Current Level Detection

Anyone ever used a SENSEFET for this? I've no experience of them.

If I correctly recall a previous discussion of them a few years back, they simply bring out a fraction of the source current (like 1/20th) from a separate pin, so you can use a lower-current sense resistor of a higher value to get the same sense voltage drop.

Have I remembered incorrectly?

CH

I worked with Colin Hunter at Ruston Electronics also in Luton back in the early nineties when I did a couple of contract design and software projects for them.

I remember him as a talented engineer if a tad irascible. Being a free spirit he enjoyed tying management in knots, An interesting bloke!

I never found him irascible. If he didn't like something he said so, but he'd always say why he didn't like it, and the reasons always struck me as sensible.

I think he might have gone further in a less class-conscious society, where he would have got a university education, and management would have taken him more seriously - they took him pretty seriously anyway, but he didn't have a Ph.D. and lots of people at Kent Instruments at the time did, including the technical director - Bob May, who wasn't all that clever. Most of the Ph.D.s bailed out at much the same time as I did - there was a wage freeze, and if you had a Ph.D. it was fairly easy to get more money somewhere else. I moved to EMI Central Research, which definitely raised my salary a lot, but I think that was something of a one-off.

I have found over the years that there is no strong correlation between academic acheivement and ability to to the job. I can think of at least two clever and creative engineers I worked with who had no engineering qualifications whatsoever. They were better at "thinking outside the box" than other conventionally educated engineers.

Yes, it's true that our class riddled society has historically and continues to hold back talented people while encouraging mouthy morons into positions of power. Johnson being a case in point.

An ideal application for a toroidal saturable reactor. There is no upper current limit, you design for the low threshold. And galvanic isolation as a bonus. Power it from an oscillator of your favourite breed, load with a resistor at the other end and measure the voltage of this L(I)R divider. A 4x NAND/NOR TTL or CMOS dinosaur should suffice for most of the required functionality with a bit of thinking, a no-brainer with two of them.

Best regards, Piotr

But why the hall, if a piece of wire will have exactly the same effect, provided the sampling current is of conveniently high frequency? Variable reactance elements are such a powerful concept, are super-reliable and can be very cheap. With a bit of research off-the-shelf components would do just fine -- one can easily pass a thick wire through the hole in an existing toroidal choke.

Here with the frequency, number of turns or the load resistor.

Best regards, Piotr

Indeed.

No, it will not. You just need a pulse generator and an AM-like voltage detector and a comparator. A 74HC132 would be enough.

Best regards, Piotr

A Neel effect sensor would be another option. But Mr Doom just wants an indicator, there is no reason to make it more complex than necessary.

Best regards, Piotr

Sounds moderately complex!

What is simple then? :)

Best regards, Piotr

Just use a high-mu toroidal choke, these have quite low B_sat. The inductor wound on an r=5mm F938 core loses its inductance (quite literally) in the 700mA-turn range. 3 turns of the current-carrying conductor will be perfectly enough for a simplistic detector, one would suffice if the value of the R in series can be adjusted.

Gentlemen, I have been using this very technique in a 50+A synchronous rectifier control circuit and in a <2uW isolated switch state sensing application. It does scale and is extremely robust. A saturable reactor is a way to go here.

Best regards, Piotr

One shunt resistor and a comparator.

Now you have to connect the battery post through that shunt resistor, to the car, and power the comparator, and bias the comparator with other resistors and some kind of reference, and observe the output somehow.

The toroid is a no-connection option, and doesn't require your junk box to have a 15A-capable shunt resistor handy. One resistor and a comparator, handed to you, wouldn't do the job; simple, yes; complete as a solution, no.

My Ph.D. is in physical chemistry, not electronics, so - to some extent - I'm an example. Getting well taught should put lots of techniques at your finger-tips, but my experience was that academically trained junior engineers preferred to give me their lecture notes and let me exploit the technique rather than call on what they'd been taught, which rather lined up with mu experience of chemistry, where the lectures served more as instruction about what could be done if you spent time in the library rather than anything immediately useful.

outside the box" than other conventionally educated engineers.

There's talent and there's training.

Boris Johnson isn't a moron, any more than Trump is. One of our friends served as his nanny when she was an undergraduate at Oxford, and tells us that he was bright kid, if rather fond of looking at the world from his own point of view. What you are complaining about is an egocentric way of dealing with the world rather than stupidity as such - though it does lead to some pretty stupid choices.

Draw up the toroid schematic and we'll discuss it.

OK, but first you draw up the battery post clamp, and the dummy battery post for the car's battery cable, with suitable insulation so it doesn't short to the chassis during testing. A mechanical drawing, of course, not just a schematic.

The pulser and detector is up to you. I would suggest an MCU, which would use two GPIO pins and nothing beyond the components already drawn.

Here is a full analog application of this concept:

Best regards, Piotr

Here is another one, same principle + synchronous sampling. More components due to the extreme low power requirements. Not sure if it is more analog or digital, hybrids work best.

R7 should be read as "tested up to an absurdly high internal switch resistance of 68 ohms"

Best regards, Piotr