will this work

default wrote on 11/1/2017 12:20 PM:

Do you have an estimate for the current in the wire, the resistance of the wire? You seem to be very concerned about the voltage across the circuit, but not at all concerned about current used by the circuit. You can lower the sense voltage with voltage dividers that only drop a small portion of the voltage, say 10% and not sacrifice much in the way of gain.

Whether this circuit will work well depends on a lot of details. Sensing a small DC value with an opamp is not so easy as the opamp has an unpredictable DC offset (defined within a range though). With the small voltages you are likely trying to detect this error will be significant. You can use a multiplexer on the input to switch the same amplifier input between the two points being measured to generate an AC signal which can be amplified more easily and allow the opamp to be isolated from the DC level of the signal with capacitor coupling.

On the other hand, there are commercial devices designed for high side current sensing. They work much the way you describe but use techniques to deal with the problems you will find designing your own. Have you looked for any of those?

I'm guessing that the current will be in the order to 1-4 amps. This is for a stand-by super capacitor bank that runs hard drives, modem, speakers etc.

Sounds like the voltage divider set to 1-2 volts under 12V may be the way to do it. That also provides another benefit - I can incorporate a pot into one leg of the differential divider and adjust for offset and use it to set the tripping level.

If I were to get that complicated I could just as easily build an isolated DC/DC converter, use it for the op amp, and an optical coupler to run the relay side of things.

I ruled that out already - I want to build it with parts I already have laying around. A fuse would work too, but I like the idea of a (simple, fast acting) resettable circuit breaker and pair of LEDs to indicate the status.

There's a 4 amp PS constantly charging the caps and the caps can keep the modem alive for >1/2 hour in a power failure. (they can also start my truck, or a fire) The few milliamps the circuit and relay will consume isn't a concern.

I remember doing something like this in a commercial supply we designed for some point of sale cash registers in the 70's. We had multiple supplies and all used high side sensing with resistors we wound from phosphor bronze wire to get a repeatable current set point. The op-amps were 747 (two to a package) and one half ran the regulator and the other the current limiting, I think the power came from +/- 15 volt supplies which were part of the larger supply.

Sensing the drop on a wire "with essentially no voltage drop" seems risky to me. If you absolutely can't add a small series resistor I'd go with a Hall-sensor to measure the current. ...Jim Thompson

Even if you are only building one of these, opamp offset will drift with time and temperature. You need to have an idea of what voltage you will be sensing so you can estimate the significance of the offset of your opamp. You didn't complete that thought above.

Complicated? What I described is a chopper amp and is the common way of amplifying small DC levels. It only requires a small, 6 or 8 pin IC as a mux and a capacitor to block the DC. If you want DC to drive a circuit it only requires a diode and capacitor on the output. You think that is complicated? I suppose you do need a 555 timer to drive the mux.

Oh, then please provide a list of your spare parts.

You can also make a mux with a few resistors and a pair of transistors. You will need inverting drive for the two transistors.

Do you have phosphor bronze wire on hand?

As Jim T has said, a magnetic sensor makes a good match to the 'no voltage drop' requirement. Hall sensors are one possibility, magnetic sensitive relays another (with some electronics to latch it off). It's pretty easy to wind a few turns of heavy wire around a reed switch. That scales up nicely, there are 'protective relays' in power stations.

For really fine discrimination, a fluxgate magnetic sensor is good (if somewhat complex).

I'll give the reed relay thing a try. Hall effect is another good one, I think there's an old keyboard around that has Hall sensors in it.

Thanks

A room of my house knee deep in tech junk. Most of the stuff in small cardboard boxes with what project they went with written on the outside. Cataloging isn't going to happen... Only the discrete parts that I stock get cataloged.

No, that was back in the 70's. We needed some ultra low value resistors for current shunts. I want to use the same technique since it did work very well, but I don't remember how the power supplies were arranged. Offset and temperature drift were not significant. I don't think the client set a specification for the current limiting just that all outputs could tolerate an indefinite short circuit.

Let me know when you get that catalog together and I'll give you suggestions based on what you have on hand. Oh wait, you must know what you have on hand as you have already said you don't have any micro switches.

Ok, I can't help you if you only want to use what you have on hand and you don't know what that is. I don't see how you can get anything done this way.

You need quite a few turns around a reed relay unless there's a fair bit of current. back in the days of dial up modems, a magazine article was published to thwart rogue dialler software. The fraud drops your connection and dials a server on a premium rate number. The reed had about 200t of wire round it and both were in series carrying the line current. Dropping the line released the reed and you had to manually hold a bypass button to establish a new connection.

someone suggested a Hall effect current sensor, its simply an assembly that places a hall effect semiconductor device in fixed proximity to a wire you're passing the current through.

If its AC - a current transformer might be cheaper.

I found a reed switch that gave the specs as 30 turns of wire at one amp should actuate. I want it to trip at 4 amps so that should be less than 10 turns.

I'm still playing with op amps and haven't tried the reed or hall switches yet.

DC

Ok this looks like a winner to me (for it's simplicity and hopefully reliability).

I tried a bare reed switch I happened to have and wound 20 turns of 22 AWG magnet wire around it in two layers. It works at .992 amps. So I'll cut that to 5 turns of wire and try again.

I also have a number of reed relays around and plan to cut the fine wire off and use the bobbin and connection pins to pretty it up.

I'm going to use a reed since that seems to be simple, easy to calculate. 20 turns of 22 AWG magnet wire around it in two layers. worked at .992 amps. I can up the wire to 14 AWG and lower the turns. The diameter of the reed is small enough so that the wire length is only a few inches and will have very low resistance.

I'll latch the switching relay 'on' with a transistor and just use the reed to ground the base so I'll only be switching a 2 milliamp resistive load. A couple of pushbuttons for on/reset and off/tripped, some LEDs and it should be good to go.