Sunday 18 February 2018

The μCurrent Custom

Working on TT7F brought my attention to low consumption modes and general power saving a lot. My efforts, though, were somewhat accompanied by the lack of sufficiently accurate instrumentation. Recently, I came across David L. Jones' μCurrent thoroughly documented on his EEVBlog and adjacent Youtube channel. After checking the parts list with a local supplier, I thought I would design my own μCurrent board and learn more about the circuit in doing so.
The result can be seen in the image above, and the Eagle files, for anyone interested, can be downloaded from here: uCurrent.brd and uCurrent.sch. As apparent from the schematic, my version is for the most part based on the original μCurrent with a couple of minor changes. The selection of modes (off, on, on-short) follows the μCurrent GOLD's options, and the current limiting resistors R4, R9, R10 and R12 now have the same value (240Ω). All resistors and capacitors are in a 0603 package for easy hand soldering.
I ordered the 30x40mm PCBs from OSH Park for $9.30 for three boards. The layout of the main section with amplifiers for the most part copies the original proven design.
The rest, on the other hand, is fairly packed to fit the much smaller dimensions. The banana receptacles are replaced with simple pins which serve as small leg stands as well.

setting range resistor amplifier absolute
1mV/nA +/- 0-1250nA +/- 3.16mA +/- 165uA +/- 165uA
1mV/µA +/- 0-1250µA +/- 100mA +/- 165mA +/- 100mA
1mV/mA +/- 0-1250mA +/- 10A +/- 165A +/- 300mA

David provides all sorts of specs for both of his versions on the website. But I was more interested in a real life question such as: if I want to measure a microcontroller's sleep current say in the μA range, will the high current prior to the microcontroller's transition to sleep damage the μCurrent? Looking at the circuit, the range in which it can measure is determined by the supply voltage, the shunt resistor, and the point at which the amplifier (MAX4239) hits the rail and starts capping the output. Considering the lowest operating voltage of 2.7V and taking into account the amplifier's output voltage swing (at worst it could get 100mV from the rail) gives the table value of 1250mA for the mA range. $$2.7V\:/\:2\:-\:0.1V\:=\:1.25V\:/\:100\:/\:0.01Ω\:=\:1.25A$$ To determine the absolute maxima the circuit should withstand, there are two main factors to look at. The power ratings of individual shunt resistors and the voltage limits on the amplifier's input pins. The resistors on my μCurrent are rated for 100mW (10kΩ), 100mW (10Ω), and 1W (10mΩ) which leads to 3.16mA current rating in the nA range for example. $$\sqrt{\:0.1W\:/\:10kΩ}\:=\:3.16mA$$ The MAX4239's datasheet states the absolute maximum rating for IN+ and IN- pins to be Vgnd-0.3V and Vcc+0.3V. Considering again 2.7V power supply leads to the allowed maximum of 165mA flowing through the μCurrent in the µA range. $$2.7V\:/\:2\:+\:0.3V\:=\:1.65V\:/\:10Ω\:=\:165mA$$ Looking at the table above, the absolute maxima for individual ranges are 165uA in the nA range due to the amplifier's input rating, 100mA in the µA range due to the resistor rating, and 300mA due to the mechanical switch rating which I haven't yet mentioned. That is probably the reason why the switch was replaced for a different type on μCurrent GOLD.

From the computations above, it is apparent that supplying μCurrent with 5.5V, which are the maxima for both MAX4239 and LMV321, would increase the measuring range and the absolute maximum rating in the nA range. The amplifier would now cap at 2650nA/µA/mA and tolerated 305μA in the nA range.
Here are a few images from quick testing. There are three 1MΩ resistors in series connected across a 2.49V pack of batteries. The μCurrent is set in the nA range (1mV/nA) and the multimeter measures millivolts in its 2000mV range. The second image shows the same setup with only two 1MΩ resistors.
To measure the current flowing through only one 1MΩ resistor, I had to switch to the μA range (1mV/µA), but I could also use the finer 200mV range on my multimeter.

The list of parts is very similar to the original μCurrent. I only had to find an alternative to TPS3809 which turned out to be TCM809 voltage monitor. And I had to go with only 1% precision 10mΩ resistor as opposed to 0.5% LVK12R010DER on the original. Overall it cost me about $33 to make which was mainly due to the fact I made only one and had to order some of the parts in larger quantities.