Hall sensor inductance

Den tirsdag den 17. november 2015 kl. 17.59.28 UTC+1 skrev John Larkin:

this one mentions that the inductance will affect the propagation delay

I guess they assume that in the big picture it is otherwise negligent

-Lasse

I'm pretty sure that one doesn't have a core. And its prop delay is probably dominated by the internal electronics. The statement that prop delay is a function of primary loop inductance doesn't make sense, since prop delay is defined as Vout/Iin.

I'm looking at rates like 10 amps per microsecond, so every microhenry drops 10 volts. I have no idea if one of those big Hall dudes adds anything like 1 uH.

Den tirsdag den 17. november 2015 kl. 18.43.17 UTC+1 skrev John Larkin:

the functional block diagram shows a core and it pretty much needs one to get the field though the hall element and not be affected too much by external fields

the one you link to are rated for a few hundred kilohertz at most and several micro seconds of propagation delay

-Lasse

No, but I wonder if you could add a winding with multiple turns and servo the flux as measured by the hall effect device to zero, thus cancelling any inductance seen by your single turn primary. Obviously primary current is then some multiple of secondary (servo) current.

Cheers

Closed-loop Hall sensors do exactly that to cancel the flux in the core, at least in a limited bandwidth. But they are expensive and use a lot of power, so I want to use an open-loop part.

In my case, I'd need a couple of hundred ampere-turns, which could get messy.

If this is related to ongoing pulse projects, you might consider a plain old current transformer. They're good for infrequent pulses (low duty cycle) and when the maximum pulse duration is less than the "sag" time constant.

I have a one of the LEMs with the three single turns, I can measure it.

Tim

-- Seven Transistor Labs Electrical Engineering Consultation Website:

I need DC response, and low cost, and modest accuracy, so Hall looks good. Some of the 300-amp parts are like $15 in quantity.

The Danfysic DCCTs are fabulous, but huge and expensive, great for lab measurements but not production.

That would be great. But I ordered a Lem and a Tamura, which will be here Thursday, and I'll measure them.

Damn! Another one bites the dust.

GMR sensors are very easy to interface to, inexpensive, moderately accurate, and this app note may be of use...

Cheers

AFAIK: Hall effect sensors aren't particularly fast (maybe 200kHz or so) - they probably don't consider the inductance to be of much consequence at the highest frequency you might use.

Because the core is split, the inductance should be pretty small for one turn. Probably a couple uH at the most.

Jon

No experience, but the core characteristics should be pretty much determined by the dimensions of the slit -- you could try just calculating it.

I try to be an off-the-shelf guy as much as possible, but maybe you need to cook up your own?

I think you can reduce that by a factor of 10.

Most Halls I've worked with start to show signs of respone delays at 25Khz or more.

There could be some I have not yet touched, if so I would like some PN#s. because it would help me on a brainstorm idea I have.

Jamie

That's got to be a function of the signal conditioning -- the Hall effect should be a speed-o-light thing.

These guys are faster than Allegro at 400KHz but limited to +/-130A:

You might consider an open-loop sensor with bus bar. This is the cheapest solution, but takes some work to get it right:

The integrated amplifiers are chopper-stabilized to reduce offsets, this limits factor for the bandwidth. But eddy currents in the leadframe also limit the bandwidth. Also skin effect in the current conductor.

I haven't seen anyone spec the inductance, it would be interesting to see what it is. Normally the loads have much higher inductance, so it's not a concern.

Maybe not quite speed of light; there's conductivity of the sensor material, and a magnetic field diffusing in (so there will be a skin depth at any given frequency).

AFAIK, this is the case. Seems likely that it's a GBW thing -- the signal is pretty small (mV), so they'd be recovering a lot of noise and squirrely behavior if they tried for much more. There's also AC or chopper stabilization tricks in them, usually.

Current sense amps aren't usually very fast, either. Your average INA138 and related parts don't pass a MHz. It's probably for the best.

Tim

I got a LEM and a Tamura Hall sensor, both open-loop Halls, about 300 amps. I measured 1-turn inductance around 30 nH for the LEM, maybe 40 for the Tamura. That was a lot less than I expected.