EEVblog Electronics Community Forum
Electronics => Metrology => Topic started by: fracton on November 30, 2024, 11:15:09 am
-
We recently had an issue with a 6 1/2 digit digital multimeter from Keitheley at work. As it still seems to be under warranty we had to send it back to the vendor for repair. However, that got me thinking about a DIY current meter that can measure 10 - 1000 nA level currents. I am thinking about using an LMC6082 or LMC662 with an arduino uno and LCD screen to put together a current meter.
Has anyone here built something similar or more robust solutions? Any help would be appreciated.
-
The 10 -1000 nA range is not yet where one would need a super low bias OP-amp like the LMC662. I would also look at the low frequency noise. So I would consider the OPA145 as amplifier. The LMC.. are more something for the sub 1 nA range.
Depending on the use case one could consider just the TIA as ampifier and still use a normal DMM for reading the output voltage.
Just the arduino uno with it's internal ADC is not ideal. A little more resolution for the ADC could help to get away with less range switching or a larger useful range for a single resistor. So one would normally add an external SD ADC chip, like MCP3421 as the low end (especially not ideal with 50 Hz mains).
-
Like this one (a simplified example only)..
An adjustment in sw would be needed..
I already collected the ADI's MAX11167 ;D (true bipolar +/-5V input, 16bit SAR, 3x3mm TDFN, there is an 18bit version too) for such a purpose..
-
Thanks! I looked around and could get hold of a couple of LF353's dual opamps.
For improving the ADC, I'm considering to use an ADS1115, which also acts as a PGA. Alternatively I could consider using a pico pi rp2040.
Please let me know if I can make much simpler design choices. Also, I'm curious to know how an implementation of LMC662 or LMC6082 would look like and what are the additional precautions to be taken for measuring sub nA currents.
-
A complication with ADS1115 and friends is their inputs are differential/single_ended and "unipolar" (meaning the inputs has to be kept between GND and +Vcc/+Vref only, cannot be negative against GND). So, after the TIA (the same schematics as above with any of your opamps above, the sub nA requires XXX Meg feedback resistor and special construction around the input as well) you have to bridge the "bipolar" output of the TIA to the "unipolar" input of most ADCs or MCUs.
TIAs design and input voltage conditioning for such ADCs have been discussed here many many times afaik, try to search in the forum.. Best when you draw a schematics and people will comment, sure.
-
First question is the bandwidth, pure DC current and dynamic /spicking patternt require quite different design approach
-
The ADS1115 can be used in differential mode, e.g. with one side at a fixed voltage of some 2 V. This voltage would be used as the virtual ground for the input and non inv input of the OP-amp at the TIA. This way one would have a usefull voltage range of the +-2 or -2 to +3 V.
If a higher supply is used at the OP-amp, one could consider a divider between the amplifier and ADC.
With the rather low currents one naturally has to expect mains hum. Ideally one would integrate / average over full mains periods to suppress hum.
The ADC of the RP2040 is relatively poor, with a kind of bug. So nominally 12 bit with quite some INL error / jumps to get more a 8 or 9 bit accuracy and useful resolution. So the 10 bit ADC in an AVR may actually be better as oversampling can work and improve the resolution. To suppress hum one would average over many samples anyway.
There is nothing that special about the LMC662. These just have very low bias and a bit more voltage noise. So they are a really good choice for very small currents, like < 100 pA, but not that good (would still work with essentially the same circuit) for higher currents, as the voltage noise is not that great. For very low currents the resistor gets really large (like 100 M to 10 GOhm) and care is needed with the isolation / leakage.
The LF353 needs some care with the common mode range at the input and the supply range.
-
https://www.tek.com/document/handbook/low-level-measurements-handbook (https://www.tek.com/document/handbook/low-level-measurements-handbook)
Shunt ammemeter is more typical than your shunt to virtual ground (half way to a feedback ammeter).
-
1uA FSD is not taxing. Any JFET input op-amp will do the job without needing to go MOSFET. OPA1641 has good AC and DC characteristics for the job. It can cope with 12V rails, so 10V FSD fits nicely with 6 digit meters. For 1uA FSD you'll need a 10M feedback resistance, so it's easy to get close tolerance. Just be careful about cleanliness in construction and screening.
-
A complication with ADS1115 and friends is their inputs are differential/single_ended and "unipolar" (meaning the inputs has to be kept between GND and +Vcc/+Vref only, cannot be negative against GND)
I am using the ADS1118 and it can go about 0.3V below the GND rail. All circuits which go down to GND Vcm will go slightly below (e.g. see input circuit of LM358). What you cannot do (obviously) is go below GND on the output of some circuit :)
Quite often it is useful to have a -0.3V supply rail, easily made with a schottky diode fed from a resistor. 0.3V does not turn on input protection diodes. Well only a little bit ;) - depends on Zout of the source.
Re PCB layout, you may need machined slots, or guard rings driven by a voltage follower.
-
Like this one (a simplified example only)..
An adjustment in sw would be needed..
I already collected the ADI's MAX11167 ;D (true bipolar +/-5V input, 16bit SAR, 3x3mm TDFN, there is an 18bit version too) for such a purpose..
Would you mind to share more project details, driever IC, reference, noise & linearity tests ?
My new toy details posted here;
https://www.eevblog.com/forum/projects/diy-evaluation-ads8354-16-bits-dual-channels-700-ksps (https://www.eevblog.com/forum/projects/diy-evaluation-ads8354-16-bits-dual-channels-700-ksps)
-
Improvements to the common transimpedance amplifier can be made when used as a test instrument like feedback around the series input protection resistor.
-
..I already collected the ADI's MAX11167 ;D (true bipolar +/-5V input, 16bit SAR, 3x3mm TDFN, there is an 18bit version too) for such a purpose..
Would you mind to share more project details, driever IC, reference, noise & linearity tests ?
I got couple of 11166 and 11167 from Maxim long time back mainly because their true +/-5V bipolar input, still not soldered into something and evaluated. It has got 12 pins (2x6) in 3x3mm package so not easy to prototype..
Evaluating 16bits 250/500kSamples SARs is not an easy exercise, so I doubt I could do it better than Maxim..
-
..I already collected the ADI's MAX11167 ;D (true bipolar +/-5V input, 16bit SAR, 3x3mm TDFN, there is an 18bit version too) for such a purpose..
Would you mind to share more project details, driever IC, reference, noise & linearity tests ?
I got couple of 11166 and 11167 from Maxim long time back mainly because their true +/-5V bipolar input, still not soldered into something and evaluated. It has got 12 pins (2x6) in 3x3mm package so not easy to prototype..
Evaluating 16bits 250/500kSamples SARs is not an easy exercise, so I doubt I could do it better than Maxim..
Don't have to be better, just real.
-
Btw here is a "pico-ammeter thread" together with my naive experiments with my "Mysterious ion detector probe" (a picoammeter basically) I made.. :)
https://www.eevblog.com/forum/projects/picoammeter-design/msg4912018/#msg4912018 (https://www.eevblog.com/forum/projects/picoammeter-design/msg4912018/#msg4912018)
-
Even the cheapest DMMs can measure µA, for nA there is the neat trick to just use it in volt range as a nA (down to pA) meter by using the input impedance of 10Meg and calculate the current (I = U / 10E6).
For DMMs with high input impedance there is usually a mode to set it to 10Meg (FixedZ).
-
Even the cheapest DMMs can measure µA, for nA there is the neat trick to just use it in volt range as a nA (down to pA) meter by using the input impedance of 10Meg and calculate the current (I = U / 10E6).
For DMMs with high input impedance there is usually a mode to set it to 10Meg (FixedZ).
I was going to recommend that however I assume the OP wants a close to zero burden voltage. A multimeter would have the burden voltage of the range full scale, so typically 2 volts or more.
-
DIY current meter that can measure 10 - 1000 nA level currents.
If I were trying (with home made method) to measure very small current (for example 1nA full range) not searching for specialized elements my first idea would be to use capacitor loaded to let us say 0.1V (to not insert too high voltage drop in the measured circuit).
10nF loaded with 1nA during 1s will get 0.1V. Then connected (I assume reed-relays) to the other capacitor (for example 1nF) will load it close to 90% of its voltage and after few cycles the second capacitor will get very close to the end voltage at first one. This way problem is converted into measuring 1nF capacitor voltage not discharging it. I would use 2 DMMs. One to set the reference voltage to be equal to this 1nF capacitor voltage (DMM on the most sensitive Voltage range should show 0) and second to measure hand regulated reference voltage.
I assume collecting charge during 1s to limit reed-relay switching times fluctuations influence on measuring result. Calibration can be a challenge. If we will have 1nA reference source the time constant can be varied to just get exactly 0.1V output.
It may not work for 1nA, but for 10nA or 100nA it should work, I think.
-
Guys, you are making this out to be a difficult problem, and it really isn't! Achieving 1uA FSD using a JFET-input op-amp like the OPA1641 is trivial. You can pretty well rely on a JFET having gate leakage current of <2pA, so that's not going to upset a 1uA FSD meter. Input voltage offset is not amplified by an electrometer, so you're likely to see a 2mV or thereabouts offset on the output for an instrument that will develop 10V at FSD, so that's 0.02% error at full scale and can be tweaked out if necessary. The biggest error comes from the current conversion resistor, but 10M is easily available in 0.1% tolerance and could be trimmed if necessary. In short, it's a simple problem, easily solved. You could add input protection and make it more complex or you could make it easy to replace the (cheap) op-amp.
If you want 1nA FSD, you'll need a 10G current conversion resistor, available in 1% tolerance at best, so the <2pA gate leakage current is still less than the error from the current conversion resistor. But you will need to be careful about cleanliness with that large a resistance.
-
Guys, you are making this out to be a difficult problem, and it really isn't!
Sure..
The OP wants to use the 662 or 6082 -> ADS1115 -> rp2040 or arduino.
He wants a schematics or a link to such a project/setup which will work fine..
That is all..
-
If I were trying (with home made method) to measure very small current (for example 1nA full range) not searching for specialized elements my first idea would be to use capacitor loaded to let us say 0.1V (to not insert too high voltage drop in the measured circuit).
10nF loaded with 1nA during 1s will get 0.1V. Then connected (I assume reed-relays) to the other capacitor (for example 1nF) will load it close to 90% of its voltage and after few cycles the second capacitor will get very close to the end voltage at first one. This way problem is converted into measuring 1nF capacitor voltage not discharging it. I would use 2 DMMs. One to set the reference voltage to be equal to this 1nF capacitor voltage (DMM on the most sensitive Voltage range should show 0) and second to measure hand regulated reference voltage.
I assume collecting charge during 1s to limit reed-relay switching times fluctuations influence on measuring result. Calibration can be a challenge. If we will have 1nA reference source the time constant can be varied to just get exactly 0.1V output.
It may not work for 1nA, but for 10nA or 100nA it should work, I think.
Integration is a great way to measure small currents, and the best way to measure the input bias current of low input bias current operational amplifiers.
-
Guys, you are making this out to be a difficult problem, and it really isn't!
Sure..
The OP wants to use the 662 or 6082 -> ADS1115 -> rp2040 or arduino.
He wants a schematics or a link to such a project/setup which will work fine..
That is all..
@iMo my bad, I thought I could brainstorm and get help from others on this DIY project.
-
Improvements to the common transimpedance amplifier can be made when used as a test instrument like feedback around the series input protection resistor.
Damn, the full app note is super interesting, and it's no wonder - written by Bob Widlar himself!
-
Improvements to the common transimpedance amplifier can be made when used as a test instrument like feedback around the series input protection resistor.
Damn, the full app note is super interesting, and it's no wonder - written by Bob Widlar himself!
The LM11 was an amazing part, with the modern equivalent being the much improved bipolar LT1012 or JFET OPA140.