Current Measurement robust against massive overcurrent

[swarfrat]

I'm thinking about throwing a teensy on my car battery - I have something which appears to be draining the battery, but it's not always on. (Charging, Starter, Battery all check out fine). So I want to monitor battery voltage and charge/discharge current while off in addition to while the car is running.

Aside from the actual process of diagnosing this in a vehicle - it got me thinking. How DO you accurately measure the 0-25A or so range without frying or impeding perfect valid loads of several hundred amps? I could put something scaled to the normal expected loads after the starter taps into the main 12V line, but that won't catch anything which might be pulling current from the starter line at other times. Assuming the starter can pull 600A or so (just going from the battery rating), that pretty much rules out any sort of resistive current sensor. I can't really think of anything that sensitive that could actually withstand (and not impede) the normal operation of the starter.

Right now, my thinking is put my current monitor after the starter tap, and just use the battery voltage as an indication of substantial load on the starter circuit. Given the cabling of the the starter system - I'm having trouble imagining anything unplanned on that circuit only sucking down a couple amps. If it shorts it's gonna smoke.

 

[Xenoamor]

What you're asking for is a hall effect sensor. With such high currents you want to be galvanically isolated (no direct connections between the IC and battery)



I won't get into the physics but I'd strongly recommend reading up on how the Hall Effect works

 

[Jp3141]

Do you know the rough magnitude of the draining current ? You should be able to measure it with the car off:

But be careful just using a DMM -- many of the loads in a car have a large capacitor at their input (50-100 uF) and if you just disconnect the battery, insert a current meter and reconnect, it is possible that the inrush current will blow the fuse in your meter.
So -- if you use a meter, disconnect the battery, insert the meter configured for current measurement, short the meter inputs together and reconnect the battery remove the short and measure current. Be careful that you don't open a door or something as the current that the lights take could also blow the fuse meter (if you have a ~ 300 mA range).

Instead, use one of these -- https://moosh.im -- they can measure down to << 1 mV -- just connect it across one of the longer wires from the battery. The wire resistance will give you an (uncalibrated) current sense reading.

 

[swarfrat]

Yes, I'm familiar with the Hall Effect, and did look at a couple Hall effect sensors I was familiar with from past projects - but nothing I was familiar with would survive hundreds of amps while measuring a couple amps. Thinking some more about the problem - I'm thinking ignoring the starter tap is perfectly valid. I'll still have a battery terminal voltage reading that would indicate if something I don't have an ammeter on sucks the battery down. What I really want to measure while running is charging current, and draining current while it's off. Those are both rather modest, and an ACS 712 can measure 25A and survive 125A, and provides galvanic isolation (which I'm losing by also reading battery voltage.) Headlights should be less than 17A or so - charging currents should be 5-10A or so (at least for good battery life), so I think a normal range 712 is about perfect. It also has the added bonus of being available in a cheap breakout board.

I don't know whats draining. It's not continuous. I threw a normal multimeter in with the car off, and saw 10mA or so with occasional 400mA loads. Even the 400mA is probably not enough to drain it overnight, so I think there's something on the order of a couple amps (like maybe a window motor circuit run amuck. I do have one which stopped functioning about the first time the car died, and the car has a "roll the windows down if some magical key code is detected" feature. One theory right now is when the car is idle, every once in a while it tries to roll down the stuck window and it wouldn't take it too long if that's a 5-10A load.

 

[drmartin]

While I love to build things and see them come to life.... In this case, I'd probably just buy a test instrument built for the job at hand.

See http://www.ebay.com/itm/UNI-T-UT210...672207?hash=item2345b9fd0f:g:xQkAAOSwD0lUd87j

for an example. Search eBay "DC Current Clamp" for many alternatives... Just be sure the fine print says "DC Current"

If you are really trying to log hours of data then perhaps something like

http://www.windandwet.com/software/datalogger/current.php

would be of interest. Note that the hall effect device and the opamp would work together to saturate and limit the input to the ADC. The circuit shown is for a 5v system but the mods needed to drive a 3.3v ADC are pretty straight forward. Just change the supply voltage to the opamp from 5v to 3.3v.

Also you can refer to the discussion thread

https://forum.pjrc.com/threads/3014...-amp-to-analog-input?highlight=protect+inputs

One example source of current sensors which withstand higher overcurrents is

http://www.ebay.com/itm/Panel-Mount...981595?hash=item567ceb47db:g:sNwAAOSw9mFWG4VI

http://www.ebay.com/itm/50-100-150A...e-DIN-Rail-or-Panel-Mount-ACS758/201448133457

or just search "ACS758" or "ACS712" (which you already know about....)