

** You f****ng well did. See above. ** You wrote what you were thinking.

** FFS imbecile - there is no delayed action with an NTC thermistor. ======================================================== ** I read your tenth witted tripe far enough - f*****ad. ** You said it was 4 amp load 9 ( rms ? ) I posted a possible solution ( small 1 ohm NTC ) which you rudely ignored .

FYI Shithead

You don't write clearly because you don't think clearly.

Fuckwit Tesla owner

....... Phil

They call a computer program "code" for a reason.

It's already obscene, partly because they buy a few at a time and won't commit to quantity purchase orders. The irony is extra thick here.

I'll tell you the story one of these days. You know the folk. Big things are happening, or even better maybe not happening.

It's a PTC thermistor, not NTC, so it starts off at a low resistance, chosen so that it doesn't warm up much in regular operation.

In-rush currents are much bigger and heating goes as the square of the current, so you can end up dumping a lot of heat into a relatively small volume of thermistor, whose resistance rise enough to limit the in-rush before it can trigger a current limit or blow a fuse.

The circuits starts up more slowly at this lower current, but once it has charged all it's storage capacitors the current decreases and with it the temperature of the PTC, so it's resistance falls.

I've never designed one in but the theory gets spelled out from time to time.

Find the right part, and the manufacturer's data is likely to be a lot more helpful.

The page itself isn't all that helpful, but it does offer links to potentially helpful application notes.

<snip>

I've never seen one used that way, but I haven't seen a PTC inrush-current-limiter either. Just heard about them. Odd that you haven't.

Not what TDK seems to think.