1. Home
  2. Electronics Design
  3. Sensing current on isolated feed

Sensing current on isolated feed

Hi,

I have a lot of field components behind isolated supplies. Invariably, I need to look at a signal representative[1] of the actual load. In each individual case, I've been coming up with "cheap kludges" to get the *minimal* job done (to keep manufacturing costs really low).

But, solving the same problem many times with slightly different criteria/constraints is getting annoying.

So, I'm looking for a "universal" approach that will scale readily (at some cost penalty over my cheap and dirty approaches).

Power to the load(s) -- there may be many attached to a single control point -- is sourced at the controller. Power to the controller is sourced "remotely". Loads are a few electrical feet away or

100+ feet distant. Loads may be subjected to nasty environments (RF, lightning, etc.) And, of course, shorts to other loads, ground, etc. "Sensing" must happen at the controller (loads may be in inhospitable places).

I'd prefer high side switching of loads but that's not imperative. Everything is DC -- 12, 24 and 48V. Loads range from ~100mA to

3-5A (I'm willing to change stuffing options for different load types). In some cases, I want to just look at DC characteristics; in others, AC (particularly on/off related transients).

Pointers to suitable circuit topologies?

Thx,

--don

[1] In some cases, this can be a really crude indication -- e.g., zero load, nominal load, high load/short circuit. In other cases, I really want to be able to *watch* the load to better characterize what the controlled mechanism/device is doing.
Reply to
Don Y
Loading thread data ...

For isolated DC current sensing, use HALL Current Sensors.. They make PCB types etc...

P.S. They aren't cheap.

Jamie

Reply to
M Philbrook

Consider magnetoresistive sensors. This article includes description of them among other methods:

formatting link 

--
To reply directly remove sj. from the domain name. This is a spam jammer.
Reply to
Stephen

For that range of voltages (12-48V) and currents (0.1 - 5A) a suitable shunt resistor and opamp into V-F converter could all thats needed. Choose shunt resistor, amplifier scaling resistors and power supply voltage dropping options at board stuffing time.

V-F in audio range is nice as it gives so many isolation options: opto, transformer, RF, telecoms etc. In emergencies your ears listening to the tones makes a handy gut-feel, low tech diagnostic tool.

piglet

Reply to
piglet

P.S. They aren't cheap.

Not at all -- little SOIC ones are under $5, for reasonable accuracy.

The industrial ones run a modest $20 (LEM, etc.), not really having better ratings, though somewhat more bandwidth, and higher isolation. These modules can get pretty fancy; VAC has some that apparently work on differential saturation, so are very accurate.

For crude sensing, I'd think about something like a self-powered opto. Maybe use some shunt diodes and resistors to set a limited voltage drop, which lights the opto. Assuming you can tolerate the ~2V drop. If not, a self-powered op-amp to do a more finessed measurement of same, perhaps. This gets increasingly complicated if you want high voltages as well as isolation (e.g., mains/rectified sensing).

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com
Reply to
Tim Williams

Years ago I experimented as a kid, taking two single sided PC bords, each having a conducting trace from on end to the other.

These two little boards were electrially joined at each end, with the boards bonded together via the traces looking at each other, but not touching in the middle.

On the other side of the boards I put a strip of graphite as resistor paths that became part of a bridge.

When passing current through this device the boards flexed and thus also changed the values of the graphite strips.

It seem to work ok for what I was using it for, but over time you'd need to zero, most likely due to the PC boards not holding the shape or what ever.

Jamie

Reply to
M Philbrook

I used this approach on some of the lower power uses. Note that you also have to worry that the load will be shorted, etc. (so, shunt resistors can become smoke emitters in short order) This limits its use (without other short circuit protection mechanisms).

The VFC only works if you are looking at reasonably static (DC) load characteristics. High frequency transients get lost in the converter.

Ideally, I'm looking to combine the requirements into an integrated solution. E.g., use characteristics of the short circuit protection (current limiter) to deduce the instantaneous load, etc.

Reply to
Don Y

Wow! That's pretty cute! But, way too pricey (even at $4, Q100) for my application. (I need a *lot* of them!) And, "all" it does is sense load current -- nothing to *limit* that current.

Reply to
Don Y

It's still pricey. And, undoubtedly overkill for my needs. I'm more concerned with repeatability (e.g., that a sensor yields similar results for similar conditions *and* that sensor A behaves similarly to sensor B).

E.g., I want to be able to see if an inductive load is now

*resistive*; if a powered DC motor is now *stalled*, etc.

I can also "learn" from past observations -- assuming the circuitry doesn't drift. So, if I "know" that a load is behaving properly at some time, I can use the observations from that time to later determine that it is NOT behaving properly.

I was thinking more along the lines of a closed loop around a FET/BJT that effectively sensed load current by how "hard" it had to drive the "switch". And, at the same time, could be designed to refuse to drive it *too* hard (i.e., affording short circuit protection).

Knowing that the driver *should* be on yet appears NOT to be on (no load current) would be used to infer the driver had gone into "shutdown" due to a fault. (Alternatively, it could stay on and just limit the current to that max level which could then be sensed and the "fault" inferred -- though the switch would need to be able to sustain that condition indefinitely)

Reply to
Don Y

ter

,

, a

intelligent switches? fpf2700 auir3315

-Lasse

Reply to
Lasse Langwadt Christensen

Have you considered smart DC/DC converter modules? Vicor makes regulated, isolated (or not) switchmode converters with logic shutoffs, that have a PR pin for paralleling: the signal on that PR pin is a good indication of the current drawn.

Reply to
whit3rd

Ooooo! These look yummy! I'll have to see how I can "watch" load current using them. Thanks!

Reply to
Don Y

Allegro has several Hall sensors, some they're trying to sell me for

*way* under a buck. Too noisy, though. I may look at their GMR sensors later this year. Meanwhile, five resistors and an opamp are good enough.
Reply to
krw

I did essentially that (but with a DC/DC that I built from discretes) for the irrigation controller: monitor current on the primary to deduce the load that the secondary sees. But, the data it produces are relatively "low fidelity" and you have to track previous loads to infer what *new* loads are doing.

Putting one converter per "drive" would probably be too pricey...

Reply to
Don Y

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.