logo elektroda
logo elektroda
X
logo elektroda
Advanced Search
logo elektroda
Build faster with M5Stack »
Build faster with M5Stack » Stick • Unit • Cardputer
ADVERTISEMENT
Dostępna jest polska wersja

Czy wolisz polską wersję strony elektroda?

Nie, dziękuję Przekieruj mnie tam

uCurrent for the poor - an adapter for measuring small currents

zgierzman  27 64320 Cool? (+32)
📢 Listen (AI):

TL;DR

  • A cheaper, slimmer clone of David Jones’s uCurrent measures small currents with a shunt resistor and amplified output.
  • It replaces the CR2032-powered, dual-opamp scheme with a simpler one-opamp circuit, two 12 V batteries, and input protection.
  • The adapter has two ranges and a 0 to 10,000 mV output, covering 0 to 10 uA in nA mode and up to 10 mA in uV mode.
  • Measured offset stayed below 1 mV, and with the input shorted the output drifted only 0.3 to 0.8 mV.
  • Calculated accuracy was better than 0.5%, and the author estimates about 1% because no calibration reference was available.
Generated by the language model.
Homemade current meter with red and black terminals and switches labeled 1mV/nA, 1mV/µA, and OFF.

I came across a project once David Jones uCurrent .
All in all, it is a very simple system. Voltage is deposited on the measuring resistor, which is then amplified x100 and fed to the output.
Unfortunately, the item is not cheap, about 70 EUR excluding shipping costs.

Screenshot of an online store page displaying the EEVBlog µCurrent GOLD current adapter.

I decided to make my version, slimmer, cheaper, but with similar (or even better in some respects) parameters.
What did I not like about the original?
- Power. One CR2032 battery and artificial zero. The circuit is complicated (like one opamp, but always), and the power supply of the output amplifier is only +/- 1.5 V.
- The use of two expensive and hard to reach opamps in the amplifier. The cascade circuit improves the frequency response a little compared to one amplifier with higher gain, but is megahertz important, for example, when measuring the current consumed by the circuit in the sleep state?

Changes made by me:
- Two ranges instead of three. Each multimeter measures the milliamps, and for the oscilloscope you can use a resistor in series.
- Easily available operational amplifier - you can buy it at TME.
- Power supply from two 12 V batteries. This gives an output voltage range from 0 to 10,000 mV, so in the nA range I can measure currents from 0 to 10 uA, and in the uV range up to 10 mA. This is 10 times more than the original.
- Place for a diode protecting the amplifier input. Just in case.
- I also resigned from the battery voltage monitoring system.

Thus, the layout was created as in the diagram. Simple, not to say primitive.

Circuit diagram of a device featuring components such as resistors, capacitors, and an operational amplifier.

The value of the capacitor C1 has been transferred from the original schematic - if necessary, I will try to improve the compensation using a generator and an oscilloscope.

A tile pattern was created behind the diagram. As the scheme is simple, the board came out in one layer without any problems, without combining with bridges etc. Easy to make at home.

PCB layout design for a current meter circuit.

And then the layout itself.

Close-up of a printed circuit board with mounted components and electrical circuits.
Prototype of an electronic device with switches and batteries.

0.1% resistors

Time for measurements:

The offset voltage is less than 1 mV. With the input shorted, the output voltage fluctuates in the range of 0.3 to 0.8 mV.

The accuracy of indications is surprisingly high, better than 0.5%. But I assume it will be safe to assume 1%. Calculated using the technical method, because I have nothing to calibrate this device with. Yet the meters I use to measure voltage and resistance have their limitations.
The voltage source was a 3V battery, a current limiting resistor in series and my adapter. The voltage is measured at the ends of this series circuit, and its total resistance, including the leads, is measured with a multimeter.

For the uA range:
At the entrance = 3.1630 V, R = 99 890 ?
And the theory. = 31.6 uA
And d. = 31.8 uA
error = 0.42%

For the nA range:
At the entrance = 2.9434 V, R = 5134000 ?
And the theory. = 573.3 nA
And d. = 574.4 nA
error = 0.19%

Enough for me.
Attachments:
  • rev3.0.sch (362.47 KB) You must be logged in to download this attachment.
  • rev3.0.brd (85.65 KB) You must be logged in to download this attachment.
About Author
zgierzman wrote 1775 posts with rating 1533 , helped 108 times. Live in city Zgierz. Been with us since 2005 year.

Comments

ptero 12 Nov 2020 14:58

1. How with the stability with ordinary resistors? 2. Are these metal plates short-circuiting the metal housing of the battery? - just scratch the paint. [Read more]

TechEkspert 12 Nov 2020 16:20

The original has a near-zero DC offset and drift through the use of autocorrelating zeroing techniques ", how can the effects be affected by using OPA192 instead of MAX4239? Maybe you will make a wider... [Read more]

zgierzman 12 Nov 2020 17:33

OPA192 landed there a bit by accident. Now I see that TME has MAX4239 in SO8 housing, but I didn't see it then (or it wasn't on offer). Besides, its power supply is +/- 2.75 V maximum, so I would... [Read more]

And! 12 Nov 2020 18:41

If you wanted to share the project with more people, the "kit" would actually be better than the assembled product, while banana connectors, as you write, are optional, what is important is PCB, acrylic,... [Read more]

piterek-23 12 Nov 2020 19:24

A great project and such gadgets I would like to see in the elektroda.pl store ;) Where did you get these battery holders? I'm looking for those, but with AAA batteries, and I can't find som... [Read more]

zgierzman 12 Nov 2020 19:50

https://www.tme.eu/pl/details/keys82/baterie-pojemniki-i-uchwyty/keystone/82/ This particular model fits AAA, N, and A23, which is exactly 12V. The only difference is the "length" spacing. https://obrazki.elektroda.pl/7141301100_1605207005_bigthumb.jpg... [Read more]

Błękitny 12 Nov 2020 21:38

Congratulations on a successful project! I have a question because I am building something similar only on the LF356, namely what is the measurement bandwidth, could a colleague measure with a current... [Read more]

Urgon 13 Nov 2020 09:26

AVE ... A very nice alternative, although I personally will not find it useful. Plus, absurdly cheap. I always thought the original uCurrent was a little too expensive compared to its intended use. Autozero... [Read more]

Janusz_kk 13 Nov 2020 09:38

But why? Secondly, it will not compensate for it as there are amplifier inputs, it is better to give an amplifier with processing. [Read more]

zgierzman 13 Nov 2020 09:55

Here you are. https://obrazki.elektroda.pl/4282371600_1605256653_bigthumb.jpg Bandwidth varies greatly depending on the range. In the uA range it is only 73 kHz. The rise time of the impulse response... [Read more]

Urgon 13 Nov 2020 13:11

AVE ... @Janusz_kk I was thinking about something like this: https://obrazki.elektroda.pl/8384328200_1605269099_thumb.jpg Q1 and Q2 should be glued together, as well as Q3 and Q4. R1 and... [Read more]

Janusz_kk 13 Nov 2020 13:57

I guessed :) except that this system does not exist here, this is how the oscilloscope inputs are made, which must be high ohmic, but here the gain from it is negligible, because we do not need such a... [Read more]

zgierzman 13 Nov 2020 14:38

In the diagram @Urgon is the LF356 which has this feature: "JFET Input Operational Amplifier". I am also eager to find out why it is worth adding a handful of discrete elements to add a JFET buffer... [Read more]

Urgon 13 Nov 2020 15:33

AVE ... There is an error in the feedback circuit in the diagram - wrong resistor values and wrong connection layout. You're right, it doesn't make sense a bit to be honest. I used the schematic... [Read more]

PiotrPitucha 18 Nov 2020 16:24

Hello Today you can buy good amplifiers looking at temperature drifts etc. In the past, these inventions did not exist and you had to deal with them :) In the Czech Amaterkie Radio there was an excellent... [Read more]

Urgon 18 Nov 2020 16:40

AVE ... IN one episode of EEVBlog a meter is shown which has a current range of 100 attoamps, or 0.0001 pA. JFET input stage, all spider-soldered on probably polyester supports. [Read more]

Jawi_P 20 Nov 2020 17:56

Very interesting. Interesting resistance at the input. But for such ranges it is probably not strange. [Read more]

Urgon 20 Nov 2020 22:59

AVE ... I will add one more curiosity from the 1960s: a circuit with an input impedance of more than 1 T?, an output impedance of 50 ? with a voltage gain of 0.98. https://obrazki.elektroda.pl/2485299900_1605909539_thumb.jpg... [Read more]

TechEkspert 21 Nov 2020 10:09

The value of the resistor on the gate of the BFX63 transistor makes it difficult to buy in a local electronics store (there probably still existed in those years). I understand that this is not a DC ... [Read more]

FAQ

TL;DR: DIY uCurrent-style adapter costs ≈ PLN 25, hits 0.19 % error, and “the accuracy of indications is surprisingly high, better than 0.5 %” [Elektroda, zgierzman, post #19039005] Why it matters: you can probe nano-amp sleep currents without a high-end instrument.

Quick Facts

• Offset voltage: < 1 mV with input shorted [Elektroda, zgierzman, post #19039005] • Bandwidth: 73 kHz (µA range) / 1 MHz (nA range) [Elektroda, zgierzman, post #19040663] • Measurable currents: 0-10 µA (nA range) and 0-10 mA (µA range) [Elektroda, zgierzman, post #19039005] • Parts cost: approx. PLN 25 (PCB, 0.1 % resistors, OPA192, switches) [Elektroda, zgierzman, post #19039443] • Power: two 12 V A23 batteries; output swing 0-10 V [Elektroda, zgierzman, post #19039005]

What is this adapter and how does it differ from Dave Jones’ uCurrent?

It is a two-range current-to-voltage converter that amplifies the shunt voltage 100×. Unlike the original three-range uCurrent, it uses ±12 V rails, drops the auto-zero section, and relies on a single readily available OPA192 instead of two rare MAX4239 devices [Elektroda, zgierzman, #19039005; #19039443].

Which current ranges and resolution can I expect?

With 100 Ω and 0.01 Ω shunts you can read 0-10 µA in the nA range (1 mV = 100 nA) and 0-10 mA in the µA range (1 mV = 1 µA) [Elektroda, zgierzman, post #19039005]

Why was OPA192 chosen over auto-zero MAX4239?

OPA192 tolerates ±18 V supplies, fits TME stock, and delivered acceptable 5 µV/°C drift in the prototype. MAX4239 can only run on ±2.75 V and was unavailable locally at the time [Elektroda, zgierzman, post #19039443]

What does it cost to build?

For a batch of a few dozen units: PCB ≈ PLN 1, battery holders PLN 5, switches PLN 6, OPA192 PLN 8, precision resistors PLN 5. Total: ~PLN 25 excluding connectors and case [Elektroda, zgierzman, post #19039443]

Is long-term or thermal stability characterised?

Not yet. The author notes no measurements of thermal drift or ageing with ordinary metal-film resistors [Elektroda, zgierzman, post #19039443]

Are kits or finished boards available?

The designer is open to supplying PCB-and-parts kits but lacks time for full assembly; estimated small-lot production would still target the ~PLN 25 parts cost [Elektroda, zgierzman, #19039443; And!, #19039613].

How do I zero the output before a measurement?

  1. Short the input terminals.
  2. Select the desired range.
  3. Adjust the offset trimmer until the multimeter reads ≤0.3 mV. This three-step nulling matches the builder’s test method [Elektroda, zgierzman, post #19039005]

What are the main limitations or edge cases?

  1. No auto-zero, so offset may drift over temperature.
  2. Bandwidth falls to 73 kHz on the 0.01 Ω shunt, limiting fast pulse capture.
  3. Metal shunt screws or finger oils can introduce picoamp leak paths; clean the board after soldering [Elektroda, zgierzman, #19040663; Urgon, #19041033].

Could a JFET input stage improve performance?

In theory, discrete JFET-buffer front-ends can push bias currents below 1 pA, but would add 12 parts and need thermal coupling; the designer felt modern low-IB op-amps render that unnecessary [Elektroda, Urgon, #19041033; Janusz_kk, #19041136].

Why drop the auto-correlating zero circuit?

Auto-zero halves bandwidth and complicates the layout. For most sleep-current work, sub-1 mV static offset and <0.5 % gain error are acceptable without extra switching artefacts [Elektroda, linuxtorpeda, post #19489620]

Can I use ordinary 1 % resistors?

Yes, but expect roughly double the gain error. 0.1 % metal-film parts enabled the measured 0.19 %–0.42 % accuracy; 1 % parts could raise total error near 2 % [Elektroda, zgierzman, post #19039005]

How fast can it capture nanosecond transients?

The 1 MHz bandwidth equates to a 350 ns full-power response, so nanosecond events will be attenuated; consider a 10× faster op-amp if sub-50 ns edges are essential [Elektroda, zgierzman, post #19040663]
Generated by the language model.
%}