# Wide range and high precision current measurement

I need to measure current over a large range from a few µA up to 2A. The application is sensitive to noise at later calculations with the measured current. My idea is to use different shunt resistors which are bypassed as needed.

Has anyone some experience with this situation and has already selected a bypass switch for that purpose? My first idea is to use a simple bistable relay but I don't know how much that would affect noise behaviour on low current measurement.

• What resolution do you actually need and do you need it over the full range? – po.pe May 5 '18 at 9:00
• For wide range measurements I have used a current mirror with a voltage controlled output gain stage. – Peter Smith May 5 '18 at 9:00
• How much noise can you tolerate? What precision do you need? How fast do you need to measure? All of these things have a bearing on how you should implement this. The switch you choose is more likely to introduce offset and drift errors than noise. @PeterSmith can you elaborate on your current mirror solution? – Loganf May 5 '18 at 11:27
• at what rate do you need the samples? if slow enough, you can feed an RC with a shunt and using timing to measure instead of voltage. – dandavis May 5 '18 at 13:17
• I already picked an AD converter which makes up to 500k samples per second but speed is not the problem here. I could live with much slower values. Unfortunately the accuracy is needed at the full range. – Gustavo May 6 '18 at 12:09

Simply bypassing a current shunt requires an impractically low resistance switch, when the currents start to get into the amps.

Far better is a current routing switch as shown below. Each switch element can now be any resistance without affecting the metering accuracy, it only affects the voltage drop across the meter. This means you can choose (for instance) a power FET for SW1, and an analogue multiplexer for SW3 and SW4. You could even use BJTs, which would not work at all well as low resistance switches, but work just fine for current steering.

simulate this circuit – Schematic created using CircuitLab

• Reed relay is even better for low noise. – Long Pham May 5 '18 at 9:22
• @LongPham reed relays are worth considering, but you have to look at all their specs. Hot switching is not great for instance, if that's important in the final application. – Neil_UK May 5 '18 at 9:37
• The series resistance to the opamp input on the highest current setting will cause offset errors and make the current noise worse. Would it be better to switch the +ve input to different points in the chain rather than switching the current path? – Loganf May 5 '18 at 11:33
• @Loganf The input current path must be switched. The amplifier could be switched to the same point as well, if you really think the extra resistance of R4 is significant. Not doing so saves an analogue multiplexer. Perhaps you'd work out for me, and reply in the next comment, what the additional offset and current noise is from a 90ohm R4, or even a 90k R7, when using something like a TL074 or a LMC6482 as the follower. – Neil_UK May 5 '18 at 13:00
• @niel_uk taking your 90k R7, and assuming 5pA/rtHz (which is pretty good for an opamp), you get 450nV/rtHz, or 142uVrms for a bandwidth of 100kHz.(nearly 1mv peak). This translates to a 100uA error when your R6 is in circuit (for 10k shunt). Maybe that level of error is okay? This didn't include the shunt resistor noise or opamp voltage noise etc etc. – Loganf May 5 '18 at 13:22

Have a look at the schematic of TX3 multimeter to see how Tek did this.

https://www.tek.com/manual/tx1-and-tx3-service-manual

• I took a look at the manual and it seems they use three shunts and route the current through a mosfet. – Gustavo May 6 '18 at 12:14
• Indeed it does. My point: TEK know what they are doing. It is a good arrangement to copy - it is protected against abuse to a high level, and Clearly works OK, these meters have 100nA resolution. – Henry Crun May 6 '18 at 20:59
• Good solution. Is the fuse 2A? There seems to be a risk of burning up the 50 ohm resistor if the voltage is high and the wrong scale is selected. Maybe the firmware is fast enough to change scales before this happens. Note that Neil's solution has the same potential issue. – Mattman944 Jun 23 '19 at 15:43