What's the hi-side current sense chip du jour?

"Joerg" schreef in bericht news:GKYRl.4517$ snipped-for-privacy@flpi145.ffdc.sbc.com...

Seen Linears LTC6100? To pricy maybe?

petrus bitbyter

That's similar to the INA193. LTC mentions just under $2 and I like LTC. However, neither Digikey nor Mouser carries it and that's usually a sign that a part hasn't made it into the mainstream. Kind of a litmus test for me ;-)

"Joerg"

Beware of problem below 20mV input voltage. Check the latest datasheet?

AD8211, AD8213 and similar AD parts. Available from digikey, for example.

A resistor in series with the - input should add some protection. Or make your own with a r-r opamp and a transistor, 40 cents maybe.

Or make a simple current-mirror-ish sort of thing, or a diffamp, with a couple of BCX71's and a couple of resistors...10 cents and nearly indestructable.

One trick, not super accurate, is to sense the DC voltage drop across the inductor and eliminate the shunt. Then the signal conditioning problem gets much, much worse.

John

The LM3485 senses the drop across the mosfet (when on!).

Even less accurate :)

Or you could sense the free wheeling diode for even worse performance :-)

This is getting worse and worse. You guys are great.

Flying capacitor, driven by the switch frequency?

IR temperature sensor aiming at the inductor?

John

Wa-a-a-ay back in the late '70's, current transformer, secondary side had a cycle-by-cycle DC restorer, no *WHADDABANG* failures, EVER ;-)

I used it in all the off-line switchers for GenRad Portable Products Division (nee Omnicomp) power supplies.

...Jim Thompson

1. Smoke detector, aimed at the load?
2. Microphone, " .
3. Light meter, " .

Cheers, James Arthur

One technique that gets used a lot in signal processing:

Multiply by zero and add what you think it should be.

Cheers

Phil Hobbs

Why not roll your own from a dual PNP? Digikey has several jellybeans:

Getting fancier, these are matched:

And slightly specialized, a bona fide mirror:

Cheers, James Arthur

Shunt + SSR chopper feeding a cheap ISDN transformer, with a bit of step-up for free. Somebody makes a VME thermocouple input module that works sort of like that.

You could counter-wind the inductor: add a second skinny winding that subtracts the AC component and feeds you clean DC, namely the voltage drop across the main current-carrying winding.

Digitize the AC voltage across the inductor and do a DSP thing to detect current based on saturation curves.

Measure the buck duty cycle and the input and load voltages and estimate current.

That last one is not totally insane.

John

Or the more general case: Skinner's Constant (aka Flannagan's Finagling Factor)...

"That quantity which, when multiplied by, divided by, added to, or subtracted from the answer you get, gives you the answer you should have gotten." --"Murphy's Law," Bloch

Grins, James Arthur

let Sloman design it, and it won't work at all.

Try Linear's direct ordering.

Sloman has little to say about electronics, and then nothing original or helpful. Ditto "Cranium." They are here to deliver insults, so prefer subjects they can't be called on.

What a choice in life.

John

Several of the Linear parts do measure the current into the inductor.

On some of the "roll your own" designs I have done, the current out of the inductor was measured. Basically it was the LM339. On section sensed the current going above the desired and yanked the charge out of a capacitor. Once the current fell below that value, the capacitor would charge back up and the power section switched back on when the voltage hit some magic value.

It took quite a while for the comparator chain and the power device to switch so the current always went above the set point a little.

I am sure that'll work but for client circuits I really prefer something that the mainstream distributors carry. If they don't then it might be or become a boutique part.

They really run up an error budget there but I couldn't see any nasties such as wild glitches. Or is there other stuff?

I am not planning to go that low, mostly I want to current-limit a buck when the sense resistor reaches 50mV or so, in order not to have to dissipate as much as a discrete transistor solution does.

The 8213 is too pricey but the 8211 looks very interesting, around a buck, thanks. It wants +5V supply but I can make one.