Current sensing - troubleshooting MAX4080 - improperly amplified low voltages.

[drogus]

I'm using the following circuit to measure current with a 8mR shunt resistor and MAX4080:



When I measure currents from about 0.7A and up I get a proper readings, ie. I get a current value roughly similar to what I get on a multimeter. I calculate the current using the following equation: v_sense / (gain * r_shunt), where v_sense is a value on OUT pin of MAX4080, gain is MAX4080's gain (60 in my case) and r_shunt is a value of the shunt resistor.

The problem is that when I try to measure anything lower than 0.7A, the reading is off. It doesn't even register any values smaller than 0.3-0.4A (ie. voltage on OUT is ~0). When current gets up to about 0.3-0.4, voltage starts increasing, but it's off until it gets to about 0.7A (it's hard to get precise readings with my limited equipment, but that's a rough description of what happens).

I checked the reading from the resistor and it seems correct, for example when a power supply shows 360mA, I get 2.9mV, for 160mA I get 1.3mV, so it looks like the problem is with MAX4080.

One thing that I noticed after going through MAX4080's datasheet is that they recommend using 0.1uF capacitor and I used 1uF. According to my (limited) knowledge it shouldn't change how an amplifier works in that way, but if you think that it may be the problem I can try swapping the capacitor with something smaller.

What can I do to troubleshoot the problem? Should I just buy another MAX4080 (or two?) and check them, because my IC is faulty?

[The Soulman]

The polarity's in your schematic are wrong, vcc should be positive and also the current through your shunt resistor
needs to flow from the terminal connected to RS+ to the terminal connected to RS-, the max 4080 measures current in one direction only, not both ways like a multimeter.
The value for the capacitor should be fine, again mind the polarity.

http://datasheets.maximintegrated.com/en/ds/MAX4080-MAX4081.pdf

[rstofer]

See the datasheet, especially note 5 where it gives a range of input voltages that will keep the output within tolerance.  Then see note 7 where they are looking for quite a high Vsense to meet other specs depending on the particular device.  Your 8 mOhm resistor and, say, 0.5A only yields 4 mV and they're talking about 100 mV or more of Vsense.  You seem to be falling off the bottom end of the Vsense requirements and that is consistent with your observations.

I didn't spend a lot of time on this, I just looked for numbers to support the idea that you need to drop a lot more voltage across the resistor.  There may be other specs that say different.  Maybe there are some app notes as well.  Note that the maximum input differential voltage is 80V!  They're not worried about a 100 mV, more or less.

In very general terms, bypass capacitors are chosen by the frequency of concern.  Smaller capacitors for higher frequencies.  That's why you sometimes see 0.1 ufd or 0.01 ufd.  It's the frequency that determines the size.

Page 5, lower left corner:
https://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf

Unfortunately, 0.1 ufd is not always the right answer.  But that's where I start since I have a reel of 4000 pieces!

[drogus]

The polarity's in your schematic are wrong, vcc should be positive and also the current through your shunt resistor
needs to flow from the terminal connected to RS+ to the terminal connected to RS-

Sorry for that, I haven't even noticed that there's a plus on the connectors. The PCB is correctly marked and the voltage indeed goes from RS+ to RS- (just like I mentioned, the voltage drop on the resistor is correct).

See the datasheet, especially note 5 where it gives a range of input voltages that will keep the output within tolerance.  Then see note 7 where they are looking for quite a high Vsense to meet other specs depending on the particular device.  Your 8 mOhm resistor and, say, 0.5A only yields 4 mV and they're talking about 100 mV or more of Vsense.  You seem to be falling off the bottom end of the Vsense requirements and that is consistent with your observations.

That seems like a possible explanation, thank you! And again I was bitten by not reading the datasheet thoroughly enough.

Do I understand correctly that in this case MAX4081 would work better (even if I don't need a bidirectional current sesnse), because it has full range listed in Note 5?

[rstofer]

Do I understand correctly that in this case MAX4081 would work better (even if I don't need a bidirectional current sesnse), because it has full range listed in Note 5?

The MAX4081 datasheet shows a value of Vsense for full scale.  You want to go to full scale so you can get resolution.  It makes no sense to use the device to, say, 1/10 of full scale.  External gain doesn't count.  All that does is multiply the value of 1 step of resolution.  So you get a staircase with steep stairs.

I suppose it works out that the '81 has a full scale of 50 mV but the analog range of outputs is still cut in half.  The output is just shifted down by 1/2 the range.

I have to get off the keyboard and hit the barbecue.  I'm the designated chef...

[rstofer]

Do I understand correctly that in this case MAX4081 would work better (even if I don't need a bidirectional current sesnse), because it has full range listed in Note 5?

The MAX4081 datasheet shows a value of Vsense for full scale.  You want to go to full scale so you can get resolution.  It makes no sense to use the device to, say, 1/10 of full scale.  External gain doesn't count.  All that does is multiply the value of 1 step of resolution.  So you get a staircase with steep stairs.
  ETA:  I don't know what I was thinking about here.  We haven't digitized it yet...

I suppose it works out that the '81 has a full scale of 50 mV but the analog range of outputs is still cut in half.  The output is just shifted down by 1/2 the range.

I have to get off the keyboard and hit the barbecue.  I'm the designated chef...

Corrected...

It would seem to me that in a unipolar application, 0 .. 100 is probably better than -50 .. 50  What appears to matter is which specific device is being used.