IMA 226 shunt monitor with multiple different shunt resistors

For a project I'm trying to build a measurement tool for USB where I can measure U, I and P using the INA226 from Texas Instruments.

The shunt resistors all have a different value to measure a current from 0.200 µA to 4A. Of course it's going to measure voltage but the INA converts it to current and later on to power.

Between the vbus and the shunt resistors I placed some Power Distribution switches so the microcontroller can chose which one would be suited from the collected measurement from the INA.

The problem is that the RDs(on) on the switch where the shunt resistor for 4A is located is almost as high as the value of the shunt resistor.

Will the INA work as configured here or do I need to place a switch so it only measures over one shunt resistor and place V+INA226 after the power distribution switch so the RDs(on) doesn't interfere in the measurement?

  • \$\begingroup\$ I don't think series resistances are an issue, because your INA shunt monitor should have differential inputs across ONLY your shunt resistor. It will just output a current gain based on the detected voltage (and then you put your own gain resistor to get a voltage out of the produced current) \$\endgroup\$ – KyranF Oct 15 '14 at 13:46
  • \$\begingroup\$ @KyranF the schematic shows the monitor across the bank of switches; you'd need either many monitors or additional switches to connect the single monitor to the selected shunt \$\endgroup\$ – Pete Kirkham Oct 15 '14 at 13:56
  • \$\begingroup\$ Yeah i thought of that but then i still can't connect it over all shunt resistors because all the traces would be interconnected and the current would always flow straight trough the smallest resistor. I was thinking of something like a demux or a switch so I don't lose the voltage and I can switch between the 5 resistors using the microcontroller. Any ideas? \$\endgroup\$ – Joris Aerts Oct 15 '14 at 13:58
  • \$\begingroup\$ to get a good and accurate measurement, you really should just have a bank of shunt monitors with their own resistors each. You are meant to do kelvin sensing, and use very short traces, and use 1% or less tolerance resistors. I would avoid trying to do it like this because it's not really how the IC was intended to be used.. You have pointed out one of the main flaws anyway, which is multiple resistors in parallel, where the switches in the circuit do indeed change the values. You could calibrate and adjust the gain/output though, if you knew which channel you were using i guess. \$\endgroup\$ – KyranF Oct 15 '14 at 14:07
  • \$\begingroup\$ @PeteKirkham yes, sorry I did not notice the very faint grey text that is meant to represent the connections of the INA226 on the schematic \$\endgroup\$ – KyranF Oct 15 '14 at 14:08

Connect all the shunt resistors in series, and put the INA226 across the ends of the string. The range switches are then outside the current sense path, so their internal resistances won't affect your measurements.

For lower currents the values of the resistors must add up to the shunt values you need, eg. 4A shunt = R5, 400mA shunt = R5 + R4, 40mA shunt = R5 + R4 + R3 etc.

In the partial circuit below I have shown FETs to represent your load switches. Note that the TPS22943 is limited to 40mA maximum current.

enter image description here

| improve this answer | |
  • \$\begingroup\$ Hi Bruce, yeah I knew about the TPS 22943 but it was just as an example. looking at your circuit I am wondering how V+INA is going to measure a voltage when only R5 is connected. \$\endgroup\$ – Joris Aerts Oct 16 '14 at 10:40
  • \$\begingroup\$ It reads the voltage through the other shunt resistors. Since the shunt resistances should be much lower than the INA266's input impedance, they should not significantly affect the voltage reading. \$\endgroup\$ – Bruce Abbott Oct 16 '14 at 20:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.