Current Sensors built from stratch.

[Zanderist]

I'm currently looking into making a current sensors however my issue is practicality due to the wattage rating of the Sense resistor.

Consider this diagram

http://www.circuitdiagramworld.com/monitor_circuit_diagram/High_Side_Current_Sense_and_Fuse_Monitor_13623.html

I did a calculation and I concluded from it that I need a resistor that is >1.25watts. I felt that may be a little big.

Another point I'm curious about is how would you go about to make it rated for say 100 Amps?

I've been doing google searches on the subject only I am finding it frustrating that it tends to only show hall effect type.

I am also not sure on what to look for an op-amp other than having it be a rail to rail one.

What I hope to accomplish is an understanding on how to make a current sensor in order to provide a tutorial on how to make and integrate with an arduino

[shandy27]

from what i can remember decrease your rsense resistance to increase current measurment.

[doobedoobedo]

Why don't you want to use a Hall effect?

[Zanderist]

Why don't you want to use a Hall effect?

I've just ordered two of them to try out, but it's big.

If I go with an opamp way of doing it I could account for size when I go to make a final build

[CraigHB]

In that respect you're limited using a current sense resistor.  If you wanted a range up to 100A (which would be very high for any circuit), you're talking 10W on a 1 mOhm CS resistor.  That's a lot.  You can find them with 3W tolerance in a 3216 metric package which isn't terribly large in terms of PCB space.  However, the problem you run into when you go too low with a current sense resistor and op-amp is noise.  The smaller the signal the larger the gain.  With big gains you amplify noise as well as signal.  The lowest I've gone myself is 5 mOhms for a 40W mod limited to 10A.  I think it would be okay with a 2 mOhm CS resistor, but haven't actually tried it.

You may run into the same problem with power consumption using a Hall effect sensor when getting into currents that high.  They do have some resistance, though less than a typical CS resistor. 

In terms of what you need in an op-amp, you need both rail to rail input and output for a current sensor.  Another thing to look at is offset.  Ideally the less offset the better.  Any op-amp with much offset will need it calibrated out.  There's also consideration for temperature drift.  The cheapo op-amps I use do drift a bit.  My resistance readings drop a couple hundredths when the mod gets warm.   

Any precision op-amp with rail to rail input and output should be good.  You can use a general purpose op-amp and it will work, but it's worth it to use a precision one, they're not *that* expensive.  A good example is the Microchip MCP6V26.  I'll probably use that one next time I design a mod board.

[doobedoobedo]

The MCP6V26 is in short supply in the UK. DO you think the LTC2050 would be a suitable replacement?

[CraigHB]

Should be good.  Generally speaking, Linear makes a lot of specialized parts you can't find from other makers.  The problem with them is their stuff is considerably more expensive than anyone else.  They're also stingy with samples.  I've used their parts before, but only when I can't find the same thing from another maker.

If you register at Microchip's site, you may be able to get free samples even if you can't find stock at online shops in your country like Mouser UK or Digikey UK, or Farnell.  Microchip also has an online store, but not sure about shipping costs for UK, might be cost prohibitive.  Though most of their stuff ships from Asia so it may not make a difference.

[doobedoobedo]

Cheers I thought they looked OK.

The UK mouser and digikey sites are actually US sites which gather UK taxes. Everything comes from the US with huge postal costs unless you make a big order (£50 before tax). I get everything from the Farnell UK/French warehouses with the very odd exception because of that...

[Ian444]

Zanderist, that schematic doesn't make sense to me, in that the resistance of the fuse needs to be taken into account, a fuse replacement could change the output of the op-amp and skew the calibration. Or am I missing something? (usually the case...) Most fuses I've checked the specs on are way greater than 2 milliohms, more like 10 to 15+.

[CraigHB]

Didn't look at that circuit diagram closely.  You mentioning that error prompted me to do it.  Yeah the fuse can't be put there.  The op-amp inputs have to be directly above and below the CS resistor.  In fact care is required in PCB design that those traces are short as practical.

Even if it's possible to compensate for the added resistance (which it's not), if a resettable fuse trips there it will put full supply voltage on the op-amp inputs which could fry it depending on the op-amp.  Some op-amps have less than supply voltage tolerance on input differential.

Another issue is the diagram shows high side sensing.  The op-amp mentioned only does low side sensing.  In other words the CS resistor should be on the atomizer's return feed (negative side).  It's possible to do high side sensing, but an "over the rails" op-amp is required and they're not all that easy to find.  I know because I've had occasion to need one for a project other than an e-cig.  I ended up redesigning the circuit to eliminate the need.

[norcalreballer]

I'm using 2 1mohm 2w 2512 shunts in parallel.  Good for up to 90a.  Why not use a shunt monitor instead of an OP amp?

[doobedoobedo]

The op amp is used as a differential amplifier to amplify the tiny voltage drop across the shunt.

There's a diagram for low side current sensing on page 12 of the LTC2050 datasheet I linked earlier.

[norcalreballer]

There's amplifiers specifically designed to read the voltage drop across a shunt resistor.  They're called shunt monitors.  Some has some very cool and useful features. 

[CraigHB]

Yes you can find shunt monitors specifically designed for that purpose.  Though I've found they tend to have a preset gain which can be a problem for resolution.  Ideally you want the full swing of the output to be proportional to the full swing of the gauge to get the best resolution.  If gain is too much greater than what you need, you can end up with an unreliable meter.  Using a precision op-amp requires a few more parts, namely a network of four resistors, but it's also most flexible in terms of setting gain.  However features such as high side sensing and and serial data outputs can make it worthwhile to use a current sense amp.

[norcalreballer]

I'm using the INA226.  i2c, only needs an input cap, reads bus voltage, has settings for conversion times(140µs minimum) and averaging, alert pin with different functions, 16bit ADC with internal reference, etc.  The only limitation is it'll only read up to +82mV.  It works great even with a 0.0005ohm shunt. There's even more to it than that.   Little on the expensive side, but totally worth it to me.

This monitor made it so I don't have to use the internal µC ADC/external reference at all in my project.  Pretty friggin' cool if you ask me.