Nanoammeter circuit

[shishirk]

Hi,

I work in research labs and one of the costlier equipment is high-end digital multimeters (DMMs). For example Keithley 2400 or 2600 series source measure units. I don't need all the functionality these meters offer and don't have budget for them. I have also seen muCurrent by eevblog, but that needs a meter which can read in ~10 microvolts for going down to pA range.

So, I am looking for a circuits/pointers, for making a homemade nanoammeter that I can interact from computer.
- pA level accuracy
- high bandwidth (~1000 readings / min)
- a large and configurable range

A related project is http://www.vk2zay.net/article/251, although I couldn't deduce the design choices relevant to my needs. Similarly, uCurrent uses MAX4239AUT, but my earlier searches led to OPA133 and similar opamps. Can you please comment on that?

Any links, resources, book references are welcome. I have never designed any substantial circuit but have background in EE. The interfacing etc should not cause too much of a problem (e.g. optically decoupled USB/RS232/Wifi links).

thanks and regards.

[Kleinstein]

Fast reading of nA range curent is very difficult. The big problems are capacitance and coupling of external signal. Allready flexible capbles are a problem here.

In such cases one often needs a special amplifier close to the source, as to keep capacitance from cables low. The amplifier for small current is usually more a transimpedance amplifier (TIA) and not a shunt and voltage amplifier. So the linked artikle is the right way to go. The TIA does not need such a low offset OP, so it can use types with even lower bias currents. Often the OP is not that critical - the more difficult part can be the feedback resistor, if more than 100 M are needed. Range switching is difficult - often a separate amplifier for different ranges is easier - also the TIA might work an a large dynamik range, some range switching can be done with the DMM following.

Measuring 10 µV resolution is not really demanding. There should be plenty of mid range DMMs to chosse from, possibly used ones.  Some of the more modern ones may also be fast enough. Usually measuring at 1 NPLC which is 50/60 readings per second is possible - going faster is difficult anyway, as you don't suppress line voltage coupling anymore.   

[Gyro]

Spooky Halloween type timing - I've just been adding my implementation to the Picoammeter design thread...

https://www.eevblog.com/forum/projects/picoammeter-design/

As Kleinstein points out, speed is going to be your problem. The TI LMC662 makes a very good op-amp choice due to its very low input current (typically around 3fA), and low temperature drift, really good for such a a cheap part. Keithley use it in at least one of their Picoammeters too. The problem is getting low enough capacitance for speed whilst maintaining its loop stability. Getting it as close to the measurement point and avoiding cables will certainly help.

[Marco]

Nevermind.

[Gyro]

  I don't understand, where does the 10^9 come in in a transimpedance amplifier? 

[Marco]

I wasn't thinking right, I was mixing up the ucurrent and the TIA and thinking fuzzy.

[Gyro]

Aha, ok 

[krish2487]

@OP
Have you actually stopped and considered why the big names in T&M command so much.
(Exception being fluke, but thats a different story.)


It is not just about achieving one goal like resolution, accuracy, stability, precision, reliability, repeatability, low noise, low drift, almost zero error(static and dynamic) for a given application.
It is about compromising (to a certain extent) on all the factors and consistently delivering on those fronts.


Instrumentation is not childs play.


As I ve said, keithley, keysight, marconi - take your pick. They don't build such instruments to a budget. The applications that demand such capabilities out of a instrument must/should/will have deep enough pockets not to care for the price of such an instrument.


Fluke, commands a premium because of the sheer reliability and ruggedness of their instruments, while not being too picky about the other factors.


Bottom Line : Stop trying to work off a utopian dream. If your application demands it. Pay for it. You can't do a tenth of the requirements for a similar price what the big names ask.

[Marco]

On the other hand, at the end of a meter of cable even a million dollar instrument might not be enough to do what a 1 dollar opamp can do at the source.

[Gyro]

@krish2487

A little over the top don't you think???

You can achieve an awful lot with some well targeted circuitry close in an focused on a particular measurement. As I pointed out, Keithley  use a $1.50 LMC662 as the front end device of one of their megabuck (well not cheap anyway) picoammeters - is it because it's because they judge it to be a well spec'd part for the job, or are they just being greedy B**** and wanting to maximize their profits. I think they're probably honest enough to think the former.

Most of us are design engineers here and if you design it right and for purpose then does it really matter whether you've spent megabucks on it?

Edit: and a lot of us enjoy playing with instrumentation circuits! Childs play or not.

10 Minutes later Still Irritated rant Edit:

As I ve said, keithley, keysight, marconi - take your pick. They don't build such instruments to a budget.

What planet are you living on? EVERY equipment manufacturer builds their equipment to a budget! That kind of 'Leave it to Megabuck equipment experts it's too difficult for us mere mortals' attitude really doesn't wash in an engineering forum!

[krish2487]

@Gyro
Not at all.

I wasn't being condescending about it. The OP works in a research lab. If he needs a specific pA/uA range and would like to achieve it using picked and chosen components. It is doable. It makes for a nice challenge actually. But, as I ve mentioned he needs to be aware of the finer nuances of instrumentation and where he can actually hit a wall. I was merely pointing it out. And having a background in EE != instrumentation.


On the other hand, if he does need the instrument for use in his lab - there is also the issue of calibration. A homebrewn solution might not or cannot be used to publishing research findings. There is a reason why the Voltage references in many such high end gear are aged and chosen. It takes manpower, time and the equipment to characterise a semiconductor, device, instrument etc. It costs a lot of money. Money, which should eventually be borne by the customer that needs that specific kind / range of performance. And being a company they need to make money and profits. If that makes the high end stuff, a sour grape for most then that is the sad truth. After all, that is the secondary goal for a company (money), the primary goal - being good at what they do.


I am not saying that homebrewn / DIY solutions won't work. I am saying that for the specs he has mentioned, he would be hard pressed to do it for any cheaper ( in terms of direct, indirect expenditure, time - however you choose to see it). IIRC there was such a similar thread about one of the HP multimeters 3458?? or so I think. Free_Electron wrote a post about it.


@the budget part
I meant a hobbyist budget. I should have been more clear in my original post. My bad. 

[Gyro]

Hmm, I'm sure you didn't MEAN to be (or at least sound) condescending, but that's exactly how that post reads, hard to interpret it any other way. The OP probably needs to clarify, but he says has a specific measurement requirement that doesn't need the power of a comprehensive instrument and he also says that he doesn't have the budget for such an instrument (Research labs have budget restrictions too in today's world).

The OP is asking for ideas and reading materials at this point. Only he knows his requirements in terms of accuracy and tractability for this particular task - maybe he can clarify further???

Anyway, hopefully this exchange has clarified some of the things that he has to think about in terms of his particular requirement. 

Oh, by the way....

@ shishirk, As nobody else has mentioned it so far, the standard reference for  measurements of this type is the 'Keithley Low Level Measurements Handbook' currently in its 7th Edition. You can download it free from the Keithley website if you enter your details, but it is easy to find on the web too.

[krish2487]

@gyro
My apologies for coming off as an ass. To the op too.

I might actually point out that the little details the op has shared can hardly be classified under "beginners" section. Hence my strongly (but wrongly ) worded reply.

Sent from my 2014818 using Tapatalk

[Gyro]

...and my apologies for having a rant, I'm usually better behaved  As I said, I suspect that the OP has probably gained some useful insight into the issues potentially involved!


@ shishirk,  I've just been re-reading your post too. I noted the bit right at the bottom "The interfacing etc should not cause too much of a problem (e.g. optically decoupled USB/RS232/Wifi links)."  Hopefully you understand that you're going to need at least a basic dvm with whatever interface you need in order to take the output of a nanoammeter type circuit (eg. transimpedance circuit). Hopefully you already have have something for this. Extending the design to include an ADC and interface would certainly be prohibitive in cost of time spent.