Whats the best Inamp for high speed current sensing?

Hi all,

I am working on a device to measure the current of a supply in real time.  The purpose of this, is to help with software development of ultra-low power 'always on' devices to be more efficient.  The biggest market for this currently is IoT.

I want to have both speed and accuracy from an amplifier and I have therefore discarded all of the purpose made current sense amplifiers as they don't provide sufficient bandwidth for my application.

The specifications required are as follows:

  • The highest voltage supply to be measured will be +5V to 0V. and the maximum current will be 1A.  I will most likely use a 50mohm shut resistor, giving the maximum differential voltage of 50mV.  I would also like for it to be accurate to the 100'suA scale, meaning that the differential voltage accuracy needs to be < 10uV (although I may use a 0.1ohm resistor for the lower scale). 
  • I need a high bandwidth across the gain range as I plan to use a 500ksps ADC to digitize the analog voltage, although will most likely be using 100-150ksps.
  • It needs to have very good common mode rejection ratio.
  • It needs to have the voltage offset as low as possible to get the best accuracy.
  • Adjustable gain is a must ideally from an external resistor, that I will drive from a digital Pot.

I have been looking at the AD8220 for a rail to rail inamp, or the AD8221 which appears to be a better device for me, but will require a dual supply to get the full swing.

Can anyone suggest any better suited iamps that I have missed, or do these 2 seem the best suited for this application?

Any support will be very much appreciated.

Many Thanks Ian.

  • Hello Ian,

    Yes, I agree that your choice would be sufficient enough in your application. But you may also want to consider AD8420 or AD8421. AD8420 is a wide supply range, micropower, rail to rail inAmp but the bandwidth is a little lower. AD8420's CMRR is also excellent. While AD8421 is a low noise low power In Amp. It has low voltage offset of 25uV.

    I also want to share our tool for in amp. It is a In amp operating range tool that will show common mode voltage vs the output voltage, this might help you in choosing the proper inamp considering your input signal and your supply. You can visit the tool here.

    Apologies for late response.

    Regards,

    Phil

  • Hi Ian,

    In addition to Phil's response, I just wanted to point out that there are a few issues using a digi-pot or solid-state mux to set the gain for a standard in-amp. There are several threads about this, such as this one: https://ez.analog.com/message/175432#175432. Also, see this short article that has some more detail: Use a PGIA to avoid getting burned by switch parasitics.

    If you need more granularity in gain than the AD8250/AD8251/AD8253 integrated PGIA family can provide, the best option is generally to take the minimum gain that you will need with the first stage in-amp and follow it with a programmable-gain op amp stage based on the digipot. There is a good discussion on how to build a good PGA with solid-state switches in the op-amp applications handbook (Section 2.2), which I would suggest reading.

    Generally, the widest bandwidth in-amps that we offer are dual-supply products. For single-supply, there might be a few products such as AD8422 or AD8226 with enough bandwidth, but I think if you check the in-amp tool that Phil linked above, you'll find that you probably need a dual supply anyway due to input/output headroom requirements. AD8421 is a great product, but it may have more bandwidth than you really need (2MHz at G=100, whereas you're talking about Nyquist rates in the 50kHz to 75kHz range). I would suggest that you consider using a dual supply and something like an AD8228 on the first stage (AD8228 is similar to AD8221, but with fixed gains, and therefore lower gain drift which will be important because 10µV in 100mV full-scale is 100ppm). Then you should have plenty of options to choose from as the op amp for your PGA.

    Best regards,

    Scott

  • Thanks Phil and Scott for some very helpful points.  I hadn't appreciated that the capacitive loading of the digpot would greatly affect the internal gain control of the inamp.

    The best option for me (seeing as I would require a greater variation in gain than the pin selectable inamp's would give) would be to use a fixed gain inamp AD8221 or the AD8253 which can be set to a gain of 10 and appears to be higher spec'ed in the area of interest.  Then to have an op-amp for the second stage, say OP1177 with a digital pot AD5160.

    I am also looking into having an analog Differential 8-to-1 Multiplexer on the front such as the ADG707 or for higher than 5V ADG1407, so that multiple supplies with their own shunt resistor, can be switched into 1 inamp and op-amp combo.  One side of the shunt sense lines being connected to SxA and the other to SxB.  This would save cost, but mostly for us, board space.

    I have chosen these multiplexers for their on-resistance flatness.  The switching response time should not be an issue, as there is no requirement to quickly switch supplies between measurements, the driving force is to enable multiple supplies to be measured without powering down the unit under test to add a shunt resistor. 

    Do you see any issues with using these multiplexers in this way?

    Best Regards

    Ian

  • Hi Ian,

    Muxes like the ADG1407 have to use larger devices to get the low R_on. Therefore they end up with higher capacitance. In-amps like the AD8253 or AD8221 have very high input resistance, so the on-resistance of the mux in series with the in-amp input doesn't introduce much error. In cases like this you can usually get better performance from a low capacitance mux like the ADG1207 instead. Other than that, it all seems quite reasonable to me.

    Best regards,

    Scott

  • Have you looked at the AD8130? it's intended as  a differential line receiver, but is very fast ( a faster variant of the AD8420 mentioned above).  It uses indirect current feedback, so for example you can refer the output signal to mid supply for bipolar input to a differential ADC. It only uses a single gain resistor, so it is possible to change the gain without the CMRR issues. and it works nearly rail to rail, so no issues with the "hexagon".