current sensor circuit

Split-drain mosfets (or multiple-collector transistors) can sample a fraction of the main current. I don't know of any discrete versions of that.

A small resistor in the pfet source works, but needs an IC to shift the signal down to ground.

fredag den 2. april 2021 kl. 19.34.18 UTC+2 skrev snipped-for-privacy@highlandsniptechnology.com:

like this?

Thank you, actually I am interested in ASIC (so, to have a MOS split is very easy) but it is interesting to know there are discrete version. I was not aware of that.

However to be more precise, once you have a split MOS sensing a part of a current my question is what to do with. I remeber a circuit using this current as a part of a Vgs loop to have an image of the MOS current, I don't remeber it precisely, I am trying to remember it.

Some time ago, I worked on a semiconductor company doing this stuff and I stumbled on this circuit different times. I had it clear in mind 10 years ago, but now it is gone. But it was very common. Perhaps somebody knows it or the concept behind..., it should be standard but I cannot find information on the web. Perhaps also a resistor was involved. I don't remeber exactly

Thank you anyway for the help.

fredag den 2. april 2021 kl. 20.43.07 UTC+2 skrev maitre Aliboron:

Obsolete from Nexperia, now a Rochester part.

What's with Rochester?

Il 02/04/2021 21:59, Lasse Langwadt Christensen ha scritto:

Yes, the circuit was something like that... (but for PMOS).

If somebody has also tips, suggestions... on current sensing directly on the inductor I would apprecciate.

Thank you.

Well, ON Semi used to make SenseFETs that worked like that, except with split sources. All obsolete apparently, and IIRC they were all NFETs.

Cheers

Phil Hobbs

They seem to make a business out of parts that have hit EOL from the original manufacturer. As I understand it, they'll buy up and store and distribute the remaining parts stocks. I believe they'll also buy completed wafers and unpackaged dice, design rights, masks, and so forth, and can manufacture new runs of otherwise-obsolete parts if there's a sufficient demand.

I've tried that in a switcher but the noise was overwhelming. By the time it was filtered, the loop dynamics was trashed.

A resistor is better, downstream somewhere that's quieter.

They seem to have got into the lifecycle management business in the last year or two--used to be, you had to email them for a quote, but now they're appearing as "marketplace" parts on the major distributors' sites.

With the current panic-buying of semiconductors, they're an important resource. We use AD8606ARMZ dual op amps to drive moderate-speed ADCs, because their output stages have local feedback that prevents the ADC kickout from driving them nuts. Three weeks ago the distys had hundreds of thousands of them, then boom!--they were gone. There were SO8s but no MSOPs.

Rochester had some of the non-ROHS (ARM) versions in stock, so all's well for now!

Cheers

Phil Hobbs

I don't know yet if, for this design, I have the right to use an external resistor (otherwise, I know there are tricks to use the parasitic ESR of the inductance for current probing, requiring no additional external components - If you have info by the way you are welcome).

Thank you anyway for the tip, it's something worth accounting.

Isolation is a way getting rid of a lot of interference, and Hall effect current sensors sense the magnetic field generated by the current you want to measure.

They end to be rather less than wonderfully accurate, but might be good enough for what you want to do.

Is someone else making rules?

The voltage drop across a saturated swicth can be used as a current sensor.

This was an attempt at a super-wideband switching current source. It didn't work very well. Better was a voltage switcher ahead of a linear final stage.

Un bel giorno maitre Aliboron digitò:

Using the ESR of an inductor to measure current looks very inaccurate. Inductor windings are usually made with common copper wire, and therefore there is a significant dependence on temperature (3900 ppm/K). Proper current-sensing resistors tempcos usually stay between 10 and 50 ppm/K.

Il 03/04/2021 18:10, dalai lamah ha scritto:

I agree about inaccuracy but there are some expedients to perform this operation in a smarter way. I don't remember details (that's why I am asking). Typical L ESR current probing makes use makes of a capacitor and a resistor in parallel to the tank inductor, as far as I can remember. About noise there should exist some tricks such as blanking the comparator signal after the switch to make the comparator "blind" during current overshoots due to discontinuity. As I told before I do not know yet if I have the right to use an external resistor for this design. I am talking about solutions suitable for ASIC design.

However, if possible, I prefer to probe from Hi-Side PMOS. I know from former colleagues' experience that this methods works fine, but I cannot remember the exact circuit arrangement.

IEEE reference articles or direct experience about the matter (both methods) are welcome and appreciated.

the inductor is not going to announce its current with a voltage. because it's always connected to a voltage source. (unless it's ringing)

maybe by adding a winding you could detect core saturation.

Super-wideband = super-noisy?