el cheapo micro-ammeter

[Kalvin]

You can split that huge resistor into three smaller ones

In an I-V converter, you try your best to avoid the use of small value resistors to minimize johnson noise. If you look at commercial picometers, they all use high value resistors, some as high as 1Tohm.

Yes, they are expensive, though glass resistors are a lot more available now, especially from China. The polymer types however aren't that great.

I didn't realize how big difference is between the circuit using only the 1 gigaohm resistor and the noisy circuit I provided until I ran the noise analysis. Very illustrative lesson learnt, thank's for pointing that out.

[dannyf]

That feedback network was typically and mistakenly used by many: even Pease and Widlar used it in some appnotes from NatSemi.

Thankfully, the test equipment manufacturers knew better - they all use large value resistors for I-V conversion.

EDIT: just to be complete, the Pease / Widlar feedback network is not without merit: it has much faster response time. The bandwidth of an I-V converter with a large resistor is typically less than 1Hz, and often less than 0.1Hz. That kind of bandwidth is not a problem for typical applications for those meters - they are meant to measure DC.

If you intend to measure faster signals, you have no choice but to use the feedback network Pease and Widlar used.

Hope it helps.

[dannyf]

try to search by this keyword on this site.

Obviously it is hard to make an objective comparison. But here is my attempt:

1) ucurrent's topology is rarely used for measuring tiny current; Burden voltage would be a consideration here;
2) ucurrent is easier to construct: the input impedance is low so it should work even without shielding;
3) ucurrent is more complex;
4) it is hard to make ucurrent to go down to lower measurements. It would be difficult to see how this can be done as a picometer.

If you want to measure small current in a circuit with large voltage drops, ucurrent would work. If you want to measure really small current (reverse leakage current on a diode, leakage current on a ceramic capacitor, etc.), I-V conversion is pretty much the only game in town.

[Zero999]

If you are on a budget and dont wan't to solder, there is something else you have to measure low currents.El-even-cheaper.
Basically a voltmeter in parallel with 10 Megaohm...a voltmeter in parallel with 10 Megaohm...a voltmeter in parallel with 10 Megaohm...

Yes, your garden variety dmm with 10 Megaohm impedance :-)

(exact value may vary but no one prohibits you from finding out )

That will work but the input impedance of DVMs isn't accurately controlled. Fortunately it's easy to measure the input impedance with another meter and take it into account.

At work there are some MCP602s in stock and some 100M resistors which should work quite well. I'll give this a go one lunchtime next week.

[codeboy2k]

Cockroft and Walton say:
El cheapo megger coming soon !

[dannyf]

I did some digging on the bias generator. With my tl082, cos is 10mv with a 100k feedback resistor and 500mv with a 10meg feedback resistor.

So id you are ok with manually subtracting that from your reading, or if your meter can measure relatives, you can get away from that - a path I intend to make with my build.

Now I think a build around a quad would be good -one opamp as rail splitter, three as i-v converters, with 100ua/v, 10ua/v, to 1ua/v. Another would be 10ua/v, 1ua/v, 0.1ua/v.

[Alex Eisenhut]

and offered a diaphragm

I've heard of putting your heart and soul into your work...

[dannyf]

Finished building the uAmmetor on a quad (TL074).

The test was measuring the reverse current in a blue led under 9v battery + 1Meg resistor.

The left picture shows the uAmmetor's reading using a 10Meg feedback resistor (0.1ua / v): 0.129V translates to 0.0129ua.

The right picture shows the uAmmetor's reading using a 1Meg feedback resistor (1ua / v): 0.011v translates to 0.011ua.

Not bad for something you can build with Radioshack parts.

[dannyf]

This build on a quad opamp has no bias generator and the breadboard contains all the parts needed for the uAmmetor.

Pretty amazing if I may say so.

[dannyf]

Here is the uAmeeter's performance with regards to Vcc changes.

TL07x is specified from +/- 3.5 to +/- 15v (7v to 30v in unipolar supply).

Here, we are using the uAmmeter to measure a small current, gain is 0.1ua / v.

We take measurements under 5v/12v/20v power supply voltage. When we switch supply voltage, we zero the meter.

The readings for the same circuit under different power supplies are:

5v: 338mv = 33.8na;
12v: 325.1mv = 32.5na;
20v: 320.0mv = 32.0na;

At 5v, the opamp is out of spec, but not far from a 9v battery's minimum output voltage (6-6.5v).

The sensitivity is the highest at 0.1ua / v and decreases as gain goes down (to 1ua / v or 10ua / v).

[dannyf]

The smallest current I was able to measure was the reverse current on a led: 2.1mv @ 0.1u/v = 0.2na. Would be interesting to see what practical applications there are for a meter like that.

[dannyf]

Played around the uAmmeter a little bit more. See the photos below.

All measurements are taken on the 10Mohm setting (=0.1ua/v). All zeroed.

First of all, very little drift in the hour or so I used it.

Now, g-s leakage current for a fqu1n60c (small mosfet by Fairchild). Immeasurable at this setting, vs. max 100na per the datasheet.

[dannyf]

The same FQU1N60c reverse leakage.

36.3mv @ 0.1ua / v = 3.6na, vs. max 100na in the datasheet.

[dannyf]

Now a p-ch mosfet: FQP3P50.

(forward) Leakage current is immeasurable: 0.00mv @ 0.1ua/v = 0.00na, vs. 100na max

Reverse leakage is about 3.6na, vs. 100na max.

[dannyf]

Leakage of a 1n4148. 17.7mv @ 0.1ua/v = 1.77na.

[dannyf]

Measured some capacitors.

1. ceramics: mostly immeasurable at this setting.
2. electrolytics: too much leakage for this setting.
3. here is polypropylene cap. 0.2mv @ 0.1ua / v = 0.02na. Cannot complain about it,

[dannyf]

A while back, I talked about TDA1308, my new favorite opamp. This is a 8pdip cmos opamp, rated at 5v.

So I decided to try it here. Initially, I ran it at 5v and no problem - a little bit limited in terms of operating range. So I ran it at 12v, no problem still. Good for 9v battery operations, and +/- 4v output for sure.

All test be low are done under 12v power supply.

First, measuring reverse current in some random led, TDA1308 vs. TL074. Feedback resistor is 100k (10ua / v).

Difference is only 0.001mv (0.01ua).

[dannyf]

2nd, Vos at 10Meg feedbakc resistor.

0.000v, vs. 96mv for TL074.

This really surprised me. Vos is specified at 10mv typical in the datasheet.

Maybe it is my lucky day,

[Kalvin]

Thanks for nice idea, dannyf. I built something similar just for fun.

I took your design as a basis for my project. However, I used a bipolar power supply of +/-12V, and added a potentiometer to be able to set the test voltage for the DUT in range +/-10V. I also added a precision full wave rectifier to the op amp's output and hooked one of those cheap 5-digit voltage display to show the rectified leakage current/voltage. As the 5-digit volt display cannot display negative voltages, I added two leds which will indicate the polarity of the test voltage. There is also a toggle switch to select which parameter will be displayed:  The test voltage or the measured leakage in uA. The op amps are LF411 with offset nulling, and a LF347 for the full wave rectifier, led polarity comparator and HI-Z buffer for the test voltage measurement. The measurement range is approximately +/- 10.000uA and the resolution is 1nA.

Edit: I used this project to evaluate those cheap 5-digit volt meter/displays and Mornsun IA0515KS-2W isolated 5V in, +/-15V out dc/dc modules. The circuit is running from a +5V Wall-wart, and the op amps are running from the (unregulated) +/-15V power supply created by the dc/dc-converter. The 5-digit voltage meter/display is operated from the +5V power supply to keep the noise levels under control. Those dc/dc-converters seem to be unregulated and have some output ripple. I used a simple LC-filtering, and the power supply quality seems to be adequate for this application. However, using a linear post-regulator with a LC-fllter might be advisable in more sensitive applications.

[dannyf]

Sounds like a fun project.

Just a word of caution - because of the large value feedback resistor plus capacitance, either discrete or intentionally added, the circuit can have very limited bandwidth, in the sub-hz range with 100gohm resistors for example.

If you need to measure ac, you may consider using lower value feedback resistors plus additional amplification, or the t-network in the feedback loop.

You effectively trade noise for bandwidth.

Hope it helps.

[dannyf]

More measurements.

1) the uAmmeter vs. a multimeter.

Measuring small current from a red led + 1.2Mohm resistor powered by a 9.26v battery.

@ 10ua/v, the uAmmeter reads 0.615v @ 10ua/v = 6.15ua. The multimeter reads 6.1ua.

I made another measurement, with a different led and got 5.16ua on the uAmmeter and 5.1ua on the multimeter.

[dannyf]

Measured the one circuit  (@ 10ua / v) under Vcc=5v/12v/20v.

The measurement remained the same.

[dannyf]

All the experiment was done so far with a TL074 (actually, I suspect it is a knock-off TL074).

Here is a LMC6042 (Ibias=2fa and Vos = 3mv)

Here is a LMC6042, with a 10Meg feedback resistor.

output offset voltage with shorted input is 3.5mv. The TL074 would have an output voltage of 100mv+ in the same condition.

[dannyf]

Now reverse current in a yellow led, @ 0.1ua / v.

The LMC6042 reads 0.87na and the TL074 reads 0.84na.

With the light turned off, the LEd would have a reading of 0.27na. So I attributed the difference between 0.87na by LMC6042 and 0.84na by TL074 to inexact location / lighting of the led between each test. Nonetheless, they are fairly close.

[dannyf]

I am going to finish the construction next.

I hoped that this has shown that measuring small current can be done fairly easily with run-off-the-mill parts.